^Ai/'s HARVARD UNIVERSITY Library of the Museum of Comparative Zoology S'^'t''^ BULLETIN OF THE Museum of Comparative Zoology HARVARD UNIVERSITY VOL. 133 CAMBRIDGE, MASS. U.S.A. 1965-1966 The Cosmos Press, Inc. Cambridge. Mass., U.S.A. CONTENTS PAGE No. 1. — GuLAR Musculature in Delphinids. By Barbara Lawrence and William E. Schevill. May, 1965 . 1 No. 2. — A Revision of the Genus Rhabdephyris in the Americas (Hymenoptera, Bethylidae). By Howard E. Evans. May, 1965 67 No. 3. — Nclda.mnrus wrightac, a New Rhachitomous Labyrinthodont from the Texas Lower Per- mian. By John Newland Chase. June, 1965 . . 153 No. 4. — The Genera of the Chilocorini (Coleoptera, Coccinellidae). By Edward A. Chapin. Septem- ber, 1965 ..." 227 No. 5. — Comments on some Recent Changes in the Clas- sification OF THE Ciidae ( Coleoptera) . By John F. Lawrence. October, 1965 273 No. 6. — The Fossil Elephant Shrews (Family Macro- SCElididae). By Bryan Patterson. November, 1965 295 No. 7. — ^Panamanian Spiders of the Genus T mar us (Araneae, Thomisidae). By Arthur M. Chicker- ing. November, 1965 337 No. 8. — The Relationships of Four Small Hispaniolan Elcuthcrodactyliis (Lei^todactylidae). By Albert Schwartz. January, 1966 369 PAGE No. 9. — Two New Fishes of the Myctophid Genus Di- aphus FROM the Atlantic Ocean. By Basil Naf- paktitis. January, 1966 401 No. 10. — The Ameiva (Lacertilia, Teiidae) of Hispani- OLA. II. Geographic A^ariation in Ameiva chryso- laema Cope. By Albert Schwartz and Ronald F. Klinikowski. March, 1963 425 No. 11. — A New Attempt to Construct Life Tables for Kent Island Herring Gulls. By Raymond A. Paynter, Jr. May, 1966 ..... \ ... 489 Bulletin of the Museum of Comparative Zoology HARVARD UNIVERSITY Vol. 133, No. 1 GULAR MUSCULATURE IN DELPHINIDS By Barbara Lawrence and William E. Schevill CAMBRIDGE, MASS., U.S.A. PRINTED FOR THE MUSEUM May 21, 1965 Publications Issued by or in Connection WITH THE MUSEUM OF COMPARATIVE ZOOLOGY HARVARD UNIVERSITY Bulletin (octavo) 1863 — The current volume is Vol. 133 Bbeviora (octavo) 1952 — No. 221 is current. Memoirs (quarto) 1864-1938 — Publication terminated with Vol. 55. Johnsonia (quarto) 1941 — A publication of the Department of Mollusks. Vol. 4, no, 43 is current. Occasional Papers of the Department of Mollusks (octavo) 1945 — Vol. 2, no. 31 is current. Proceedings of the Nev^ England Zoological Club (octavo) 1899-1948 — Published in connection with the Museum. Publication terminated with Vol. 24. The continuing publications are issued at irregular intervals in num- bers which may be purchased separately. Prices and lists may be obtained from the Publications Office of the Museum of Comparative Zoology, Cambridge 38, Massachusetts. Peters' Check-List of Birds of the "World. Volume 1 is out of print; volumes 2 to 7, 9, 10 and 15 are currently available. Phylogeny and Evolution of Crustacea. Proceedings of a confer- ence held at Cambridge, Mass., March 6-8, 1962. H. B. Whittington and W. D. I. Rolfe, editors. Cambridge, Mass., 192 pp., 80 figs., 1963. Cloth bound, $6.75. Fishes of the Gulf of Maine, by Henry B. Bigelow and William C. Schroeder. Washington, viii - 577 pp., 1953. Photo-offset reprint, $6.50. Publications of the Boston Society of Natural History The remaining stock of the scientific periodicals of the Boston Society of Natural History has been transferred to the Museum of Comparative Zoology for distribution. Prices for individual numbers may be had upon request. Bulletin of the Museum of Comparative Zoology HAEVAED UNIVEESITY Vol. 133, No. 1 GULAR MUSCULATURE IN DELPHINIDS By Barbara Lawrence and William E. Schevill CAMBBIDGE, MASS., U.S.A. FEINTED FOE THE MUSEUM May, 1965 Bull. Mus. Comp. Zool., Harvard Univ. 133(1) :l-65, May, 1965 No. 1 — Gular Musculature in Delphinids^ By Barbara Lawrence and William E. Sciievill TABLE OF CONTENTS PAGE Introduction 5 General account of Delphinidae 6 Superficial Layers 7 M. sphincter colli profundus 8 M. sphincter colli primitivus 8 M. auriculolabialis 8 M. orbicularis oculi 9 M. orbicularis oris 9 M. nasolabialis 9 Ear muscles 10 Muscles of the Neck and Throat 11 Interramal and Tongue Muscles 12 Mylohyoid muscle 12 Digastric muscle 12 Geniohyoid muscle 13 M. styloglossus 14 M. hyoglossus 14 M. genioglossus 14 M. palatoglossus IG Hyoideal Muscles 17 Sternohyoid muscle 17 Sternothyroid muscle 17 Thyrohyoid muscle 18 Occipitohyoid muscle 18 Interhyoid muscle 18 M. hyoepiglotticus 19 Muscles of the Pharynx 19 M. stylopharyngeus 21 M. palatopharyngeus 22 M. pterygopharyngeus 24 Occipitothyroid muscle 25 M. thyreopharyngeus 26 Homologies with Phocoena 26 Discussion 29 1 Contribution No. 14.37 from the Woods Hole Oceanosraphic lustitution 4 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Globicephala melaena 50 Muscles of the Xeck and Throat 50 Sternomastoid muscle 50 Mastohumeralis muscle 50 Scalenus muscle 50 Mylohyoid muscle 51 Digastric muscle 51 Geniohyoid muscle 51 M. styloglossus 52 M. hyoglossus 52 M. genioglossus 52 M. palatoglossus 53 Sternohyoid muscle 53 Sternothyroid muscle 53 Thyrohyoid muscle 53 Occipitohyoid muscle 53 Interhyoid muscle 54 Muscles of the Pharynx 54 M. stylopharyngeus 55 M. palatopharyngeus 55 M. pterygopharyngeus 56 Occipitothyroid muscle 56 M. thyreopharyngeus 57 Conclusion 57 Acknowledgments 58 Literature Cited 58 LAWRENCE AND SCHEVILL : GULAR MUSCULATURE 5 GULAR MUSCULATURE IN DELPHINIDS INTRODUCTION A scattering of authors in the past ninety years have pub- lished on different aspects of the head and throat musculature in the odontocetes. No two have gone about their dissections in the same way, the names used for the muscles vary, and in some instances the inadequacy of their material has led to conflicting statements on the number and position of the muscles involved. Our investigations of sound production in the larynx have pointed up the need for a series of more detailed dissections of the entire region, as well as for correlating the descriptions of some of these early workers with each other and with the condi- tions that we actually found. Our primary concern has been to establish the true relation- ships of the hyo-laryngeo-pharyngeal muscles or, to put it differ- ently, to study the muscles which lie between the throat car- tilages and the base of the cranium, medial to the basioccipital plates and posterior to the palate. Whether or not the great development of the basioccipital plates is related to the peculiar structures associated with the ear bones or to the evolution of an intranarial larynx, the combined result has been a character- istic arrangement of the muscles in this region which is com- mon to many odontocetes. Since accurate dissection of these deeper layers depends on an understanding of the more super- ficial ones, the latter also have been figured and described. The muscles external to the pterygoid plates are quite distinct from those of the throat region, and so are omitted. The boundary between these two regions is a tough, fibrous mem- brane which extends from the tip of the pterygoid hamulus to the tissue investing the bulla on each side, and lies across the notch in the pterygoid plates. It separates the muscles of the bony nares from the pterygoid muscles. The complex of air sinuses and retia external to this has been well described by Fraser and Purves (1960a, pp. 65-68). The first section of this paper deals with four rather closely related genera represented by the following species : Delphinus delphis Linne 1758, Stenella plagiodon (Cope) 1866 and -S'. styx (Gray) 1846, Tursiops truncatus (Montagu) 1821, and Lagcno- rhynchus acutus (Gray) 1828 andL. alhirostris Gray 1846. Herein they will usually be referred to by generic name only. The pur- pose of this section has been not to compare the four genera, but 6 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY to establish the common delphiuid arrangement of muscles in this region. For this reason, the dissections have rarely dupli- cated each other. ]\Iost of the figures are of Dclphinus, but to shoAV certain details some figures of Tursiops and some of Lagenorliynchus are included. The specimens studied are those reported on in the "Functional anatomy of the delphinid nose" (Lawrence and Schevill, 1956), and include, in addition, a specimen of Lagenorhynchiis alhirostris and one of Stenella styx, as well as a representative of the Phocoenidae, PJiococna pliocoenn (Linne) 17o8, all collected in the western North Atlantic. The second section is a clarification of Murie's classic de- scription of the throat region in Glohicephala mclacna (Traill) 1809. Although also a delphinid, this is a much larger form which differs further from the above four genera in its relatively shorter rostrum and broader skull. In the following, for ease of description, the muscles consid- ered are grouped and discussed in three divisions : the superficial layers, the muscles of the neck and throat, and the pharyngeal muscles. GENERAL ACCOUNT OF DELPHINIDAE In the hope of standardizing muscle names in the cetaceans, we have not followed one author but have selected what seem to us the most appropriate terms. Huber's terminology (1934, pp. 117-119) has been used for the superficial layers. For the hyo-laryngeo-pharyngeal region, the names are chiefly from Murie (1873), Kernan and Schulte (1918), and Howeir(1927). When the names used are Latin adjectival forms, the word "musculus" is to be understood as preceding the name; when the adjectives are anglicized, the English word "muscle" may be understood to follow the name. In order to avoid confusion between the bones and similarly named muscles, we have followed Howell (1927) and earlier authors in using the "-hyal" ending for different components of the hyoid apparatus, as for instance thyrohyal, while in these cases we have used "-oid" for muscles, as for instance ihyro- hyoid. Since the thyro- and basihyal elements fuse early, it is not possible to tell whether certain muscles attach on both or on only one or the other. This is an unimportant detail ; never- theless, it should be remembered that in some instances what we call a thyrohyal attachment may reallv be a thvrobasihval LAWRENCE AND SCHEVILL : GULAR MUSCULATURE ( attachment and vice versa. The current names for the laryngeal cartilages are so universally accepted that the standard forms thyroid, cricoid, and arytenoid have been used. This is less con- fusing than in the case of the hyoid apparatus, for in the larynx parallel names for muscles are not in common use. To make clear the interrelationship of the muscles as well as to compare better our findings with those of other workers, the material was actually dissected in a number of different ways. The sequence in which the muscles are presented here is chosen as the easiest for purposes of identification, as they are exposed in progressively deeper ventrolateral dissection. Additional de- tails of the more complicated muscles based on other types of dissection are also illustrated. These are essential if function as well as identification is to be understood. Aside from Murie's work on Glohicephala, the most useful of the earlier dissections of this region in odontocetes have been made b,y: Schulte and Smith (1918), and Kernan and Schulte (1918) of Kogia; by A. B. Howell of Neomeris (1927), and Boenninghaus of Phocoena (1902) ; by Fraser and Purves of Delphinus (1960a and b) ; and by Hein (1914) of Monodon. Where our dissections overlap and their findings differ sig- nificantly from ours, these are discussed ; otherwise their work is merely noted. Superficial Layers Huber's (1931:, pp. 117-120, fig. 4) detailed description of the facial muscles of Tursiops, published posthumously, is an in- tegral part of his carefully reasoned series on the evolution of facial musculature. In another paper (Lawrence and Sehevill, 1956), we have discussed his account of the blowhole muscula- ture in the light of our own findings. His description of the superficial facial muscles matched closely what we observed in the four genera dissected, and the following comments are mainly supplementary. These outer layers have been considered by us only because of their relation to the deeper hyopharyn- geal regions which are our primary concern. The blubber in the region of the head and throat is closely bound to the underlying muscle. Some of the outermost layers of the sphincter colli actually insert in the dense inner layers of the blubber. Ventral to the eye and surrounded by the sphincter colli profundus, auriculolabialis, and orhicidaris oris, there is an extensive area of fat lying directly beneath the 8 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY blubber and against the jaw. The remnant of the outer ear lies buried in this. Distally, this vestige is a very slender tube sur- rounded by dense fibrous tissue and embedded in fat. Prox- imally, the tube widens and is partially enclosed by small, curved cartilages of a characteristic shape. 31. sphincter colli profundus (Fig. 1, spf). This extends in a thin, transverse sheet from the symphysis of the jaw to the region behind the flipper, beyond which it was not dissected. This muscle was studied only in Delphinus and Lagenorhyn- chus acutus. In these, an aponeurotic area occurs on the throat over the sternohyoid. Posterior to this, especially laterally between the forelimb and the ear, the sphincter colli profundus is thicker. Along each side, at about the level of the eye, it meets the sphincter colli primitivus. Dorsal to the forelimb, from the line where these two meet, a small bundle of fibers converges to insert in the flipper. Anterior to this, a very few fibers pass ventrally to attach to the fibrous mass surrounding the auditory tube, and deep to these a few other fibers merge with those of the auriculolahialis. M. sphincter colli primitivus (Fig. 2, spt). This lies dorsal to the sphincter colli profundus and is very difficult to dis- tinguish from it along the line where the two meet. Only a slight difference in the direction of the fibers suggests that, instead of a single thin sheet of muscle wrapping around the side of the face and throat, there are in reality two muscles. A small, doubtfully distinct, anterior segment (Fig. 2) attaches in the dense fibrous tissue investing the slender tube which passes from the minute orifice of the outer ear to the vestigial ear cartilages. This may be what Huber (1934, p. 117) refers to as "a small vestige of the extrinsic postauricular musculature." This is the only part of sphincter colli primitivus figured. M. auriculolahialis (Figs. 1, 2, al). This arises mainly along the rim of the zygomatic process of the squamosal external to the glenoid fossa and posteroventral to the root of the zygo- matic arch. Posteriorly, it is partly overlain by the ear cart- ilage, and a small part of the muscle takes origin in the tissue here. From its origin, auriculolahialis passes anteriorly, ventral to orbicularis oculi, diminishing rapidly in size, to insert as described later. The primary squamosal origin is fairly con- stant in all the forms dissected, but the origin of the thin layer which is associated with the vestiges of the outer ear varies. In Delphinus this is in the fibrous mass surrounding the auditory LAWRENCE AND SCHEVILL : GULAR MUSCULATURE 9 tube, distal to the cartilage; in Stenella plagiodon it is dif- fusedly iu the tissue investing the ear cartilage itself, while in Lagenorhy nchus acutus there is no apparent connection with the cartilage, but the attachment is in the connective tissue mass between cartilage and skull. The relation of the insertion of auric ulolabialis to orbicularis oculi, the zygomatic arch, and the connective tissue beneath the latter varies. In Delphinus and Stenella plagiodon we found auriculolahialis inserting in part on the arch itself, while in Lagenorhy nchus acutus the in- sertion was in the connective tissue ventral to the arch. In all three there was a certain connection with orbicularis oculi, very slight in the last genus, more definite in the two former. In Tursiojjs, Huber says that auriculolahialis passes from the pre- auricular region to fuse with orbicularis oculi and gives no other insertion for it. Possibly all of this is individual, not generic, variation and an indication that auriculolabialis in delphinids is degenerating. This muscle is the tympano-zygomaticus of Howell (1927, p. 22). M. orbicularis oculi (Fig. 2, oc). AVe dissected this in detail only in Stenella plagiodon. Here, as in Huber 's Tursiops, it forms a well-developed sphincter around the eye. A few of the outer fibers anteriorly and posteriorly attach to the margin of the orbit, while anteroventrally some pass downward into fibrous fatty tissue between the eye and orbicularis oris. M. orbicularis oris (Fig. 2, or). This is a weak and very fat muscle whose fibers radiate from the connective tissue near the corner of the mouth and end in the fibrous inner layer of blub- ber. M. nasolabialis (not figured). The muscles between the eye and the melon were dissected in detail only in Stenella plagio- don and did not match Huber 's figure of Tursiops (1934, fig. 4A, p. 118). In this region, he shows a single rather extensive nasolabialis continuous with the anteriormost fibers of orbicu- laris oculi. The situation in Stenella was a little different; here we found two small sheets of fibers which were separated from the orbicularis oculi by the maxillary bone. The more pos- terior took origin on a narrow area on the maxillary, dorsal to the anterior part of the eye, and passed anteriorly to end in the fibrous mass beneath the melon. The more anterior sheet arose in front of this, possibly on the lacrymal as well as the maxil- lary, and passed anterodorsally to end in the same region as the posterior. Both are poorly developed and no attempt 10 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY has been made to decide of which layers they are vestiges. Ear muscles (Fig. 2). These were not dissected out in detail, but an apparently single small muscle was found beneath the sphincter colli primitivus with fibers directed anteroventrally to insert in the fibrous tissue on the inner side of the upper end of the car cartilage and surrounding the slender auditory tube. Huber (1934, pp. 134-35, fig. 11) gives a thorough account of the ear muscles in Monodon, and Murie (1873, p. 250, fig. 29) discusses them in GloMcephala. Fraser and Purves (1960a, p. 62) refer to them rather generally in the course of a more detailed account of other aspects of the ear. Discussion. Huber 's identification of the muscles in this region is built on evolutionary studies which have led him to the conclusion that the platysma and the sphincter colli pro- fundus derive from the sphincter colli primitivus. In cetaceans he recognizes both the primitivus and profundus, but says that the platysyna colli et faciei is lacking. Earlier authors have tended to include the superficial musculature of the head and neck in the panniculus carnosus (Murie, 1873, figs. 57, 58; How- ell, 1927, p. 19; Schulte and Smith, 1918, p. 15). Though Murie (1873, pp. 272, 273) refers to the anterior portion as the platysma, the others do not make such a distinction. The dorso- ventral division into a sphincter colli primitivus and profundus emphasized by Huber does not show in their figures. This separation, though not very definite in our dissections, is still sufficiently apparent to reinforce Huber 's opinion. Kesteven (1941, pp. 74-75), while recognizing the same muscle arrange- ment as that found by Huber and ourselves, believes that sphinc- ter colli primitivus and profundus are actually anterior ex- tensions of the panniculus carnosus and do not belong to the superficial facial musculature. Schulte and Smith's description (1918, pp. 14-15) of the sphincter colli musculature (called panniculus carnosus) in Tur- siops does not entirely match Huber 's and our dissections. Their aponeurotic area on the throat anterior to the pectoralis is as we found it. We differed, however, in not finding, on the side of the neck immediately anterior to the forelimb, an aponeurotic layer with which the ventral aponeurosis was continuous. On the contrary, in this region the muscle was well developed, passing ventrally across the sternohyoid, in part to join its fel- low, in part to end at the pectoralis. Posterior to this and dor- sal to the forelimb, there is a distinct separation between the LAWRENCE AND SCHEVILL : GULAR MUSCULATURE 11 thicker band of fibers passing anterior to tlie forelimb and the thinner layer which converges to insert on the forelimb. The development of an aponeurosis in this area may vary individu- ally; we found it only in Lagenorliynchus acutus, and here it was always separated from the aponeurotic throat patch by the flipper. Huber 's derivatives of the sphincter colli profundus — the auriculolabialis (tympano-zygomaticus) , orhicularis oculi, and orbicularis oris — were all found by Howell. It is probable that the ''bundles" of Schulte and Smith (1918, p. 14) "caudal to the angulus oris" are the same as the orhicularis oris. Muscles of the Neck and Throat The muscles described in this group are ventral only, and for the most part associated with the hyoid apparatus. The more superficial and anterior are described first under the heading of interramal and tongue muscles, the more posterior and deeper are described second under the heading of hyoideal muscles. Correlated with the development of a fusiform body, the neck in cetaceans has been shortened and thickened and the muscles of the throat strongly developed. The hyoid elements also are well developed, with their main portions ossified even in young animals. In adult individuals, the basi- and thyrohyals form a single, broadly crescentic bone extending across the throat for the full width between the basicranial plates. Anteriorly on each side, a cartilaginous ceratohj^al connects each basihyal with a three-sided, rod-like stylohyal, which is also ossified. The sty- lohyal attaches to the exoccipital by means of a cartilaginous tip which may be a separate element and is sometimes called the tympanohyal. Functionally, the expanded hyoid apparatus is important as an attachment for a number of muscles in this region, where the typical cetacean relation of jaw and occipitals makes a more standard mammalian arrangement impossible. The digastric or depressor mandibulae has its origin here, and a very thick sternohyoid helps to fix the basi-thyrohyals in place as the jaw is opened. Counteracting this backward pull is a moderately strong geniohyoid and the much expanded interhyoid which draws the thyrohyal up and forward towards the stylohyal, Avhile the latter is in turn lifted by the palatopharyngeus. The association of the tongue muscles with the hyoid appa- ratus is a more normal one and needs little comment here. 12 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Interramal and Tongue Muscles The muscles of the interramal region and the tongue are easy to distinguish and not too different from those of other mammals. Mylohyoid muscle (Figs. 2, 8, my). This is a thin, coarsely fibered, transverse muscle that covers the throat from the symphysis of the jaw to the hyoid region. Anteriorly, a few of the more external fibers insert in the tough membrane along the lower margin of the jaAv, while the main mass of the muscle turns upward to insert in the tissue internal to the lower part of the jaw. Posteriorly, the muscle passes between the genio- hyoid and the digastric, becoming increasingly fat towards its insertion, with the digastric, in the fatty tissue inside the jaw. Here, the direction of the fibers is dorsal and slightly postero- dorsal, and the transition from muscle to fat mass is so gradual that it is not possible to say where the one ends and the other begins. This is characteristic of the digastric as well. The con- nection with the hyals is vestigial and lateral only. Here, on each side, a thin sheet of the mylohyoid passes internal to the digastric to attach with it via the same aponeurotic sheet on the thyrohyal. JMedial to this, the mylohyoid ends in an aponeu- rosis which merges with that attaching the geniohyoid to the basihyal. Anterior to this aponeurosis, the fibers of each side meet in an ill-defined raphe which is bound, in the midline, to the geniohyoid. In Stenella plagiodon, the anterior portion of the mylohyoid was difficult to separate from the overlying sphincter colli profundus. The mylohyoid is identified as such by Schulte and Smith, Howell, and Hein. Boenninghaus figures but does not discuss it. Digastric muscle (Figs. 2, 3, 8, 9, d). This arises chiefly from the thyrohyals and probably in part from the basihyal also; laterally, a few fibers take origin along a ligament which runs from the tip of the thyrohyal to the tissue investing the tym- panohyal cartilage. The digastric becomes increasingly more fatty as it passes anterodorsally to terminate in the fatty tissue around the lower margin of the jaw. The same gradual transi- tion from muscle to fat seen in the mylohyoid makes it not pos- sible to define exactly the insertion of the digastric. Generally speaking, it is around the posterior third of the jaw. Some variation in the exact direction of the fibers showed in the dif- ferent individuals dissected. LAWRENCE AND SCHEVILL : GULAR MUSCULATURE 13 The digastric of this paper is the same muscle as Schulte and Smith's hyomandihularis and Howell's mono gastric. Hein's digastric, although figured at insertion only, is clearly the same as ours. Geniohyoid muscle (Figs. 3, 8, gh). Situated between the digastric muscles and internal to the mylohyoid, this muscle is very distinct from the underlying complex of tongue muscles, from which it is sej^arated by sheets of loose connective tissue. It takes origin in an aponeurosis across the basihyal external to the hyoglossus and internal to the mylohyoid muscles. As described above, the aponeurotic insertions of both mylohyoid and geniohyoid are closely bound together. The belly of the muscle is thick and rather short and the insertion is in a long, rather thick aponeurosis inside the jaw at the symphysis. A few of these tendinous fibers are also closely bound to the 'mylohyoid. Some variation was noted in the different forms dissected. In Dclphinus, its origin extended a little distance on to the thyrohyal where it was overlaid by the insertion of the digastric and its insertion was on the inner surface of each ramus of the jaw as well as at the symphysis. In Stenella plagiodon and Lagenorliynchus acutiis there is no overlap of digastric and geniohyoid at their origin. For the most part this appeared to be a single muscle, though in Lagenorhynchiis acutus the proxi- mal end of the belly internally showed a slight separation into two parts. Schulte and Smith show and describe (1918, fig. 11, p. 37) a geniohyoid which is much like ours; in their figure 12, page 39, the origin shown for the genioglossiis is probabh'- that of the geniohyoid. Howell describes a geniohyoideus wliich differs from ours in taking origin along the entire cranial borders of the basihyal instead of transversely across the middle, and suggests that the insertion is in the tongue, not the mandible. His figures (1927, fig. 4, p. 10, fig. 9, p. 25) of the geniohyoid, showing an extensive thyro- as well as a basihyal origin, look more like our hyoglossus. Hein found no geniohyoid, but does not take this as proof of its absence in Monodon. In his specimen only the hyoid attach- ments of the superficial throat muscles were left, and here, as described above, the aponeurotic attachments of mylo- and ge7iiuhyoid are one. For this reason, we agree Avith Hein's sup- position that a geniohyoid does occur in Monodon, and further 14 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY believe that it is much like the geniohyoid found in our dis- sections. The geniohyoid is figured but not discussed by Boenninghaus. 31. styloglossus (Figs. 3, 4, 8, 9, sg). This muscle has its origin on the anterior surface of the lateral end of the stylo- hyal, dorsal to the attachment of the interhyoideus. It is charac- teristically thick at its origin, often blunth" wedge-shaped with the wide end lateral, and the tapering end sometimes extending along as much as a third of the length of the stylohyal. From its origin it passes anteriorly, ventral to the hyoglossus and dorsal to the geviioglossus to insert in the under surface of the tongue anterior to, and in part also lateral to, the medial part of the hyoglossus. Where the two meet at their insertion, they mingle. Sehulte and Smith, Howell, and Hein all find a styloglossus which differs little from ours. M. hyoglossus (Figs. 4, 8, 9, /(). This is a paired muscle, that of each side being indistinctly divided into two sections (Figs. 3, 8). Its origin laterally is from the anterior border of the thyrohyal between the mylohyoid and interhyoid, and medially from the basihyal (processus lingualis) internal to the ge^iio- hyoid. From its origin the muscle passes forward between the styloglossus and palatoglossus, radiating in a thin layer of coarse fibers to surround the oropharynx almost completely. The lateral section, which is also the most posterior, inserts on the dorsal surface of the palatoglossus. The medial section sends a few thin bundles to the sides of the oropharynx, but for the most part passes internal to the gcnioglossus to mingle with the styloglossus at its insertion in the under surface of the tongue. The degree of distinctness of the two parts varied in the genera examined. The hyoglossus of Sehulte and Smith as well as that of Howell is clearly the same as ours, although they differ in certain respects. The former is described as a smaller muscle with origin from the ceratohyal. The latter is said to have its origin on stylo- and ceratohyal; this may be an error, as Hein in Monodon and we, as well as Boenninghaus, in Phocoena, found the attachment of the hyoglossus to be much as in the Delphinidae. Hein's hyopalatinus is the lateral portion of our hyoglossus. M. gcnioglossus (Figs. 3, 4, 8, 9, gg). This muscle lies internal to the geniohyoid and takes its origin medially in a long raphe from the ventral surface of the tongue and pharynx. From its LAWRENCE AND SCHEVILL : GULAR MUSCULATURE 15 origin it passes anterolaterally, spreading over the insertions of hyoglossus and styloglossus to insert principally in the lining of the anterior part of the buccal cavity. A few fibers also mingle with the mylohyoid to insert in the lower inner margin of the jaw posterior to the geniohyoid. At its caudal end, a narrow, scarcely muscular band of the genioglossus lies internal to the liyals, with which it has almost entirely lost its connection. In Tursiops, a vestigial connection remains, via a thin aponeurotic sheet, to the dorsal surface of the basihyals as well as to the dorsal surface of the geniohyoid. While our failure to find such a connection in the other three genera does not mean it is necessarily always absent, the primary posterior insertion of the genioglossus is medially in the ventral wall of the pharynx. Where the palatoglossus wraps around the pharynx, the origin of the genioglossus is from the ventral surface of the palato- glossus, while anterior to this it is in the root of the tongue. M. genioglossus is not always clearly distinct at its origin, and, in Lagenorhynchiis acutus at least, where the lateral section of the hyoglossus inserts on the palatoglossus, parts of the three mingle so as to be indistinguishable from each other. Schulte and Smith as well as Howell find a similar genio- glossus. Boenninghaus ' genioglossus s. s. (1902, p. 66, pi. 1, fig. 1) and the anterior part of Hein's genioglossus are homo- logous with our muscle. In addition, Boenninghaus describes a posterior extension of the genioglossus which he calls the genioepiglotticus. Boen- ninghaus describes and figures this (1902, p. 66, pi. 1, fig. 1) as a medial, ventral band of muscle passing from the pharynx to the epiglottis. This in effect would bind the wall of the pharynx to the epiglottis anteroventrally. We found no such connection. Posteroventrally, the tongue muscles end with the insertion of palato- and genioglossus (Fig. 8). Caudal to this, the pharynx passes dorsal to the cerato- and basihyals and divides to go between the epiglottid and thyroid cartilages on each side (Fig. 6). In this region, as described under palato- pharyngeus, there is a continuity of muscle from pharynx to epiglottis. This, however, is lateral and posterior to the division of the pharynx. Medially, where the pharynx divides, the larynx is strongly bound to the hyals by the hyocpiglotticus (see below) but this muscle is quite distinct from the pharyngeal muscles. Boenninghaus further distinguishes a glossoepiglottimis which he says goes from the dorsal surface of the base of the tongue 16 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY and merpres with genioepiglotticns (1902, p. 66, pi. 1, fig. 1). In our dissection of Phocoerm a careful search was made for both of these muscles. Only two extrinsic muscles were found inserting on the anterior part of the epiglottis, the hyoejnglotti- cns and the anterior portion of the thyropalatine section of the 'paJafopharyngeus. The arrangement of these did not significantly differ from that found in the Delphinidae. In both groups, posterior to the margin of the palatoglossus, where the pharynx lies dorsal to the hyoid complex, there is a well defined non-muscular area. Anterior to this are the muscles of the oropharynx ; posterior to this are the muscles of the hyoid apparatus and the nasopharynx. Along the lateral walls of the divided pharynx, a few fibers (see under palato- glossus) pass between these two regions. They have no con- nection with the hyoepiglotticus, from which they are separated by a distinct space. Whether these belong to the palatopharyn- geal complex or to that of the tongue muscles, or whether they are vestiges of totally other muscles is not clear. In any event, they are so poorly develoj^ed as to have little functional sig- nificance. Hein (1914), apparently following Boenninghaus, also de- scribes a complex genioglossus, distinguishing a posterior por- tion of this as the genioepiglotticus. Where he figures this as medial (pi. 5, fig. IX, muscle 54), we believe this to be an error, though possibly it could be part of our hyoepiglotticus. Where he figures it as lateral (fig. X, muscle 54), it is appar- ently a better developed part of what we describe below as a posterior extension of the combined palato- and genioglossus. Muscle 67 of figure X, also called genioepiglotticus, appears to be part of the thyropalatine section of our palatopharyngeus. Hein, like us, was unable to identify a glossoepiglotticus. 31. palatoglossus (Figs. 4, 6, 8, 9, pg). This surrounds the oropharynx for almost its entire length, and mingles without sharp boundary with the various tongue muscles which lie external to it, the hyoglossus laterally and the genioglossus ventrally. Dorsally, the palatoglossus is closely bound to the bony palate, and in the midline fibers from opposite sides meet, but do not form a conspicuous raphe. For the most part, the palatoglossus is here thinly covered by the lateral section of the hyoglossus, although anteriorly it extends beyond this latter. From the posterior margin of the combined palatoglossus and genioglossus a very few fibers extend, as described above, along LAWRENCE AND SCHEVILL : GULAR MUSCULATURE 17 the lateral walls of the divided pharynx to join a similarly indistinct section of the palatopJiaryngeus. As stated earlier, this fiber tract is probably what Ilein has called gc7iioepiglotti- cus. It may also include the glossocpiglotticns of Boenninghaus. Schulte and Smith have no muscle corresponding; to this, but Howell, Hein, and Boenninghaus all report a palatoglossus which, except as described above, is the same as ours. Hyoideal Muscles Identification of those muscles which fix the position of the hyoid relative to the sternum, and those muscles which fix the position of the larynx relative to both the sternum and hyoid apparatus presents no serious problems. Although their arrange- ment differs in certain characteristic ways from that in other mammals, homologies are not difficult to understand. With the deeper layers the situation is different. The great transverse extension of the thyrohyal bone and the rather nar- row space separating the thyrohyal from the stylohyal bones have resulted in rearrangement of the muscles of this region. A single large mass, presumably equivalent to the cerato-hyoideus {=interhyoideus) of other mammals, fills the whole interspace between these two. Of the two other muscles usually found in this region, the stylohyoideus and occipitohyoideus {—jugulo- hyoideus), there is only one vestigial remnant. The attach- ments of this small muscle differ from both of the above, and whether or not it is the displaced homologue of either, it is not possible at this time to say. From its position, we tend to be- lieve it is the occipitohyoid and have so called it. Sternohyoid muscle (Figs. 1, 2 sli). This is a thick muscle arising on the anterior part of the sternum, and covering the thyroid cartilage medially as it passes forward to insert broadly on the basi- and thyrohyal bones. There is little disagreement over the name of this muscle although, by misprint, Howell (1927, p. 26) calls it as well as the next muscle sternothyroideus. Sternothyroid muscle (Fig. 4, st). This arises on the anterior margin of the sternum and, in Stenella plagiodon at least, from the first rib near its sternal attachment. Thence it passes anteriorly to insert on the outer surface of the posterior horn of the thyroid. Schulte and Smith (1918, p. 39), reversing Howell's error, call this as well as the foregoing muscle ster- 7iohyoid. ]8 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Thyrohyoid muscle (Fig. 4, th). This is a thin, paired muscle lying ventral to the strong aponeurotic sheet which connects the thyroid and epiglottid cartilages. The ihyrohyoid takes origin on the side of the thyroid cartilage anterior and ventral to the insertion of the sternothyroid. Thence it passes anter- iorly and a little ventrally, crossing the ventral part of the thyroid insertion of the occipitothyroid, to insert on the posterior margin of the basihyal and on the posteromedial border of the thyrohyal as well. In some instances as in Lagenorhynchus acutus, there is a well developed fibrous sheet across the angle between these two, to which the thyrohyoid also attaches. Schulte and Smith, Howell, and Hein all identify this same muscle. Occipitohyoid muscle (Figs. 3, 8, oh). This is a thin, small muscle which has its origin behind the bulla in the region of the cartilaginous tip (sometimes called tympanohyal) of the stylohyal, whence it passes ventrally across this cartilage to insert on the dorsal tip of the thyrohyal. There is some varia- tion in exact origin, as the stylohyal and the exoccipital are closely bound together and invested with tough fibrous tissue which also covers the bulla. In Delphinns and Lagenorhynchus acutus, the origin was from this tissue where it overlay the stylohyal. In Tursiops, we found it in the tissue between the stylohyal and the bulla. Schulte and Smith (1918, pp. 36, 37, and 38) describe a depressor mandihulae which they homologize with the occipito- hyoideus of Rapp and Stannius. According to them it has its origin in the dense fibrous tissue in the region of the articula- tion of the jaw and inserts in the tip of the hyoid (=thyrohyal) continuously with the mylohyoid. Undoubtedly this is the hom- ologue of our muscle, although the mylohyoid in our dissections did not extend this far laterally, and the origin of our occipito- hyoid was more posterior. Boenninghaus (1902, pp. 61-63, fig. S) describes and figures this muscle, which he considers to be homologous with the hyopharyngeus or constrictor medius. Neither Howell nor Hein mention such a muscle. Interhyoid muscle (Figs. 3, 4, 8, ih). This paired muscle fills the space between the hyal bones on each side. It arises from most of the dorsal surface of the thyrohyal as well as antero- dorsally on the basihyal, and passes anterodorsally to insert around the posterior surfaces of stylo- and ceratohyals. Medially, the two muscles are separated by the hyoepiglotticus. The lateral extent of the stylohyal insertion of the interhyoid shows LAWRENCE AND SCHEVILL : GrLAR MUSCULATURE 19 some variation. In Lacicnorlryncliu^ acufus and Drlphrnus, a small portion attached ventral to the origin of the styloglossus; in Stcitdla plafiiodon the attachment was more medial with almost no overlap with the styloglossus. Schulte and Smith call this muscle ccratohyal or hyoidcus lotus, Howell calls it ceratohyoidcus, and Hein uses hyostylohj/oi- deus but gives intrrhrioidcus as a synonym. Boenninghaus does not refer to it. M. hyo epiglottic us (Figs. 5, (i. he). This is a strong, single muscle. Its origin medially on the dorsal surface of the basihyal separates the left and right interhyoid: it is long anteropos- teriorly, but rather narrow. In Stenclla plagiodon and Lageno- rhynclius acntus, its origin extends also onto the ceratohyal, where it is dorsal to the interhyoid. From its origin the tibers pass posteriorly and a little ventrally to insert on the anterior surface of the epiglottid cartilage about one-third of the dis- tance up from its base. Ventral to this nuiscle between the posterior border of the basihyal and the lower part of the epiglot- tis is a strong ligament. Kernan and Schulte describe a double h yoepiglottic muscle, while Howell and Hein botli tind it single. Boenninghaus figures but does not describe a hyocpifiloftieiis whicli looks like ours. Muscles of the Pharynx In order that the arrangement of the pharyngeal muscles in the odontocetes be understood, the relation of the hyoid a])paratus and the laryngeal cartilages to each other and to the skull needs to be discussed further. In these mammals the larynx is permanently intranarial. Consequently, it lies close to the base of the skull and to the posterior bony nares, into which the arytenoepiglottid cartilages protrude (.Fig. 10). While the thyroid cartilage lies posterior to the thyro- basihyals, the much elongated arytenoepiglottids pass anterodors- ally internal to the hyoid apparatus, between the basioccipital- pterygoid plates and into the nares. These plates, which extend on each side from the exoccipital to the bony nares, project ventrally as flanges between the basicranial region medially and the bulla and prebnllar region laterally. This is a typ- ically odontocete arrangement, as is a corresponding ventral development of the exoccipital posterior to the bulla. Whether or not this modification is primarily to help to isolate the bulla acoustically from the throat region is of peripheral interest for 20 BT'LLETIN : MUSEUM OF COMPARATIVE ZOOLOGY the moment. The signifieant thing is that, coincident with these changes, the tympanohval has shifted its cranial attach- ment to the exoccipital behind the bnlla, and as the exoccipital has moved ventrally, so also has the stylohyal. The shift has also moved the thyrohyal too far ventrally to articulate with the thyroid. Such an articulation is further made impossible by a transverse development of the flattened thyrobasihyals so great that this complex spans the space between the basiocci- pital plates (basioccipital crests of Fraser and Purves, 1960b, p. 24, fig. 7; p. 25, fig. 8). The thyroid cartilage, no longer sujiported by the hyals, is directly attached to the base of the skull by strongly developed paired lateral muscles, the occipito- th 11 raids. Further rearrangements of the muscles of the pharyngeal region are caused by the fact that the pharynx, instead of passing dorsal to the larynx, divides in front of the hyocpiglot- ticus to pass on each side of this muscle and the elongated arytenoepigiottid cartilages (Fig. 10). It passes between these cartilages and the horns of the thyroid on each side and forms a single ])assage again posterior to the arytenoepiglottids and internal to the occipiioihyroid muscle. This arrangement of the pharynx, the position of the hyals, and the attachment of the thyroid to the base of the skull effectively isolate the pharynx from any hyal attachments of the constricting muscles. The bones and cartilages of the base of the skull are not the only structures that have been modified by the requirements of an aiiuatic existence. Because of the permanently intranarial position of the larynx, the functions of those pharyngeal muscles which arc oft;ni called constrictors differ from those of other mammals. Primarily, in the odontocetes, they are concerned with holding the larynx in place in the bony nares. This is accomplished in two ways : by means of a sphincter around the tip of the arytenoepigiottid cartilages, and by drawing the larynx up and forward into the back of the bony nares. Of the muscles involved, two, the palatopharyngcus and the ptery- (jopharyngcus, hold the larynx in place by means of a powerful sphincter. x\ third muscle in this region, the occipitothyroid, binds the larynx firmly to the ventral surface of the skull. The only muscle that has a sole constricting function is the thyreo- pharyngcus. Because of these changes, and without a detailed study of innervation, it has not been possible to homologize with any degree of certainty the muscles of this region with LAWRENCE AND SCHEVILL : GULAR MUSCULATURE 21 those of other mammals. For instance, the sphincter muscle of the posterior nares was found to be made up of two main units, called here the pterygopharyngeus and palatopharyngeus and very probably homologous with those of other mammals. The same is true of the ihyreopharyngius, which might be found to include the cricopharyugeus. Whether or not the two parts of the hyopharyngeiis are represented by the entirely differently placed occipitotJiyroid is uncertain. Two other mammalian muscles normally found in this region, the tensor and levator palati, are either missing or much modi- fied because the soft palate as such has ceased to exist in the Delphinidae, and their function is obsolete. Instead, there extends posterior to the hard palate, for a short distance, a thickened aponeurosis. This is called by some authors (e.g. Kernan and Schulte, 1918) the velum palati; we have followed Fraser and Purves (1960a and b) in calling it the palatine aponeurosis. From its dorsal surface the palatopharyngeus takes origin in part. Laterally, from the tips of the pterygoids to the tissue investing the bulla the palatine aponeurosis is continued as the pterygoid ligament. .1/. styJopharyngeus (Figs. 6, 7, i), sp). This musch' takes origin narrowly on the dorsomedial side of the stylohyal near its cranial articulation and expands somewhat as it passes anterodorsally across the palatopharyngeus, to mingle with it and the ventral border of the pterygopharyngeus at its insertion in the lateral walls of the nasopharynx. It follows closely the course of the eustachian tube which opens into the nares internal to the insertion of the stylopharyngeus. Both Schulte and Smith (1918, pp. 38-39) and Kernan and Schulte (1918, p. 261) describe a similar stylopharyngeus. Howell gives no details of this postnarial region. Fraser and Purves (1960a, p. 68, fig. 85; 1960b, p. 21, fig. 6) call this the levator palaM. Boenninghaus describes and figures (1902, pp. 45-17, pi. 1, fig. 3) a stylopharyngeus in Phocoena with styloid origin as in our stylopharyngeus, but with the two parts joined and lying ventral to the pharjmx. This does not agree with our dissec- tion of Phocoena, in which the stylopharyngeus is paired, and, except for being less distinct from the pterygopharyngeus at its insertion, resembles closely that of the Delphinidae. Lack of knowledge of the Delphinidae and the fact that stylo- and pterygopharyngeus in Phocoena are more completely merged in 22 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY the Avails of the pharynx have probably led Boeniiinghaus to a mistaken identification of the distal ends of these two muscles. His figure 3 of plate 1, which purportedly illustrates this region, has certain errors which are further confusing and arc dis- cussed in more detail below. M. palatopharijnrjcus (Figs. 4-7, 9, pp, ppt). The palaio- pharijngcus as identified here is both complex and not well defined. It is a A'ery thick muscle which has its origin over a large area extending from the anteromedial wall of the bony nares to the pterygoid margin of the palate and across the narrow palatine aponeurosis to the pterygoid ligament and stylohyal. Within the nares (Fig. f)), its attachment is prin- cipally on the pterygoids posterior to the vomer and on each side of a raphe which extends from the vomer to the margin of the palate and separates the two halves of the muscle. At its dorsal margin, the paJaiopharyngcus mingles, with no sharp boundary, with the ventral margin of the ptcriif/opJiarii)ifi(i(s (Fig. 9). For this reason, it is hard to define exactly the extent of the insertion of each on pterygoid and palatal bones. From its origin, part of the palatopharyngeus sweeps more or less horizontally around the tip of the arytenoepiglottid cartilages in a strong sphincter (Figs. 7, 9), the outer layers joining those of the tJii/rcopharyngeus in the mid-dorsal region. Some pass beneath the thgreopharyngeus to join, via tendinous bands, the fibers of the opposite side ; a few end on top of the thyreopharyn- geus. A second part, consisting of the innermost layers of the paliitopharyngeus, forms the arcus palatopharyngeus. A third part of the palatopharyngeus, called the pars thyropaJa- ti)ius, passes posteroventrally to insert between the thyroid and epiglottid cartilages. While the main mass of the muscle is internal, it can be understood best in detail if we examine it as it shows in pro- gressively deeper lateral dissections. When the thyrohyal bone and its attached muscles are cut away, the thyropalatine section of the palatopharyngeus may be seen attaching on the laryngeal cartilages in the region where the oropharynx divides to pass around the arytenoepiglottid cartilages (Fig. 5). Its exact insertion varies somewhat in the genera examined. In Del- phinus, it is on the lower half of the inner surface of the thyroid cartilage, the adjacent part of the pharynx and the lower part of the epiglottid cartilage, with only the suggestion of a division between the epiglottid and more dorsal portions. In Stenella LAWRENCE AND SCHEVILL : GULAR MUSCULATURE 23 plagiodon the insertion is distinctly double. The posterior part is on the pharynx and on the inner face of the body of the thyroid in a narrow vertical band extending ventrally from the anterior border of the notch. The anterior and also more ventral insertion is between the thyroid and the epiglottid, chiefly on the former, but with a few strands to the latter. In Lagcnorhynclius albirostris there are two separate thyroid attach- ments as in Stenella plagiodon, although the muscle is not double. The intervening fibers insert in the tissue between the epiglottid and thyroid cartilages. In this genus attachments to the wall of the pharynx are poorly developed. From the anterior border of the thyropalatine portion of the palaio pilar yngcus a few fibers (not figured) pass forward be- tween the interhyoid and the wall of the pharynx, which here lacks muscle fibers. Some of these fibers end on the dorsal surface of the interhyoid, to which they are loosely bound by connective tissue. Anterior to this and somewhat more dorsally, a few fibers extend on top of the combined genio- and palato- glossus. This is the portion of the palatopharyngeus referred to in our discussion of genioglossus. From its laryngeal insertions, in all three genera, the thyro- palatine portion of the muscle widens as it passes anterodorsally towards the posterior bony nares. A posterior segment of the external portion attaches (Figs. 5, 7) strongly to the stylohyal ; anterior to this, the external layers pass internal to the stylo- hyal (Figs. 5, 6), to which they are loosely bound by con- nective tissue, and attach to the pterygoid ligament. Deep to these external thyropalatine layers, but not distinct from them, the main mass of the palatopharyngeus (Fig. 7) passes up the bony nares internal to the stylopharyngeus. If the skull is so cut away that the nasopharynx can be laid open along the midline posteriorly and its mucous membrane dissected away from the underlying muscle, this latter is ob- served as a single mass indistinctly divisible on the basis of texture and direction of fibers. By far the greatest bulk of this muscle is made up of rather coarse, anteromedially directed fibers which are the combined palato- and pterygopharyfigeus; in this inner aspect there is no real separation between the two. In the midline anteriorly are paired raised bundles of a more finely fibered layer which is probably Boenninghaus ' pars in- terim of the palatopharyngeus. At their upper end the fibers of this layer are as coarse as and indistinguishable from the 24 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY undeiiyinf}^ pfenigopliaryvfjcus with which it mingles as the two converge laterally towards the opening of the eustachian tube. At their lower end the paired bundles separate to encircle the arA^enoepiglottid spout in a strong sphincter muscle, which also is more finely textured than the more external layers with which it merges. In its pharyngeal aspect this sphincter, the arcus palafopharyngeus, is separated laterally from the overlying muscle by deep pocl^ets. The coarser mass external and lateral to the sphincter is probably what is .sometimes called the pars externa of the palaiopliaryngcus which medially, in part, passes ventral to the pars interna to insert in the raphe which is closely attached to the wall of the bony nares. Tn Korjia, Kernan and Schulte (1918, pp. 261-262) describe a palatopharynqens which includes all of the component parts found in our dissections, although they do not separate a pars interna or externa. "What we have called a tliyro palatine portion, they refer to as the "superficial portion of the palato-pharyn- geus." Their emphasis on the unity of this muscle is equally applicable to our dissections. In their palatopharyngcus the blending with pterygopharyngrtis at the insertion of stylopharyn- geus, the involvement in the arcus palatopharyngeus, and origin from velum and pterygoid all parallel the situation in the delphinids. Differences, which include a band of fasciculi pass- ing beneath our thyreopharyngeiis, as well as no mention of any stylohyal attachments, are surprisingly slight between fam- ilies with such markedly unlike crania. In Dclpliinus, Fraser and Purves' palatopharyngeus pars in- terna and externa (1960b, pp. 17-22, figs. 3-6) passes up the nares as does our palatopharyngeus and is probably the same as our muscle, although in their figure 3 the ventral portion, in a somewhat misleading fashion, appears to wrap around the ventral surface of the larynx. Hein's discussion of the pharyn- geal region is incomplete, but his constrictor pharyngeus su- perior is our palatopharyngeus. Connuents on Boenninghaus' Avork will come at the end of this section. M. ptery go pharyngeus (Fig. 9, ptp). This muscle lies dorsal to the palaiopharynge^is and together with it surrounds the tip of the larjnix. Its origin is from the anterior wall of the upper part of each bony naris. When the nasopharynx is removed from the bony nares and the external arrangement of the muscles examined, the stylopharyngeus is seen partially to separate the pterygo- and palatopharyngeus laterally. LAWRENCE AND SCHEVILL : GULAR MUSCULATURE 25 The pierygopharyngcus passes outward as well as postero- ventrally to meet its fellow of the opposite side in an ill-defined raphe in the mid-dorsal line of the nasopharynx. Posteriorly, in the midline, it is separated from the thyrcopharyngeus and the transverse fibers of the palatopharyngeus by a small aponeu- rosis. When the lining of the naso]:)harynx is removed, no separation between palafo- and pterygopJiaryngeus is found. Boenninghaus distinguishes a small paired, medial muscle overlying the pierygopharyngeus dorsallj^ which he calls the salpingopharyngeus. We found no such distinct layer in this region, although, at its dorsal insertion, the most posterior fasciculus of the plerygophuvyng'. iih was distinct i'or a few millimeters. Internally, when the mucous lining was removed, a faintly differentiated longitudinal bundle of slightly finer fibers was observed. This rather finer bundle gradually merged dis- tally and externally with the surrounding coarser fibers and was quite distinct in direction as well as structure from the ventral, internal fiber bundle identified earlier as Boenninghaus' pars inicrna of the palatopharyngeus. Further comments on Boenninghaus will come at the end of this section, and are applicable also to Fraser and Purves, who follow Boenninghaus. Kernan and Schulte describe a pierygopharyngeus which closely resembles ours. Neither Howell nor Hein deal with the nasopharynx. Occipitothyroid muscle (Figs. 4-7, 9, ot). This strong muscle draws the thyroid cartilages, and thus the whole larynx, up towards the base of the skull. Its outer portion has its origin in the fibrous tissue in the region of the bulla and on the margin of the basioccipital plate anterior to the hypoglossal canal. Thence its inner portion extends broadly across the basioccipital to meet its fellow of the opposite side. Its insertion is on the external surface of the anterior horn of the thyroid cartilage and on the adjacent walls of the pharynx. Details appear to vary in the genera examined, though the main plan is as de- scribed above. In Delphinus and Lagcnorhynchus alhirostris some of the posterior fibers wrap around the pharynx, passing on top of the thyrcopharyngeus in so doing. In Tursiops a small slip also takes origin on the tip of the stylohyal cartilage. In Stenella plagiodon the outer and inner sections seem somewhat more distinct at their origins. Parallel dissections were not made of each genus, so that complete comparative comments are not possible. 26 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY In Schulte and Smith (1918, fig. 11, p. 37) the muscle labeled inferior constrictor is probably the outer part of our occiijito- thyroid; in Kernan and Schulte (1918, p. 260) it is the rostral portion of the inferior constrictor (thyreopJiaryngeus) which is equivalent to our occipitothyroid. The attachment on the thyroid cartilage is the same, but there is no mention of any cranial attachment. This is the muscle called by Boenninghaus laryngo- pharyngeus seu constrictor pharyngis inferior (1902, pp. 59-61). Fraser and Purves (1960b, pp. 20, 21, 137, fig. 6) identify this muscle as the constrictor pharyngeus or superior constrictor muscle. ilf. thyreopJiaryngeus (Figs. 5-7, 9, tp). This muscle takes origin on the posterior horn of the thyroid cartilage and passes anterodorsalh^ across the pharynx, medial and internal to the occipitothyroid. The thyroid origin of this muscle is the prin- cipal one in all four genera examined. In Delphinus the origin extends across the cricothyroid articulation as well, and in Lagcnorhyncluis alhirostris a few fibers medial to this take origin on the circular layer of the oesophagus. From its origin, the thyreopharyngeus passes anteromedially across the circular muscle of the oesophagus and the palato pharyngeus to meet the ptcrygopharyngeus. In Delphinus and Lagenorhynchus al- hirostris, an inscriptio intervenes between the insertion of the pterygo- and thyreopharyngeus, though not in TursiojJS and Stenella plagiodon. This is the muscle identified by Kernan and Schulte (3 918, p. 260) as the caudal portion of the inferior constrictor (thyreo- pharyngeus). Interpreting Boenninghaus' text and figures in the light of our dissections of Phocoena as well as of the Del- l)hiiiida(', it seems clear that his longitudinalis oesophagi is the same as our thyreopharyngeus . He gives an anterior cranial attachment for this which we did not find, and which does not seem possible in view of the way the occipitothyroid passes dorsal to the thyreopharyngeus in this region. Hein's constrictor inferior pharyngeus is probably the same as our thyreopharyn- geus, although it is too incompletely described for certainty. Homologies with Phocoena. Boenninghaus in his careful work (1902) on the phar.ynx in Phocoena makes a diligent effort to find and identify the various muscles found in land mammals, especially man. This emphasis on homology leads to the recog- nition of more layers than are apparently distinct in the delphin- ids, and is one of the reasons why he found it impossible to LAWRENCE AND SCHEVILL : GULAR MUSCULATURE 27 relate his work to Murie's and to Macalister's on Glohicephala. Our dissection of Pliocoena showed an arrangement of the pharyngeal muscles very similar to that found in the delphinids, and one which did not always match Boenninghaus' plates. For this reason Boenninghaus' discussion of the pars superior of the pharynx needs to be considered in some detail. In his introductory comments, Boenninghaus himself emphasizes that the muscle of the pars superior is essentially a single mass (1902, pp. 38-41), exactly as we found it to be. If this is not kept well in mind, his subsequent division of it into as many parts as he does is misleading. Primarily, he identifies a constrictor of the pars superior pharyngis which he says is made up of three distinct muscles in terrestrial mammals: the ptertjgopharyngeus, palatopharyngeus (pars externa and interna) and the thyreopalatinus. "We con- sider the pterygopharyngeus to be better dcA'cloped than does Boenninghaus, and restrict the palatopharyngens to the more ventral part of this musculature. This discrepancy between the limits of his pterygo- and palatopharyngeus and ours stems from his effort to separate the two on the basis of pterygoid or palatal attachments. Since the relationship of these bones in Fho- coena is different from that in the delphinids, while the relation- ship of the two sections of muscle is much the same in both, it is better to follow an apparent though slight natural division in the muscle mass itself. In the phocoenids, the palatal bones intervene medially be- tween the pterygoids and extend to the rim of the bony palate, but in the delphinids these bones are broadly separated from the margin of the bony palate by the pterygoids. Thus, if identification were to be based on muscle attachment only, the positions of pterygo- and palatopharyngeus would be re- versed between delphinids and Phocoena. Actually, as described above, a distinct though not pronounced separation between the two muscles (Fig. 9) occurs approximately at the level of the insertion of stylopharyngeus. Since in other mammals (cf., Sisson, 1917, p. 407) the stylopharyngeus inserts in the wall of the pharynx between pterygo- and palatopharyngeus, the sep- aration is best made here. Thus the more posterior part of our pterygopharyngeus is equivalent to part of Boenninghaus' pars externa of palatopharyngeus, as well as his associated salpingo- pharyngeus. 28 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY While it is possible to correlate Boeuuiughaus ' pterygo-, palato- and thyreopharyngeus with the muscles observed by us in Pho- coena, his text and especially his figures for salpingo- and stylo- pharijngeus and the levator veli are confusing. If his figure 3 (pi. 1, 1902) is an anterior, slightly ventral view with the oropharynx cut away at the entrance to the sinus pyriformis, then the stylopharyngeus is wrongly identified ; the medial muscle anteroventral to the epiglottis is the pars externa of the palato- pharyngeus which, lateral to its insertion on the dorsal surface of the palatine aponeurosis, attaches also to the stylohyal. The stylopharyngeus lies dorsal as well as a little lateral to this and fol- lows the course of the eustachian tube into the bony nares. At its upper end its course is much like that figured by Boenninghaus (1902, fig. 3, no 2) for salpingopliaryngcus. We found that the stijlopharyngem of Phocoena differs from that of the Delphinidae in merging more completely with pterygopliaryngeus, so that it was impossible to tell, in the region of the eustachian tube, which slips were pterygo- and which stylopharyngeus. At its ventral end, Boenninghaus (1902, p. 42) says his sal- pingopharyngeus meets in an inscriptio the longitudinalis oes- phagi (= our thyreopharyngeus). His description of the relation between these two leaves little doubt that his salpingopharyngeus is a slip which we could not separate from the main mass of the pterygopliaryngeus either in the delphinids or in our adult Phocoena. As far as its insertion at the mouth of the eustachian tube is concerned, in Phocoena the stylo- and pterygopharyn- geus were completely merged in this region. In the delphinids, where the insertion of the stylopharyngeus is more discrete, this seems to be the muscle surrounding the opening of the tube. Boenninghaus' levator veli {^levator palati) is that part of pterygopharyngeus with which the pars interna of palatopharyn- geus merges. The difference in fiber direction as illustrated in his plate 1, figure 2, layers 1 and 3, does not indicate distinct layers, but, as shown in our dissections, merely a gradual curving around and converging of a single muscle mass towards its dorsal insertion. At the mouth of the eustachian tube we did not find the discrete insertion figured by Boenninghaus. For these reasons, as well as because Boenninghaus himself empha- sizes the lack of separation of the muscles he discusses in this region, we have not considered levator veli to be a separate muscle. No tensor veli (= tensor palati) as described by Boen- ninghaus was found. Fraser and Purves, although generally LAWRENCE AND SCHEVILL : GULAR MUSCULATURE 29 following Boennmojhaiis, do not agree with his identification of the tensor palati and believe (1960b, pp. 11, 19, fig. 4) that it lies lateral to the pterygoid bones and thus external to the area with which the present paper deals. Discussion The highly specific modifications of the gnlar region in the odontocetes are for the most part centered around changes in form and in position of the larynx, changes that are directly related to the needs of an aquatic environment. The advantages of an intranarial larynx in terms of underwater feeding are apparent ; to what extent either the intranarial position or the form of the larynx is also an advantage in phonation is not yet explained. In any event the laryngeal cartilages both enclose and pass through the pharynx instead of lying for the most part ventral to it. The elongated arytenoepiglottid cartilages, on each side of which the oropharynx passes, have already been de- scribed. Ventral, as well as anterior to these, lie the postero- ventrally extended palatal and pterygoid bones. A deep air sinus divides these into inner and outer laminae, the inner forming the anterior wall of the bony nares while the outer extends the roof of the mouth and hence the pharynx downward as well as posteriorly almost as far as the larynx. Together these bones separate oro- and nasopharynx. The posterior ex- tension of the palate, such that a purely ventral movement of the larynx cannot disengage it from the posterior bony nares, helps to insure the intranarial position of the larynx. This development of the palate also directs away from the posterior nares the food being swallowed. The rather complicated rearrangement of the pharyngeal muscles to insure the intranarial position of the larynx may seem confusing in terms of homologies and comparisons with other mammals, but functionally is very simple. The thyroid cartilage, having lost its connection with the thyrohyal, is suspended from the base of the skull by a very strong muscle, the occipitothyroid. Lateral to the epiglottis this muscle also draws the walls of the pharynx apart. Anteriorly and quite distinct from this muscle is the palatopharyngeus, which lifts the anterior part of the larynx up and forward into the bony nares. The laryngeal attachments on each side are near the ventral part of the larynx, and the muscle passes external to the pharynx. Both the occipitothyroid and the palatopharyngeus have an 30 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY outward as well as an upward pull which insures that the pharynx is not squeezed shut where it passes on each side between the arytenoepiglottids and the thyroid. In addition to providing a strong muscular connection be- tween the larynx and the skull, part of the palatopharyngeus forms a complete sphincter around the expanded tip of the arytenoepiglottids, which further helps to hold the larynx in position in back of the nares. While the primary function of these muscles has become a supporting one, the palatopharyngeus is also involved in shorten- ing the pharynx during swallowing. Contraction of that part of the muscle which has its origin in the anterior wall of the nares and which surrounds the pharynx external to the arcus, not only grasps the arytenoepiglottids more tightly, but it also draws the dorsal wall of the pharynx anteriorly. Posterior to the arytenoepigottids, where the pharynx is again a single pas- sage, the thyreopharyngeus is, as usual, a constrictor. The func- tion of the stylopharyngeus is to spread the walls of the naso- pharynx external to the arcus. It is apparently also involved in opening the eustachian tube. Powerful as are the muscles holding the larynx in place, there are no opposing muscles to exert the posteroventral force which would be needed to withdraw the larynx from the back of the bony nares. The only laryngeal muscles with a posterior pull are the poorly developed sternothyroids. Tension on these spreads the wings of the thyroid cartilage rather than moving the entire larynx. Ventral to the larynx, and also with a posterior pull, lies the strongly developed sternohyoid. Tension on this moves the hyals only slightly, with no corresponding shift of the larynx. Anteriorly, there is only one extrinsic laryngeal muscle, the hyoepiglotticus. The action of this is to move the epiglottis away from the arytenoids, thus opening the tip of the larynx. Obvi- ously, when the larynx is opened it will not at the same time be withdrawn from the nares, and the anterior slightly dorsal direc- tion of the fibers from epiglottis to hyals ensures that this will not happen. Kecent authors (Lilly, 1961; Brown, 1962) who have assumed that the larynx in odontocetes can be and is withdrawn from the nares, have done so on the basis of certain phenomena which they wished to explain, and not on the basis of the anatomy involved. Lill}' (1961, p. 237), discussing Tursiops truncatus, LAWRENCE AND SCHEVILL : GULAB MUSCULATURE 31 states, "The larynx can also be disengaged from the naso- pharynx entirely by the throat muscles connected to the hyoid bone and dropped down in order to discharge water into the mouth from the upper bronchial tree and trachea." Aside from the improbability of such a need, the absence of a thyroid- thyrohyal joint, the lack of a suitable muscle connection between larynx and hyals, and the position of the thyrobasihyals anterior to the larynx, all prevent the muscles acting on the hyoid from moving the larynx with it. Further, the position of the strongly developed sternohyoid immediately ventral to the larynx inhibits any downward movement of the larynx, especially when this muscle is contracted as it would have to be if the hyoid were being pulled posteriorly. In one of his interesting and informative notes on behavior and pathology of captive cetaceans, Brown (1962, pp. 63-64) at- tributed the death of a captive Glohicephala scammoni to "laryn- geal occlusion induced by inspiration of a stone," although the stone was not found in the animal. It was picked up after having fallen out of the severed head and subsequently inter- preted as the cause of a laryngeal lesion. He supposes the fol- lowing sequence of events : the stone was swallowed ; the animal vomited, retracting the larynx as it did so; the stone then passed up into the nasopharynx ; the larynx was reinserted in the nasopharynx and the stone inhaled. There are a number of anatomical difficulties with this explanation. The lack of muscu- lature for retracting the larynx has already been discussed. "Within the larynx, the paired arytenoids lie very close to each other, and to the epiglottid at their base, narrowing the passage in this region to a slit. When the larynx is opened, this is done by the hyoepiglotticus, which separates the cartilages at their tips; there is no additional musculature to enlarge this slit -like passage. Between the epiglottid and the arytenoids at their base, and somewhat ventral to the main air passage, is a distensible pouch. A large object passing down the larynx might lodge here, and it is difficult to see how it could then be moved down the main air passage. As evidence that laryngeal occlusion has occurred, lesions of the floor and side walls of the larynx in the region of the thyroid are offered. This is the region where the pharynx passes on each side of the arytenoepiglottids, separating the thyroid from the rest of the larynx and trachea. What seems more likely is that the stone, lodged in the pharynx here, may have been responsible for the lesions. 32 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Brown further states that a Tursiops whose jaws were forcibly- held apart convulsed, vomited, and then expelled vomitus through his blowhole. Though Brown believes this is evidence of laryngeal retraction, it is far more likely that the vomitus was passed up the nares between the sphincter and the aryteno- epiglottids, as is anatomically entirely possible for a liquid. Brown further theorizes that withdrawing the larynx may be a means of facilitating "the passage of large items of food." Actually, even if the larynx is artificially withdrawn from the nares, the length of the arytenoepigiottids is such that they still project across the pharynx (Fig. 11). With the larynx in place, the highly stretchable pharynx on each side of the epiglottis can pass any mass of food that has passed the isthmus faueium and that can be accommodated by the oesophagus. This is pos- sible because of the vertical expansion of the passage and the lateral spread of the horns of the thyroid. In a dead animal, it is perfectly possible to disengage the tip of the larynx by reaching up the throat and into the nasal passage with the fingers. In a specimen in which the pharyngeal muscles had not been cut, pulling the sternothyroid muscle as strongly as possible either ventrally or posteriorly did not with- draw the larynx from the nares, nor did tilting the larynx force the tip down and so out of the nares. The possibility of jaw motion moving the larynx was also in- vestigated. The attachments of the digastric and other anterior throat muscles to the hyals and the independence of the hyals from the thyroid make this seem unlikely. In addition, prying the jaw of a dead animal open as widely as possible produced no detectable shift in position of the larynx. Adapted as the musculature of this region is to holding the larynx in place intranarially without interfering with the swal- lowing mechanism, it is not surprising that the larynx is only with difficulty dislodged. In the absence of any morphological arrangements to accomplish this and since there is no need for the animals to do so, it seems clear that the odontocetes do not voluntarily move the larjaix in and out of the nares.^ Aside from studying the hyo-pharyngeo-laryngeal region as a functional unit, perhaps the most important result of our dissections has been our ability to reconcile apparently very dis- similar dissections by earlier workers, and to establish that in the five principal odontocete families the arrangement of the muscles and cartilages in this region is remarkably similar. Few 1 See note on page 35. LAWRENCE AND SCHEVILL : GULAR MUSCULATURE 33 investigators of cetacean anatomy have studied this region in detail. The difficulties involved in the necessary deep dissec- tions have caused most authors, preoccupied as they usually have been with the general anatomy of a particular species, to limit their attention to the more superficial throat and hyoid muscula- ture. Others, who have studied the larynx in detail, have illus- trated and described a larynx already severed from the skull. The most complete account of the throat region in the Delphinidae is still that of Glohicepliala published by Murie some ninety years ago. This is discussed in detail in the second section of this paper. Murie 's earlier work on Lagenorhynchus (1871) only touches on the region studied here. In this earlier paper, the function of the musculature of the posterior nares is dis- cussed (p. 144), but the individual muscles are not described. Of the hyoid muscles, two are considered in detail, the hyo-epi- glottic which is the same as ours, and the hyoideus which equals our interhyoid. Certain other muscles are figured but not de- scribed. Of these, his geniohyoid (pi. 5, fig. 8) looks like our genioglossus and his genio-hyoglossus like our hyoglossus. We have not been able to identify his levator palati. Kesteven's work (1941, pp. 78-79), which treats of some of the tongue and hyoideal muscles in Delphinus dclpliis, is of particular interest in that he describes the innervation of these. His jugulo-hyoideus {— our occipitohyoid) is supplied by the glossopharyngeal nerve, which supports Boenninghaus ' belief that this muscle is homo- logous with the constrictor medius rather than the digastric. His genio-hyoid, genio-glossus, and stylo-hyoideus {= our interhyoid) resemble ours. His description and figures of stylo-glossus and hyo-glossns are reversed ; the long narrow muscle which he calls the hyo-glossus should have its origin at the end of the stylohyal and is the styloglossus; the more medial muscle should have its origin on the hyals and is the hyoglossus. His stylo -pharyngeus and superior constrictor are not readily identifiable. More complete accounts have been published of some of the other odontocetes. For the Phocoenidae, Boenninghaus' account (1902) of the throat of Flwcoena deals in careful detail Avith the same region we have studied, and goes considerably further than we have in discussing development and homology. The lack of agreement of some of his plates with our dissections of delphinids led us to check his findings on an adult Phocoena. The discrepancies were mostly in the pharyngeal region and have been analyzed earlier. Briefly, it may be said that, where 34 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY differences occurred, the arrangement of these muscles in Fho- coena was more like that in the delphinids than like Boenning- haus' plates. Subsequent to Boeuninghaus ' work but without commenting on it, A. B. Howell (1927) described the myology of the neck and throat, exclusive of the pharynx, of Neomeris, another phocoenid. For this reason a synonymy of his muscula- ture has been included. The most recent comprehensive studies of this region in the Physeteridae are those of Kogia by Schulte and Smith (1918) on the muscles of the neck and throat, and by Kernan and Schulte (1918) on the muscles of the pharynx. Both of these studies have already been related to our work. Since these workers themselves have amply commented on the findings of earlier investigators, we have not repeated their comparisons. For the Monodontidae, Hein's account (1914) of Monodon monoceros is the most pertinent. Though it is primarily of the larynx, it includes detailed descriptions of the interramal, tongue, and hyoideal muscles, and a synonymy of his names has been incuded in our text. His observations on the pharyn- geal region, incomplete as they are because the hyolaryngeal complex had been cut away from the skull, suggest an arrange- ment very similar to that found by us. "Watson and Young's earlier paper (1880) on Delphmapterus leucas includes the same interramal, tongue, and hyoideal muscles that we found, with the exception of the mylohyoid, occipiiohyoid, and hyoepiglot- ticus. Of these, what they identify as mylohyoid is our digastric, while the other two as well as our mylohyoid are lacking. Howell (1930) described the interramal and throat musculature of Monodon as resembling that of Neomeris and uses the same names for muscles in both papers. A recent rather hasty examination of Ziphius (Ziphiidae) suggests to us that it, too, is much the same. The intranarial larynx, and its associated structures, seems to be such a successful adaptation for aquatic life that, once ar- rived at, it has varied little. Presumably this was a modification that occurred early in this adaptation. The arrangement of the structures involved is so peculiar and so different from that in land mammals that it seems unlikely to have been developed re- peatedly. Those familiar with cetaceans will know that the families involved diff'er so widely in skull and external characters that the similarity in the myology of this region is all the more remarkable. LAWRENCE AND SCHEVILL : GULAR MUSCULATURE 35 NOTE ADDED IN PRESS In support of his theory that * ' during feeding and swallowing the larynx is freed, laid in the bottom of the pharynx and food passed over it ' ' (in "Animals in aquatic environments: adaptations of mammals to the ocean," Chap. 46, Handbook of Physiology — Environment, Amer. Physiol. Soc, p. 744, 1964), Dr. John C. Lilly (personal communication 9 October 1964) reports of Tursiops that after he has forcibly disengaged the larynx by hand, the animal re-engages it without assistance. While this observation is good evidence of the effectiveness of the muscles drawing the larynx into tlie back of the bony nares, it does not also show that the animal itself can voluntarily disengage the larynx. It does suggest that much of the time the muscles lifting the larynx are more important than the sphincter muscles in holding it in place. This is as would be expected, because contraction of the sphincter muscles holds the tip of the larynx shut, and during breathing and probably during phonation it needs to be open. In swallowing, however, the sphincter muscle would seem to be the more important, for contraction of this around the tip of the epiglottis would seal off the nares from the food passage, which here is directed somewhat dorsally as well as posteriorly, and so passes diagonally across the entrance to the bony nares. The atypical position of the passage, more posterior than ventral to the nares, and the absence of a soft palate, make some such arrangement for keeping the food out of the nose essential. Dr. Lilly also writes (ibid.), "there is not room between the larynx and the bone of the lower jaw to pass food of the size which these animals Qormally swallow. . . . during feeding, one can palpate the throat region and find the larynx being pulled downward and pressed outwards during the swallowing act. ' ' Actually the larynx lies postei'oventral to the jaw (see Figs. 4 and 10) so there is no constriction of the passage between jaw and larynx. Prob- ably what was felt were the ceratohyals which lie anterior to the larynx and near the external surface. Because of their close connection with the rest of the hyoid apparatus to which the tongue muscles are attached, move- ment of these during swallowing is to be expected. Posteriorly the larynx itself lies deep in the neck, internal to the thick sternohyoid. That the larynx could not be "laid in the bottom of the pharynx" is clear also because the interrelations of aryteno-epiglottids, thyroid, and pharynx make the postulated action an impossibility. The thyroid and ary- teno-epiglottid cartilages are firmly attached to each other (see Fig. 17) so that if they are moved ventrally they move as a unit. Further, since the pharynx itself is enclosed on each side between the thyroid and aryteno- epiglottids (Figs. 10, 11), and posterior to this the thyreo-pharyngeus wraps around the pharynx (Figs. 5-7, 9, 16), if the tliyroid moves ventrally, it takes the pharynx with it. Dr. Lilly's observations would seem to show that internal jjressure in the food passage may push the whole larynx ventrally unless this pressure is counteracted by the occipitotJiyroid muscle. Moving the larynx ventrally in this manner would not pronate 36 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY the aryteno-epiglottids. It is not a muscular action and would be invol- untary as far as the porpoise is concerned. Aside from the anatomy involved, it seems unlikely that the intranarial larynx, an arrangement of such patent utility to an air-breather eating under water, would be undone and by-passed just when it is most needed. EXPLANATION OF FIGUEES The drawings are all of actual dissections of single individuals, except for Figures 8 and 9. That part of the skull shown by broken lines was not drawn from the dissection but added later for purposes of orientation. Figures 1-7 are of progressively deeper dissections of a specimen of Dclphinus delphis. Figures 8 and 9 are diagrammatic in that they are based primarily on a dissection of Lagenorhynchus albirostris, but have certain details added from a subsequent dissection of L. acutus. Figures 10 and 11, of the laryngeo-pharyngeal region, are photographs of Stenella Styx, but are characteristic of all the Delphinidae. The rest of the figures are all of Gloiicephala melaena. Figures 12-15 are of progressively deeper dissections of a single individual; Figures 16 and 17 are of two other indi- viduals. ABBREVIATIONS USED IN THE FIGURES MUSCLES al auricidolahialis ca cricoarytenoid do circular layer of oesophagus ct cricothyroid d digastric gg genioglosstis gh geniohyoid h hyoglossus he hyoepiglotticus 111 hyoglossus lateral portion hm hyoglossus medial portion ill. interhyoid mil mastohumeralis mps masseter, pars superfidalis my mylohyoid oc orbicularis oculi oh occipitohyoid or orbicularis oris ot occipitothyroid pg palatoglossus pni. pectoralis major pp palatopharyngeus ppt palatopharyngeus, pars fhyropalatinus ptp pterygopharyngeus sc scalenus sg styloglossus sh sternohyoid sin sternomastoid sp stylopharyngeus spf sphincter colli profundus spt sphincter colli primitivus st sternothyroid t temporalis th thyrohyoid tp thyrcopharyngeus LAWRENCE AND SCHEVILL : GULAR MUSCULATURE 37 STRUCTURES OTHER THAN MUSCLES A arytenoid cartilage NP AC auricular cartilage 0 AM angle of mouth OP B blubber P BC buccal cavity PA BH basihyal bone PTL BU bulla R C cricoid cartilage S CH ceratohyal cartilage SH CJ condyle of jaw T CO occipital condyle TH E eye TO EP epiglottid cartilage TPH F fat TR FL flipper W II humerus X J jaw (in each instance, right) Y ]\I meatal tube and investing fibrous tissue Z ML melon ZP nasopharynx oesophagus oropharynx pharynx palatine aponeurosis pterygoid ligament raphe sternum stylohyal bone thyroid cartilage thyrohyal bone tongue tympanohyal cartilage trachea see explanation, Fig. 10 see explanation, Fig. 10 see explanation, Fig. 10 zygomatic arch zygomatic process of squamosal bone FIG. 1 Delpldnus delphis Figure 1. Ventral view of gular region with blubber cut away to show superficial layer of muscles. 38 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY Belphinus delphis Figure 2. Ventrolateral view with sphincter colli profundus and most of the peetoralis major cut away. Of the sphincter colli pnmitivus, only the part associated with the ear is shown; the small ear muscle posterior to this was not identified. Figure 3. Ventrolateral view of interranial and tongue muscles. The mylohyoid has been entirely removed. The geniohyoid and digastric have been folded back at their insertions to show the underlying layers. LAWRENCE AXD SCHEVILL : GULAR MUSCULATURE 39 ";^ M spt FIG. 2 mh FIG. 3 40 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Delphinus delphis Figure 4. Ventrolateral view of the hyoideal muscles. The occipitdhyoid has been cut away, causing the tip of the thyrohyal to sag ventrally. The origins of the hyoglossus, styloglossus and genioglossus have also been removed. Figure 5. Same view as Figure 4. The thyrohyal bone and the external muscle layers have been removed to show the deeper layers of the hyolaryn- geal region, and the hyoid cartilages have been drawn away from the base of the skull. The pharynx has been dissected away from the bony palate, and pharynx and tongue folded down as a single mass. The zygomatic process of the temporal bone, and the frontal bone dorsal to the orbit, have been partly removed. LAWRENCE AND SCHEVILL : GULAR MUSCULATURE 41 FIG. 4 FIG. 5 42 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Delphinus delphis Figure 6. Same view as Figure 4. The tympanohyal has been cut away from the occipital posterior to the bulla, and the stylohyal pulled ven- trally and rotated outward to show the styloplmryngeus going into the posterior bony nares, as well as the attachments of the palatopharyngeus. Part of the jaw has also been removed. The right side of the pharynx is shown passing lateral to the epiglottis. Figure 7. Details of the palatopharyngeus. All of the hyals except part of the stylohyal have been removed. The cranial attachment of the oceipitothyroid has been cut and the larynx has been pulled ventrally. In order to show the inner layers of the palatopharyngeus which surround the arytenoepiglottid cartilages, the pars thyropalatinus has been cut away from the pterygoid ligament and this whole outer part folded down. For the same reason the styloplmryngeus has been cut near its insertion and folded up. Part of the thyroid cartilage has also been removed to show the laryngeal attachments of the pars thyropalatinus. The main mass of the palatopharyngeus may be seen passing into the bony nares. LAWRENCE AND SCHKVILL : GULAR MUSCULATURE 43 Fl FIG. T 44 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Lagenorhynchus albirostris and L. acutus Figure 8. Ventral view of hyoid apparatus and tongue muscles. The more superficial layers are on the lower half of the figure ; the area of insertion of the mylohyoid is shown on this side only and, on this side, the origin of the hyoglossiis has been cut away entirely to show the underlying interhyoid. In the upper half of the figure the deeper layers are shown; parts of the stylo- and hyoglossus have lieen removed and both muscles folded outward to show their relation to palato- and genio- glossus at their insertion in the tongue. Figure 9. Nasopharynx and the entire laryngeo-pharyngeal region, in- cluding bones and cartilages, cut away from the skull. The relation of stylo-, palato-, and pterygopharyngeus at their insertion in the walls of the nasopharynx is shown here. The approximate outline of the larynx within the nasopharynx is indicated by broken lines. The approximate boundary internally between palato- and pterygopharyngeus is shown by tlie straight broken line. LAWRENCE AND SCHEVILL : GULAR MUSCULATURE 45 ,TPH FIG. 9 ptp 46 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY 5!1 "^ '■^ -(H o 03 • iH r— ( 1 — 1 - +^ . ,£3 o a -4— ho '3 03 03 "to 03 o o -Q h A 03 -^ -u +-1 f— 1 bis ;-4 bti =3 o s o 03 o -1-H (1 03 4^ <» ^ ■*J O te > 3 PI 0) <0 03 O O ft P "k 43 ^ B 43 C be 03 ■1^ 03 3 o '3 00 g ■ rH bi) (3 •M > a xi c3 05 03 !3 1— < _o M .r-t a P o OJ M c5 'S 03 o S ;^, -+J ^ f^ -c 5 -f^ S •^ P4 a; o O bf O o -*— e3 • rH ft oc • r-t Xi o m CO 09 M .g ^H to *^ -M be c3 _^ ,_, 60 .a V4 n3 "3 ^ ^ o CO 0) o 03 o o "k o tn c; 'fS 03 03 a [J' 0) 0 -4-J 03 o 'o ^ 4S '•4^ ^ o +j ^ o >^ -M ^ (D -C c;_| o "cS in -M O ft • iH O o • rH c5 (13 CD .1^ £ 03 o 03 Xi -rH ft o a> 1 — 1 IB o ;h c5 ^ es =1 bD ■^ CD 03 ft O 03 Pi o Xi ■*^ o o « 03 03 ft S ft O r— 1 C3 03 tT o ro =4-1 03 .a o o 03 -rH o 00 'c3 ■*^ PI 03 > o 03 03 «— H 1—1 be 03 c3 ft •rH "3 03 ,2 O ft 03 53 H-J o Ph 'P. m 03 X3 ^ H o ft 03 LAWRENCE AND SCHEVILL : GULAR MUSCULATURE 47 48 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY o in o OI bo • r-f o3 o -l-t the small, smooth-eyed, hairy-bodied species (subgenus Rhah- depyris) differ from Laelius only in their more complete wing venation ; and the larger, smooth-eyed, smooth-bodied species (subgenus Chlorepyris) closely approach Epyris and other gen- era with paired pits on the scutellum. It is important that the systematics of Rhaidepyris be well understood if progress is to be made with the remaining genera of this complex. I do not have enough material to treat the Old World fauna at this time, but I present here a preliminary review of the 41 known American species. The generic diagnosis and key for separation from other Epyrini which I presented in 1964 (Bull. Mus. Comp. Zool., 132: 91-96) will suffice for present purposes. A review of the structure and terminology employed in this series of papers will also be found there, and the acknowledgments and sources of material listed there may also be taken to apply to the present paper. An alphabetical listing of the abbreviations used for body structures and for museums will be found at the conclu- sion of this paper. Also appended are a check list and an index of the American species. Although in 1964 I placed Trichotepyris and Chlorepyris in the synonymy of Rhahdepyris, I now find it convenient to use these names for subgenera. The types of Trichotepyris and Rhahdepyris, sensu strict o, are Palaearctic species which I have not seen. It seems probable that our species are consubgeneric with the Palaearctic species, but I cannot be certain of this. Kieffer used the name Chloreprjris to apply to species with paired scutellar pits connected by a narrow groove, but as so defined Chlorepyris presents no real morphological gap from such species of Rhahdepyris as, for example, origenus Kieffer. 1 Supported by a grant from the National Science Foundation, GB-1544. 70 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Chlorepyris is here redefined in a broader sense, to include all glabrous-eyed species in which the body setae are short, pale, and subappressed. It is difficult to know which of the three subgenera should be regarded as most primitive ; actually each is rather generalized but has at least one specialized feature. All of our species of Rhahdepyris, scnsu stricto, are small and black, while both Trickofepyris and Chlorepyris show trends toward larger size, bright metallic colors, and blotching on the wings. In the latter subgenus one finds unusual modifications of body sculp- ture, including a tendency for the scutellar groove to be widened into pits on the sides and reduced to a thin line medially. In all three subgenera the wing venation is much alike (see fig. 63 in Evans, 1964) ; I have therefore made no mention of venation in the keys and descriptions. The male genitalia are also relativelj^ uniform in structure. They are not wholly without characters, but the differences seem to me too small to justify the difficulties in extracting and mounting these minute structures. KEY TO SUBGENERA OF EHABDEPYRIS 1. Eyes covered with short setae ; males with antennal segment three reduced to a small ring-joint closely consolidated with four (Fig. 18) B. TRICHOTEPYRIS Kieffer Eyes glabrous; males with antennal segment three of variable length, sometimes short and transverse but always well separated from fourth segment (Figs. 19-24, 71-73) 2 2. Body and major wing veins clothed with coarse, fuscous setae; pronotum short, its posterior margin paralleled by a foveolate groove; small, black species A. EHABDEPYRIS Kieffer Body and wing veins with only fine, pale setae; pronotum fairly long, its posterior margin not paralleled by a foveolate groove; small to moderately large species of black or metallic green or blue coloration C. CHLOREPYRIS Kieffer A. Subgenus Khabdepybis Kieffer Rhahdepyris Kieffer, 1904, Bull. Soc. Hist. Nat. Metz, (2)11: 32 (type species: R. myrmecophilus Kieffer 1904; designated by Kieffer, 1906). Suhgeneric characters. — Small species of black coloration, without metallic reflections ; eyes glabrous, but body and major wing veins clothed with rather large, subappressed to suberect black setae ; scape and legs with shorter black setae ; middle EVANS : REVISION OF RHABDEPYRIS 71 tibiae without spines. Mandibles small, in females with two sharp apical teeth and a series of three smaller teeth basad of these (sometimes indistinct), in males with five strong teeth; base of mandibles far removed from bottoms of eyes, the malar space greater than width of mandibles at their base; antennal scrobes not margined; males with third antennal segment much longer than second, nearly as long as fourth segment. Pronotum rather short, sloping strongly to the collar, its posterior margin paralleled by a f oveolate groove ; notauli often not reaching anterior margin of mesoscutum ; scutellar groove strong ; pro- podeal disc only slightly (1.1-1.3 x) wider than long, with five to nine discal carinae, transversely striate, the side-pieces never striate but more or less alutaceous or beaded. Meso- pleurum rather irregularly ridged and pitted, the foveae not clearly formed, the lower fovea, when discernible, divided into two by a vertical or oblique ridge. Claws very weakly dentate, the tooth tending to slope outward. Remarks. — I have not seen the type species, myrmecopJiilus, and consequently use Rhahdepyris as a subgeneric name rather tentatively for this complex. The six known American species have all remained undescribed up to the present time. I am familiar with the males of only two of the six species. KEY TO SPECIES OF SUBGENUS KHABDEPTRIS Females 1. Front angle of ocellar triangle less than a right angle (as in Fig. 1) 2 Front angle of ocellar triangle a right angle or slightly greater (as in Fig. 2) 4 2. Coxae and femora wholly bright nifo-castaneous; propodeum with nine discal carinae ; front femora robust, about 2.3 x as long as wide 1. mellipes n. sp. Coxae and femora wholly brown or black ; propodeum with seven discal carinae ; front femora not quite as robust, measuring 2.5-2.7 x as long as wide 3 3. LFW 2.2 mm; scape pale castaneous; antennae elongate, third segment about 1.0 X as long as wide; scutellar groove rather thin and shallow 2. huachucae n. sp. LFW 1.6-1.8 mm; scape blackish except paler apically; antennae short, third segment 1.0-1.2 x as long as wide; scutellar groove relatively wider and deeper 3. muesehecTci n. sp. 4. Head rather long and vertex much produced above eye tops ; WH 0.93 X LH; WF 1.44 x HE; OOL 1.35 x WOT; a larger species. 72 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY LrW 2.0 mm 4. gracilis n. sp. Head more nearly circular in anterior view, the vertex only moderately produced above the eye tops (Fig 2); WH 0.98-1.00 x LH; WF not over 1.25 x HE; OOL less than WOT; very small species, LFW under 1.8 mm 5 5. Front very narrow, WF 0.90 x HE; third antennal segment wider than long; scutellar groove wider on the sides than medially; front femora 2.4 X as long as wide 5. minutulus n. sp. Front wider, WF 1.23 x HE; third antennal segment longer than vride (Fig. 2); scutellar groove arching, not wider laterally than medially; front femora 2.9 x as long as wide 6. nigriscapus n. sp. Males Front angle of ocellar triangle less than a right angle; head considerably wider than high 3. muesehecTci n. sp. Front angle of ocellar triangle approximately a right angle; WH/LH = 1.0 4. gracilis n. sp. TABLE I. SUMMARY OF SOME CHARACTERS OF FEMALES OF SUBGENUS RHABDEPYRIS Species Locality LFW (rrim . ) WH/LH WF/HE OOL/WOT Propodeal carinae Front angle ocellar triangle (degrees, approximate) 1. mellipes Fla. (type) 2.0 1.00 1.15 1.33 9 70 2. huachucae Ariz, (type) 2.2 0.92 1.06 1.15 7 80 3. muesebecki Honduras (type) 1.8 1.01 1.12 1.30 7 70 "Mexico" 1.7 0.98 1.27 1.30 7 80 Guerrero 1.6 0.95 1.12 1.36 7 80 Costa Rica 1.8 0.96 1.09 1.15 7 80 Bolivia 1.7 1.00 1.09 1.16 7 80 4. gracilis Calif, (type) 2.0 0.93 1.44 1.35 7 90 5. minutulus Peru (type) 1.3 1. 00 0.90 0.80 7 90 b. nigriscapus Argentina (type) 1.7 0.98 1.23 0.87 9 100 1. Rhabdepyris (Rhabdepyris) MELLIPES new species Holotype. — $ , FLORIDA : Orange Co., 22 March 1930 (J. E. Sadler, Fla. Fruit Fly Trap Survey) [USNM, No. 67,535]. Description of female type. — Length 3.0 mm ; LFW 2.0 mm. Black ; palpi and mandibles testaceous, the latter with the teeth rufous; antennae pale castaneous, the flagellum somewhat dull; tegulae testaceous; legs pale rufo-castaneous except the front and hind coxae somewhat infuscated ; wings subhyaline. Clypeus obtusely angulate, more sharply angled on the midline, the median carina strong, arched in profile. WH/LH = 1.0; front narrow, WF .57 x WH, 1.15 x HE ; front angle of ocellar EVANS : REVISION OF RHABDEPYRIS 73 triangle less than a right angle; OOL 1.33 x AVOT (Fig. 1). Vertex broadly rounded off a considerable distance above eye tops, distance from eye tops to vertex crest equal to slightly more than half HE. Front strongly alutaceous, somewhat shining, punctures shallow and inconspicuous, separated by 2-4 x their own diameters. First four antennal segments in a ratio of about 20:6:6:7, segment three (like the following segments except the last) very slightly longer than thick. Pronotal disc 1.6 x length of mesoscutum along the midline, its posterior margin paralleled by a series of large but shallow foveae ; pro- and mesonota alutaceous and shallowly punctate like the front ; notauli narrowly tear-shaped, diverging in front ; scutellar groove quite broad, deflected backward but not en- larged at each end. Propodeal disc 1.2 x as wide as its median length ; disc with numerous longitudinal carinae, of which three are complete, six others nearly so; surface aside from the carinae transversely striate ; surface of declivity and side-pieces somewhat beaded. Front femora 2.3 x as long as wide. Remarks. — This species is known only from the type. It is the only member of this complex with pale legs. 2. Rhabdepyris (Rhabdepyris) huachucae new species Holotype. — • $ , ARIZONA : Cochise Co., Huachuca Mts., Ramsey Canyon, 22 March 1956 (F. G. Werner & G. D. Butler) [MCZ, No. 30,936]. Description of female type. — Length 4.0 mm ; LFW 2.2 mm. Black; palpi and mandibles testaceous, the latter with the teeth rufous; scape pale castaneous, flagellum dull, light brown below but much darker on the upper side; tegulae testaceous; coxae and hind femora black; front femora and to a lesser extent the middle femora suffused with blackish toward the middle, the legs otherwise pale castaneous; wings hyaline, the veins and stigma light brown. Clypeus somewhat rounded except with a small median tooth formed by the tip of the strong, arched median ridge. Head higher than wide, WH .92 x LH; front narrow, WF .59 x WH, 1.06 x HE; front angle of ocellar triangle slightly less than a right angle, OOL 1.15 x WOT. Vertex rounded off a distance above eye tops equal to slightly more than half HE. Front strongly alutaceous, somewhat shining below but rather dull at the ocellar triangle; punctures very small, separated by 2-4 x their own diameters. 74 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY First four antennal segments in a ratio of about 26 :9 :10 :12, segment three 1.6 x as long as wide, all flagellar segments considerably longer than wide. Pronotal disc 1.6 x as long as mesoscutum along the mid- line, its posterior margin paralleled by a shallow, weakly fove- olate groove ; surface of pro- and mesonota uniformly alu- taceous, obscurely punctate; notauli slender, slightly attenuate and divergent anteriorly; scutellar groove forming an arc. Propodeal disc 1.18 x as wide as its median length; disc with five parallel carinae, also two additional weaker carinae beside the median carina, otherwise transversely striate ; declivity beaded, weakly striate below; side-pieces wholly beaded. Front femora 2.7 x as long as wide. Remarks. — This species is known only from the type. 3. Ehabdepyris (Rhabdepyris) muesebecki new species Holotype. — 9 , HONDURAS : intercepted at quarantine on banana debris from Honduras, at Galveston, Texas, 18 Febru- ary 1935 (C. P. Trotter; No. 887) [USNM, No. 67,536]. Description of female type. — Length 3.4 mm; LFW 1.8 mm. Black ; palpi light brown ; mandibles testaceous except infus- cated at extreme base ; scape black except apical .3 pale cas- taneous like the following segment, remainder of antenna dark castaneous on upper side, light yellowish brown below; tegulae light brown; coxae and femora dark brown, legs otherwise testaceous ; wings hyaline, veins and stigma very pale. Clypeus prominent, obtusely subangulate except with a small acute median angulation ; median carina high, arched. Head slightly wider than high, AVH 1.01 x LH; front rather narrow, WF .58 X WH, 1.12 X HE ; ocelli small, front angle of ocellar tri- angle less than a right angle, OOL 1.30 x WOT. Vertex broadly rounded off a distance above eye tops equal to about half HE. Front evenly alutaceous although moderately shining, punctures shallow but rather distinct, separated by 2-4 x their own diameters. First four antennal segments in a ratio of about 19 :6 :6 :7, segment three 1.1 x as long as thick, segment eleven not longer than thick. Pronotal disc 1.4 x as long as mesoscutum; posterior margin of disc paralleled hy a row of foveae ; pro- and mesonota less strongly alutaceous and much more shining than the front. Notauli tear-shaped, diverging toward the front ; scutellar groove EVANS : REVISION OP RIIABDEPYRIS 75 curved backward and slightly widened on each end. Propodeal disc 1.20 X as wide as its median length, its features exactly as described for the preceding species; declivity and side-pieces uniformly alutaceous, somewhat shining. Front femora 2.5 x as long as wide. Allotype. — $ , HONDURAS : La Ceiba, 21 March 1916 (F. J. Dyer) [USN^I]. Description of male allotype. — Length 2.0 mm; LFW 1.7 mm. Black; palpi and mandibles pale, as in female; antennae wholly dark brown, scape almost black; legs dark brown, except front tibiae and tarsi testaceous ; wings hyaline, veins and stigma brownish. Clypeus angulate, the median ridge subangulate in profile. Eyes prominent; head wider than high, WH 1.07 x LH; front fairly broad, the eyes convergent below, WF .61 x WH, 1.35 X HE ; front angle of ocellar triangle slightly less than a right angle; OOL 1.20 x WOT. Front strongly alutaceous, rather weakly shining, with shallow punctures distributed about as in female. Antennae elongate, first four segments in a ratio of about 13 :5 :9 :11, segment three 1.6 x as long as wide, seg- ment eleven 1.9 x as long as wide. Thoracic dorsum somewhat more shining and less alutaceous than the front; foveolate groove of pronotum, notauli, and scutellar groove all essentially as in female. Propodeal disc 1.2 x as wide as its median length, with five discal carinae and distinct lateral and sublateral carinae, otherwise transversely striate ; side- pieces alutaceous, somewhat shining. Paratypes. — MEXICO : 1 $ , with tomato, 30 August 1943, intercepted at quarantine at Brownsville, Texas, 54653, lot no. 43-11640 [USNIVI] ; 1 6 , "San Rafael Jicoltepec" ( ? = Jilo- tepec) [USNM] ; 1 $ , 3-6 mi. S Cuernavaca, Morelos, 4000 feet, 17 April 1959 (H. E. Evans) [MCZ] ; 1 9 , Chilpancingo, Guerrero, 4600 feet, June (H. H. Smith) [BMNH]. COSTA RICA: 1 9, San Jose, 1940 (H. Schmidt) [Sec. Agri., Sao Paulo, Brazil]. BOLIVIA: 1 5, Espia, Rio Bopi, July (W. M. Mann; Mulford Exped., 1921-22) [USNM]. Variation. — The paratypes show only insignificant variation in size (LFAV 1.6-2.0 mm). The Morelos male has the legs rather pale, all the tarsi being testaceous, the tibiae only partially inf uscated ; otherwise there is little variation in color. There is a certain amount of variation in head shape and width of the front, the Mexican female taken at Brownsville having a con- siderably broader front than any other; the Costa Rica and 76 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Bolivia females liave the narrow^est front and also have the lateral ocelli less far removed from the eye margins (Table I). The propodeum shows little variation in shape or sculpturing. It is conceivable, though I think not probable, that I am con- fusing more than one species under one name. 4. Rhabdepyris (Rhabdepyris) gracilis new species Holotype. — 9 , CALIFORNIA : Sacramento, 25 August 1932 (no collector given) [CAS]. Description of female type. — Length 3.2 mm; LFW 2.0 mm. Black, except last abdominal segment suffused with dark reddish brown ; palpi light brown ; mandibles testaceous on apical half ; scape dark brown, paler apically, flagellum castaneous, darker on upper side than below ; tegulae light brown ; coxae and femora dark brown, legs otherwise light brown ; wings hyaline, veins and stigma light brown. Clypeus obtusely angulate except acute at the midline ; median carina low except subangularly produced toward the base. Head higher than wide, WH .93 x LH; front broad, AVF .66 x WH, 1.44 x HE ; ocelli small, in a rather broad triangle, the front angle approximately a right angle; OOL 1.35 X WOT. Vertex produced well above eye tops, distance from eye tops to vertex crest equal to over two-thirds x HE. Front strongly alutaceous although moderately shining, the punctures small and shallow, separated by 2-3 x their oAvn diameters. First four antennal segments in a ratio of about 19 :7 :7 :8, seg- ment three about 1.3 x as long as thick, segment eleven barely longer than thick. Pronotal disc 1.35 x as long as mesoscutum, its posterior margin paralleled by a f oveolate groove ; surface of pro- and mesonota moderately shining, obscurely punctate ; notauli strong on posterior half of mesoscutum, tapering and diverging an- teriorly ; scutellar groove rather wide, deflected backward but barely widened on each side. Propodeal disc 1.10 x as wide as long, with seven discal carinae, otherwise transversely striate ; declivity and side-pieces beaded. Front femora 2.8 x as long as wide. Allotype. — $ , MEXICO : 10 mi. W Durango, Durango, 12 July 1954 (J. W. MacSwain) [CAS]. Description of male allotype. — Length 2.7 mm; LFW 1.9 mm. Black ; palpi brown ; mandibles testaceous on apical third, the teeth rufous ; scape black, flagellum dark brown ; legs dark EVANS : REVISION OF RHABDEPYBIS 77 brown; wings hj-aline. Clypeus broadly subangulate, with a small median tootli ; median ridge arched in profile. WH/LH = 1.0; front broad, WF .64 x WH, 1.42 x HE; front angle of ocellar triangle very slightly exceeding a right angle; OOL 1.33 X WOT ; vertex broadly rounded oif a considerable distance above the eye tops. Front alutaceous and weakly punctate about as in the female. First four antennal segments in a ratio of about 15 :5 :11 :12, segment three 1.8 x as long as wide, segment eleven about twice as long as wide. Thoracic dorsum weakly alutaceous, obscurely punctate; no- tauli very short, strong only on the posterior third of the mesoscutum; scutellar groove arcuate, slightly widened on each side. Propodeal disc 1.1 x as wide as long, with five longitudinal carinae, somewhat beaded beside the median carina but else- where transversely striate ; declivity and side-pieces beaded. Remarks. — This association of the sexes seems probable on the basis of available material, but only a great deal more collecting will solve this matter finally. I have seen only these two specimens assignable to this species. 5. Rhabdepyris (Riiabdepyris) minutulus new species Holotype.— 9, PERU (C.H.T. Townsend Coll.) (no further data) [USNM, No. 67,537]. Description of female type. — Length 1.9 mm; LFW 1.3 mm. Head and thorax black; abdomen dark castaneous; palpi and mandibles testaceous ; scape and following two antennal segments testaceous, remainder of antenna dark brown except somewhat paler on the under side; coxae and femora dark brown, middle and hind tibiae medium brown, legs otherwise testaceous; wings hyaline, veins and stigma light amber. Clypeus obtusely angu- late, with small median tooth. WH/LH = 1.0 ; front very narrow, WF .53 x WH, .90 x HE ; ocelli in a broad triangle, front angle about a right angle; OOL .80 x WOT. Vertex rounded off a short distance above the eye tops, distance from eye tops to vertex crest equal to less than one-third x HE. Front strongly shining below, more weakly shining and strongly alutaceous above; punctures indistinct. First four antennal segments in a ratio of about 14 :5 :2 A, segments three and eleven both wider than long. Pronotal disc 1.3 x as long as mesoscutum, its posterior margin paralleled by a rather weak series of small foveae ; surface 78 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY moderately shining, obscurely punctate. Mesoscutum with the notauli strong on the posterior half ; scutellar groove fairly wide, wider on the sides than medially. Propodeal disc 1.14 x as wide as long, with five longitudinal carinae and two addi- tional, weaker carinae beside the median carina, otherwise trans- versely striate ; declivity and side-pieces shining, rather weakly alutaceous. Front femora 2.4 x as long as wide, 6. Rhabdepyris (Rhabdepyris) nigriscapus new species Eolotype.— S, ARGENTINA: 5 mi. N Jujuy, 15 Febru- ary 1951 (Ross & Michelbacher) [CAS]. Description of female type. — Length 2.6 mm; LFW 1.7 mm. Black ; palpi and mandibles testaceous ; scape black except paler at tip, antennae otherwise dark brown above, light brown on under side ; coxae and femora dark brown, legs otherwise pale castaneous ; wings hyaline, veins and stigma light brown. Cly- peus broadly rounded apically, with a small median tooth. WH .98 X LH; front of moderate width, WF .64 x WH, 1.23 x HE ; ocelli in a broad triangle, OOL .87 x WOT (Fig. 2). Distance from eye tops to vertex crest equal to slightly less than half HE. Front rather strongly alutaceous, somewhat shining, with small, shallow punctures which are separated by 3-5 x their own diameters. First four antennal segments in a ratio of about 19:7:6:7, segments three and eleven both about 1.2 x as long as wide. Thoracic dorsum alutaceous, moderately shining, obscurely punctate; pronotal disc 1.2 x as long as mesoscutum, its posterior margin paralleled by a rather strong series of foveae. Notauli very short, barely longer than wide ; scutellar groove arching, not wider on the sides than medially. Propodeal disc 1.10 x as wide as long, with five longitudinal carinae between which are four other, weaker carinae, otherwise transversely striate; declivity weakly transversely striate ; side-pieces somewhat bead- ed. Front femora 2.9 x as long as wide. Remarks. — This species is similar to the preceding in many ways, but there are so many minor differences that the two are unlikely to be conspecific. Both species are known from the types only. EVANS : REVISION OP RHABDEPYRIS 79 B. Subgenus Trichotepyris Kieffer Trichotepyris Kieffer, 1906, m Andre, Spec. Hymen. Eur., 9: 376 (type species: B. paUidipennis Kieffer, 1906; designated by Muesebeck & Walkley, 1950) (proposed as subgenus of FJiahdepyris) . — Kieffer, 1914, Das Tierreich, 41: 346 (placed in synonymy of Bhabdepyris) . Suh generic characters. — Small to medium-sized bethylids ; black, head and thorax sometimes with metallic reflections, abdo- men sometimes brownish or in part rufous; ejes densely clothed with short hairs ; body with short, fine, mostly pale hair, without the strong dark setae of the preceding group (exception: nigropilosus) ; middle tibiae with or without spines. Mandibles large, in the male terminating in five sharp teeth, in the female with four or five teeth of variable development ; base of mandi- bles fairly close to bottoms of eyes, malar space less than or about equal to width of mandibles at their base (exception: nigropilosus) ; male with third antennal segment very small, shorter than second segment, closely consolidated with the much larger fourth segment. Pronotum moderately long, with or without a foveolate groove paralleling its posterior margin; notauli usually complete or nearly so ; scutellar groove strong ; propodeum of variable shape, with from three to seven longitu- dinal carinae, with or without transverse striations; declivity and side-pieces with fine striations in most species. Mesopleurum with an elongate upper fovea (sometimes divided) and a large lower fovea which is often incomplete above. Claws dentate, the tooth distinct, erect or sloping outward to some extent. Rem 23. blantoni 24. fulgens" BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY PLATE 5 Mesopleuia of female lihabdepyris Fig. 25. B. werneri n. sp., type Fig. 26. E. apaohe n. sp., type Fig. 27. E. texanus n. sp., type Fig. 28. R. carolinianus n. sp., type Fig. 29. E. luteipennis n. sp., type Fig. 30. R. nigerrimus n. sp., type Fig. 31. E. musoarius (Westwood), plesiallotype Fig. 32. E, origcnus Kieffer, type Fig. 33. E. virescens n. sp., type Fig. 34. E. subviridis (Kieffer), plesiallotype Fig. 35. E. tricolor n. sp., type Fig. 36. E. violaceus n. sp., type EVANS : REVISION OF RIIABDEPYRIS 25 werneri 26 opache 27. texanus 28. carolinianus 29. lufeipennis 30. nigerrimus 31. muscorius 32. origenus 33. virescens 34. subviridis 35. tricolor 36. violoceus BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY PLATE 6 Mesopleura of male Khabdepyris Tig. 37. E. muscarius (Westwood), specimen from Chapada, Brazil Fig. 38. B. vesculus n. sp., type Fig. 39. E. virescens n, sp., allotype Fig. 40. B. blantoni n. sp., type Fig. 41. B. viridissimus (Kieffer), type Fig. 42. B. fulgens (Brues), plesiallotype Mandibles of female BJiabdepyris Fig. 43. B. megacephalus (Aslimead), type Fig. 44. B. iverneri n. sp., type Fig. 45. B. fortunatus n. sp., type Fig. 46. B. oarolinianus n. sp., type Fig. 47. B. subaeneus Kieffer, type Fig. 48. B. angusticeps n. sp., type Fig. 49. B. luteipennis n. sp., type Fig. 50. B. muscarius (Westwood), plesiallotype Fig. 51. B. virescens n, sp., type Fig. 52. B. subviridis (Kieffer), plesiallotype Fig. 53. B. viridissimus (Kieffer), plesiallotype Fig. 54. B. fulgens (Brues), type EVANS : REVISION OF RIIABDEPYRIS 37 musconus 38 vesculus 39 virescens 40 blantoni 41 \/iridissimus 42 fulgens 43 megacephalus 44, wernen 45. fortunatus 46. carolinianus 47 subaeneus 48. angusticeps 49. luteipennis 50. muscorius 51. virescens 5Z. subviridls 53. viridissimus 54. fulgens BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY PLATE 7 Claws of posterior tarsi of Ehabdepyris Fig. 55. E. luteipennis n. sp., type female Fig. 56. E. luteipennis u. sp., allotype male Fig. 57. E. quinquelineatus Kieffer, type female Fig. 58. E. nigerrimus n. sp., type female Fig. 59. E. musearius (Westwoocl), plesiallotypc female Fig. 60. E. musearius (Westwood), type male Fig. 61. E. vesoulus n. sp., type male Fig. 62. E. virescens n. sp., type female Fig. 63. E. virescens n. sp., allotype male Fig. 6-1. E. hlantoni n. sp., type male Fig. 63. E. suiviridis (Kieffer), type male Fig. 66. E. viridissimus (Kieffer), plesiallotype female Fig. 67. E. viridissimus (Kieffer), tj-pe male Fig. 68. E. fulgens (Brues), type female Fig. 69. E. violaoeus n. sp., type female Fig. 70. E. violaoeus n. sp., allotype male Basal segments of antennae Fig. 71. E. luteipennis n. sp., allotype male Fig. 72. E. quinquelineatus Kieffer, plesiallotype male Fig. 73. E. subviridis (Kieffer), type male Fig. 74. E. fulgens (Brues), type female EVANS : REVISION OP RHABDEPYRIS cr:^ cz:^ C3^ c:::^^ 55. luteipennis ? 56, luteipennis ^ 57 quinquelineatus ? 58. nigerrimus * C^/ / 59. mjscarius ? 60. muscarius S 61. vesculus s 62. virescens s 63. virescens s 64. blantoni i 65. subviridis J 66. viridissimus ? ^^ C2^ CZ^ C2^ 67. viridissimus cJ" 68. fulgens ¥ 69. violaceus ? 70. violaceus i 71 luteipennis cT XDZI] 72. quinquelineatus d" 73. subviridis i 74. fulgens ? Bulletin of the Museum of Comparative Zoology HARVARD UNIVERSITY Vol. 133, Xo. 3 NELDASAUBU8 W RIGHT AE, A NEW RHACHITOMOUS LABYRINTHODONT FROM THE TEXAS LOWER PERMIAN By John Newland Chase Ohio Wesleyan University, Delaware, Ohio With Five Plates CAMBRIDGE, MASS., U.S.A, PRINTED FOR THE MUSEUM June 25, 1965 Publications Issued by or in Connection WITH THE MUSEUM OF COMPARATIVE ZOOLOGY HARVARD UNIVERSITY Bulletin (octavo) 1863 — The cui^ent volume is Vol. 133. Breviora (octavo) 1952 — No. 223 is current. Memoirs (quarto) 1864-1938 — Publication terminated with Vol. 55. Johnsonia (quarto) 1941 — A publication of the Department of Mollusks. Vol. 4, no. 43 is current. Occasional Papers of the Department of Mollusks (octavo) 1945 — Vol. 2, no. 31 is current. Proceedings of the New England Zoological Club (octavo) 1899-1948 — Published in connection with the Museum. Publication terminated with Vol. 24. The continuing publications are issued at irregular intervals in num- bers which may be purchased separately. Prices and lists may be obtained from the Publications Office of the Museum of Comparative Zoology, Cambridge 38, Massachusetts. Peters' Check-List of Birds of the World. Volume 1 is out of print; volumes 2 to 7, 9, 10 and 15 are currently available. Phylogeny and Evolution of Crustacea. Proceedings of a confer- ence held at Cambridge, Mass., March 6-8, 1962. H. B. Whittington and W. D. I. Rolfe, editors. Cambridge, Mass., 192 pp., 80 figs., 1963. Cloth bound, $6.75. Fishes of the Gulf of Maine, by Henry B. Bigelow and William C. Schroeder. Washington, viii - 577 pp., 1953. Photo-offset reprint, $6.50. Publications of the Boston Society of Natural History The remaining stock of the scientific periodicals of the Boston Society of Natural History has been transferred to the Museum of Comparative Zoology for distribution. Prices for individual numbers may be had upon request. Bulletin of the Museum of Comparative Zoology HAKVAED UNIVEESITY Vol. 133, No. 3 NELDA8AUBUS W BIGHT AE, A NEW RHACHITOMOUS LABYRINTHODONT FROM THE TEXAS LOWER PERMIAN By John Newland Chase Ohio Wesleyan University, Delaware, Ohio With Five Plates CAMBRIDGE, MASS., U.S.A. FEINTED FOR THE MUSEUM June, 1965 Bull. Mus. Comp. Zool., Harvard Univ., 133(3) : 153-225, June, 1965 No. 3. Neldasaurus wrightae, A New Rhachitomous Labyrinthodont from the Texas Lower Permian^ By John Newland Chase Department of Zoology, Ohio Wesleyan University, Delaware, Ohio TABLE OF CONTENTS Introduction 156 Acknowledgements 157 Systematic Description 158 Generic and Specific Diagnoses 158 Descriptive List of Materials 159 Anatomical Description of Genus and Species 160 General Skull Morphology 160 Bones of the Dermal Skull Roof 163 The Sensory Canal System 168 Palate and Palatoquadrate 169 Parasphenoid and Braincase 176 Occiput 179 Lower Jaw 182 Vertebral Column 188 Ribs 193 Appendicular Skeleton 196 Scales 203 Other Occurrences of Neldasaurus 205 Discussion 206 Comparison of Neldasaurus with other Genera of Trimerorhaehoids 206 Interrelationships of Trimerorhachoid Genera 215 Trimerorhachoid Origins and Relationships 220 Literature Cited 222 1 This paper is part of a thesis submittefl to the Department of Biology of Harvard University as partial fulfillment of the requirements for the degree of Doctor of Philosophy, August, 1962. 156 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY INTRODUCTION Since their discovery in 1875, the Lower Permian redbeds of north central Texas have been an important source of fossil vertebrates. Until recent decades, however, the fauna of this region (mainly described by Cope, Case, and Williston) was de- rived from a very limited series of horizons, which include the upper portions of the Wichita Group and the lowest part of the overlying Clear Fork. Collections made in the uppermost beds of the Clear Fork by Olson (1948, 1951 a-c, 1955) and in the Lower "Wichita formations by Romer and associates (Romer, 1935, 1947, 1958) have extended our knowledge and given us some concept of the faunal assemblages that preceded and fol- lowed those of the typical Lower Permian faunas. This paper is concerned with certain amphibian remains recovered from the lower levels of the Wichita. A number of expeditions from the Museum of Comparative Zoology at Harvard College have been made to the Texas red- beds from 1934 onward. In most cases these had the special ob- jective of obtaining remains from the lower horizons of the Wichita Group — the Putnam, Moran and Pueblo formations in descending order. The finds have included numerous identifiable amphibian specimens (mainly from the Putnam and Moran) which, for the most part, pertain to genera already well known from the higher, "classic," Wichita formations, the Admiral and the Belle Plains. These include such forms as T rimer orhachis, Eryops, Broiliellus, Tersomius, Parioxys and Acheloma. How- ever, the lower formations showed indication of a somewhat dif- ferent and more primitive faunal assemblage, particularly in the occurrence of the primitive rhachitome, Edops (Romer and Witter, 1942). A further indication of the faunal difference between the upper, typical, Wichita and the lower Wichita formations was the recovery in the earlier trips to these lower beds of remains of an obviously new rhachitome too fragmentary to permit ade- quate description. Vertebral and limb materials showed some similarity to T rimer orhachis, a form not identified with cer- tainty below the Putnam (Olson, 1955), but such skull remains as were recovered indicated a narrow, attenuated snout quite different from that of Trimerorhachis. A collecting trip from the Museum of Comparative Zoology in 1954 under the direction of Dr. Alfred S. Romer resulted in the CHASE : NELDASAURUS WRIGHTAE 157 recovery of a nearly complete skull of this problematical form, with attached lower jaws and some postcranial material in a common block. The certain association of the skull with post- cranial materials made possible for the first time a comprehen- sive description of this new form and the assignment of the specimens in previous collections to a new genus. The fossil material described in this paper is assigned to a new genus and species of rhachitomous labyrinthodont for which I propose the name Neldasaurus wrightae. Both generic and speci- fic designations are in honor of Miss Nelda E. Wright, Research Assistant and Editor of Publications of the Museum of Com- parative Zoology, who discovered the holotype specimen. ACKNOWLEDGEMENTS My appreciation is due to several persons whose aid I re- received during the course of this study. I should like first to express my sincere thanks to Dr. Alfred S. Romer, under whose inspiring direction this work was undertaken and whose con- tinued interest and advice have been of invaluable assistance. I am indebted to Dr. Donald Baird of Princeton for many helpful suggestions and permission to use his information on the Linton fauna ; to Dr. Nicholas Hotton III of the United States National Museum for information on Acroplous vorax; to Dr. Edwin H. Colbert of the American Museum of Natural History for access to specimens of Trinierorhachis ; to Dr. Craig C. Black of the Carnegie Museum for the loan of specimens ; to Dr. Robert Carroll then at the Museum of Comparative Zoology for informa- tion pertaining to early edopsoids ; to Dr. Carl Cans of the Uni- versity of Buffalo and Mr. Frank White of the Biological Labo- ratories at Harvard for assistance in taking photographs of the specimens. Thanks are also due to Mr. Arnold Lewis of the Museum of Comparative Zoology at Harvard, who gave freely of his time and knowledge in introducing me to the techniques of fossil preparation. Thanks are due also to Dr. Ernest E. Williams and Professor Bryan Patterson of the Museum of Comparative Zoology for critical reading of the manuscript. I am indebted to Miss Nelda Wright for many valuable sugges- tions during the preparation of this paper. A year of this study was made possible by a Danforth Teacher Study Grant, for which I am very grateful. Preparation of the manuscript was aided by a grant from the Shell Assist Fund. 158 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY SYSTEMATIC DESCRIPTION^ Class AMPHIBIA Order TEMNOSPONDYLI Suborder EHACHITOMI Superfamily TRIMERORHACHOIDEA Family TRIMERORHACHIDAE Neldasaukus, gen. n. Type species. Neldasaurus wrightae sp.n. Generic Diagnosis. — A low-skulled labyrinthodont amphibian which closely resembles T rimer orhachis in size, many skull pro- portions, body and limbs, dermal roof pattern, and palate. Neldasaurus differs from T rimer orhachis in the following re- spects : orbits midway between snout and occiput ; snout nar- row ; lacrimal elongate but not reaching the external naris ; nasals, prefrontals and frontals elongate ; external nares close together, their long axes more or less parallel ; jugal entering the orbital border; prevomers elongate and the choanae broadly separated from the anterior border of the interpterygoid vacui- ties ; larger number of teeth than in T rimer orhachis in the lower jaw and the marginal series of the upper jaw; a tusk pair on the ectopterygoid ; a foramen in the lower jaw for the re- ception of vomerine tusks ; opisthotic and prootic unossified ; pleurocentra as high as the intercentra; clavicles meeting in front of the interclavicle. Neldasaurus wrightae sp. n. Holotype. — MCZ 2200, including a nearly complete skull and attached lower jaws, central and neural arch vertebral elements, a partial pectoral girdle, limb bones and remnants of dermal armor. Type Locality and Horizon. — Terrapin School, Section A- 1266, BBB + CRR Survey, Archer County, Texas ; Moran For- mation, "Wichita Group, Permian. Collected by A. S. Romer party, 1954. Diagnosis. — As for genus. Referred Specimens. — UCZ Nos. 1371, 1381, 1438, 1463, 2404, 2406, 2407, 2516, 2518, 2519. 1 The classification used here follows Homer (1947). CHASE : NELDASAURUS WRIGHTAE 159 DESCRIPTIVE LIST OF MATERIALS The material consists of the holotype and the following speci- mens, all from the Wichita Group of the Texas Lower Permian. MCZ 1371. A partial skull and left lower jaw, dorsal verte- brae — some in articulated series, ribs, a partial clavicle and iuterclavicle, a nearly complete forelimb, dermal scales and armor. There is also what appears to be a pelycosaur foot bone. Putnam Formation, Table Branch, South Fork of the Little Wichita River, G. AV. Stell Survey, Section A-382, 10 miles west of Anarene in Archer County, Collected by R. V. Witter party, 1936. MCZ 1381. Fragmentary remains of several small individuals. Same data as for MCZ 1371. MCZ 1438. Poorly preserved remains of skulls and postcra- nial materials of at least two individuals of little use for pur- poses of description. Putnam Formation, 1 mile west of Archer City in the Archer City bone bed. Section 151, American Tribune New Colony Subdivision, southwest part Archer County. Col- lected by R. V. Witter party, 1936. MCZ 1463. Fragmentary cranial and postcranial materials in a limey nodule. Section 16, Falls County School Land, Archer County, about 7 miles southwest of Anarene. Collected by R. V. Witter party, 1936. The formation is either Putnam or Moran. MCZ 2404. Fragmentary portions of lower jaw and postcra- nial materials. Moran formation at Terrapin School, BBB -f CRR Survey, A-1266 in southern Archer County. Collected by A. S. Romer party, 1950. MCZ 2406. A large number of discrete skull and postcranial materials, mostly fragmentary, including the proximal ends of a humerus and a femur. There is also a Trimerorliachis-like intercentrum, a complete ring central element and scraps of a larger labyrinthodont. Moran Formation near Padgett in the northeast part of William Tryndale Survey, Young County. Collected by A. S. Romer, 1942. MCZ 2407. Fragmentary vertebral materials, the distal ends of a radius and a femur and portions of a pterygoid. Collection site as for MCZ 2406. Collected by R. V. Witter, 1942. MCZ 2516. Separate and disarticulated elements, mostly in- complete ; including parts of a lower jaw, a right femur, an ilium, and vertebrae. Moran Formation, %^ mile north and 1 mile west of Padgett Schoolhouse, Young County. Collected by R. V. Witter party, 1936. 160 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY MCZ 2518. Fragmentary skull and jaw materials, numerous partial vertebral elements, ribs, pectoral plates and two ilia, probably all pertaining to one individual. Moran Formation in Padgett, the northwest part of the William Tryndale Survey, MCZ Field No. 9, Young County. Collected by A. S. Eomer, 1951. MCZ 2519. A partial intercentrum. Moran Formation west of Cottonwood Creek, about Section 48, County School Land, Archer County. Collected by A. S. Romer, 1954. ANATOMICAL DESCRIPTION OF GENUS AND SPECIES General Skull Morphology Of the three principal skull specimens, the holotype, MCZ 2200 (Pis. 1 and 2), is the most complete. The figures and de- scription are primarily based on this specimen, but supplemen- tary information was obtained from MCZ 1371 (PI. 3), MCZ 1438 (PL 4), and also the fragments included in MCZ 2518. The holotype skull is described first. The dermal roof, although fairly complete, is displaced to the left in relation to the palate and has been subjected to post- mortem crushing so that the skull appears to have retained its original depth only in the right suborbital region. Horizontal pressures have caused some overlapping of the medial boundaries of the central dermal bones. Intrusion of the articular portions of the lower jaws from be- low partially obscures the relationships of the quadrate region. The anterior extremity of the snout is incomplete in all of the specimens, but the portions that remain permit a reasonable res- toration. The right lower jaw of the holotype has retained a fairly normal position throughout much of its length so that most of the lateral skull margin is visible. However, the posterior quarter of the ventrolateral border of the skull roof has been damaged and details of structure are hard to trace. Post-mortem flattening of the skull and damage to the skull margins introduce some uncertainty in the determination of cranial dimensions. However, the figures obtained from the re- constructed skull roof, the palate and the lower jaw, give the general dimensions, as set forth in Table 1, with a fair degree of accuracy. Determination of the depth of the skull was based primarily on the transverse dimensions of the palate and the resultant restoration of the skull roof to fit these dimensions. CHASE : NELDASAURUS WBIGHTAE 161 The skull (Figs. 1, 2, 3 and 5) is of moderate size. The snout is narrow but bluntly rounded anteriorly, the skull increasing in breadth toward the back, having its greatest width in the oc- cipital region. Allowing for a slight backward projection of the condyle from the occipital border, the suspensorium is about 1.5 cm behind the level of the condyle. However, the otic notch is not as highly developed as in Trimerorhachis, and the contour of the central posterior rim of the skull table is less concave than it is in that animal. The occiput slopes downward and backward at an angle of approximately 23° from the vertical in the preserved skull. But, though some degree of slope may be normal, the occiput may have been more nearly vertical in life. The depth of the skull midway between the orbits and the end of the snout is approximately 13 mm, and the height of the cheek below the orbit is 14 mm. The height of the occiput is about 39 mm. Thus, the depth of the skull increases from front to back, the increase being considerably more pronounced in the postorbital region, the snout being generally flattened. Determination of the original skull roof topography is difficult because of post-mortem distortion. The evidence seems to indi- cate a more or less flat or slightly concave surface in the central region of the skull table bounded by a low ridge on each side which runs from the tabular to the outer posterior corner of the orbit. The outer face of the ridge slopes gently downward, be- coming confluent with the cheek, and the inner face dips down to the central region of the skull table. The mid-orbit to snout length is approximately 49.8 per cent of the midline skull length. The dorsally directed orbits are oval in shape and somewhat smaller in relation to overall skull size than those of Trimerorhachis'^ , though, as in Trimerorhachis, the interorbital space is narrow. The inner border of the orbit is slightly elevated above the surface of the skull roof, the smooth, medial margin having a depth of 3 mm. The rest of the margin, which shows no appreciable elevation, is not as thick. A small parietal foramen is present in the midline between the parietals, 2.9 cm behind the posterior orbital border. The dis- tance from the orbit to the parietal foramen is equal to 38 per cent of the skull table length, a shorter distance than that seen in Trimerorhachis, where the comparable figure is 50 per cent. 1 This comparison is based on the ratio of skull width X skull length/orbit width X orbit length. 162 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY In agreement with the dimensions of the rostrum, the external nares are not as widely separated as they are in other tri- merorhachoids. The medial margin of the naris is slightly ele- vated above the surface of the internarial region, and the opening is somewhat constricted in its posterior third. The lateral border of the naris is separated from the skull margin by a dorsal ex- posure of the premaxilla and the maxilla of approximately 5.5 mm. Skull MCZ 1371 (PI. 3) exhibits overall dimensions that are closely comparable to those of the holotype, and the dimensions given in Table 1 can be accepted with a fair degree of certainty. Extreme flattening makes the determination of skull height dif- ficult, but it appears to agree closely with that of MCZ 2200. The preserved portions of skull MCZ 1438 (PI. 4) appear to represent a somewhat larger animal than the preceding forms. Overall length cannot be determined, but the dimensions of the preorbital region suggest proportions similar to those of MCZ 2200 and 1371. The dermal roof bones of MCZ 2518 are disarticulated and incomplete so that comparison of general dimensions of this specimen and the others is impossible. TABLE I Measurements (in mm) of Dermal Skull Eoof MCZ MCZ MCZ No. No. No. Length 2200 1371 1438 Snout to end of skull table 158 — — Snout to quadrate 173 174 — Snout to mid-orbit 78 80 86 Orbit 20 20 (est.) — External naris 13 — 14 Width Snout behind naris 40 42 46 Orbital region 70 74 — Quadrate level 120 — — Orbit 16 15 (est.) — External naris 8 — 8 Internarial 10 — 12 K eight Mid-rostral 13 — 18 (est.) Suborbital 14 14 (est.) 15 (est.) Occiput 39 — — chase : neldasaurus wrightae 163 Bones of The Dermal Skull Roof (Figures 1 and 2) Some sutures on one or another of the skull specimens, MCZ 2200, 1371, 1438, 2418, were readily traced, but the location of others, which were not recognizable per se, was based on the sculpture patterns of individual bones. Centers of ossifica- tion were determined by considering their relationship to the sensory canal system, which is typically associated with the cen- ter of growth in each bone (Bystrow, 1935; Westoll, 1943; Par- rington, 1949). I believe that the general pattern of the skull roof of MCZ 2200, as figured, is reasonably accurate. The dermal roof pattern closely resembles that of Trimerorha- chis. Differences in the shape of individual bones in the two animals are obviously due in large measure to the differences in skull proportions mentioned above. The dermal bones of the skull roof are treated here as com- prising four groups: (1) the paired elements of the dorsal mid- line, (2) the circumorbital bones, (3) the bones of the temporal region that form the lateral margin of the skull table, and (4) the marginal tooth-bearing bones and the cheek bones. (1) The nasals are long, roughly rectangular bones which ex- tend more than half the distance from the naris to the orbit, their shape sharply contrasted to the squarish nasals in T rimer orhachis. The elongate frontals, with over half their length extending into the preorbital region are also in contrast to conditions in T rimer orhachis, in which form the nasals make a much greater contribution to the midfacial area than do the frontals. The parietals form most of the central portion of the skull table, extending from the posterior border of the frontals to the postparietal suture. A parietal opening of modest size is located centrally between them. The roughly pentagonal postparietals are relatively small bones. Posteriorly, they turn down over the rim of the skull table in an occipital exposure. A stout occipital flange of the postparietal, which rests on a dorsal extension of the exoccipital, bounds the supraoccipital region laterally and forms the medial rim of the posttemporal fenestra (Fig. 5). The tabular does not appear to take part in the formation of the descending flange. (2) The circumorbital series, including the prefrontal, post- frontal, postorbital, jugal, and lacrimal, is complete. 164 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY PMX Fig. 1. Dorsal view of skull roof of Neldasauriis wrightae, n. gen., n. sp., restored, X -75. Portions of the sensory canal grooves and some sutures on the right have been restored to agree with conditions on the left. Abbreviations: EO, exoccipital; F, frontal; IT, intertemporal; J, jugal; L, lacrimal; MX, maxilla; N, nasal; P, parietal; PF, postfrontal; PMX, premaxilla; PO, postorbital; PP, postparietal; PEF, prefrontal; QJ, quad- ratojugal; SQ, squamosal; ST, supratemporal; T, tabular. CHASE : NELDASAURUS WRIGHT AE 165 The prefrontal is a narrow bone which forms the anteromedial margin of the orbit, ending halfway back on the medial rim. The postfrontal is narrow anteriorly where it joins the pre- frontal to complete the medial rim of the orbit. Behind the orbit it becomes somewhat wider. The jugal is a large bone approximately 7 cm long, and the longest skull roof bone, aside from the maxilla. Unfortunately, the suborbital region in all of the specimens is poorly preserved or damaged, but a break in the surface contour of the right cheek below the orbit in MCZ 2200 and the pattern of the sculp- ture in this area suggest that there was a lacrimal-jugal suture about halfway back on the lateral rim of the orbit. If this inter- pretation is correct, the jugal forms at least a small part of the orbital border. The postorbital forms most of the posterior orbital rim, meet- ing the jugal laterally on the outer margin of the orbit, though the exact position of the suture is not known. The lacrimal is elongate. It forms much of the anterolateral rim of the orbit and, though it fails to reach the naris, it extends four-fifths of the distance from the orbit to the naris in the holotype skull. (3) Primitively, the lateral margin of the skull table includes an intertemporal bone as well as supratemporal and tabular. The posterior limits of the intertemporal, a relatively small bone here, were difficult to trace but the anterior and medial bounda- ries were well defined on MCZ 2200 and MCZ 2518. The tabular is the smallest bone in the skull table. Laterally it has a short contact with the squamosal, excluding the supra- temporal from the margin of the otic notch, a feature which tends to accompany modest development of the latter (Homer, 1947, p. 24). Medially, where it meets the postparietal, the tabular forms most of the dorsal margin of the posttemporal fossa. Midway in its posterior border the tabular has a ventral projection, near the inner edge of the otic notch, which is the posterior end of a low ridge that continues anteriorly under the skull roof for a distance of 7 mm. The depth of this ridge, or flange, is approximately 3 mm.i Immediately anterior to its posterior border the ventral sur- face of the tabular is excavated so that a depression is formed. 1 Case (193.5) mentions a "low riisoso line" rniinine: forward and Inward on the under surface of the tabular from the inner edge of the otic notch in his Trimero- rhachis specimen 16002, which may be comparable to the structure described here. 166 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY bounded medially by the "flange" described above and posteriorly by the posterior rim of the tabular. Lat- erally this fossa, which is about 9 mm wide, is bounded by a ventrally- directed process from the posterome- dial edge of the squamosal. Little of the underside of the skull roof of this region can be seen and its precise character, partially visible only in the holotype, is difficult to interpret be- cause of its imperfect condition. (4) The boundaries of the marginal tooth-bearing bones, the premaxilla and maxilla, were hard to find in the rostral region. Their outlines as fi- nally determined are a composite of the patterns that could be seen on parts of four specimens, MCZ 2200, 1371, 1438, and 2518. The premax- illa, seen best in MCZ 1438, appar- ently had a small dorsal exposure an- teriorly, separating the anterior rim of the external naris from the rostral border by a distance of approximately 3 mm. The union of the premaxilla with the nasal posteriorly appears to have been close to the anterior narial rim on the medial side. Medially, the premaxilla joins its fellow of the op- posite side, and posteriorly, lateral to the naris, it joins the maxilla near the middle of the lateral rim of the naris. Fig. 2. Skull of Neldasaurus wrightae, in lateral aspect. Eestored, X -75. Abbreviation: Q, quadrate. Other abbreviations as in Figure 1. CHASE: NELDASAURUS WRIGIITAE 367 The maxilla extends along the lateral margin of the skull for about two-thirds of the skull length, its dorsal exposure dwin- dling rapidly towards the rear. Posteriorly, it comes close to the quadratojugal. Conditions here are uncertain, but the jugal appears to enter the skull margin for a short distance between the maxilla and quadratojugal. The postorbital cheek region is formed in typical fashion by the squamosal and quadratojugal. Posteriorly the squamosal has a contact with the anterior portion of the quadrate and forms part of the outer margin of the otic notch. On the inner posterior margin of the cheek it has a ventral and anteriorly directed proc- ess, about 6 mm deep, which joins the quadrate ramus of the pterygoid on its lateral edge. The quadratojugal forms the posterolateral margin of the skull below the squamosal. Conditions in the left quadrate region of MCZ 2200, as near as can be determined, are like those in MCZ 1371, where it is better preserved. The posterior edge of the quadratojugal, forming a shallow, concave rim, is separated from the lateral margin by a sharp angle. The inner posterior end of the bone forms a medial projection posterior to the sutural surface for the squamosal. The posterior rim of this medial process probably formed part of the border of the quadrate foramen. The sutural face on the quadratojugal for articulation with the squamosal, therefore, does not reach the anterior edge of the foramen, and the squamosal does not reach its border. The remainder of the border of the quadrate foramen was apparently formed by the quadrate. Conditions here ap- pear to be comparable to those seen in T rimer orhachis (Case, 1935, fig. 4; Williston, 1915, fig. 3). The pattern of the dermal roof bones of MCZ 1371 (PL 3) ap- pears to be identical to that of the holotype. This specimen offers further evidence of the relations of the lacrimal, which is here prevented from reaching the posterior narial border by a union of the maxilla and the nasal. Also, although the situation is not entirely clear, the lacrimal appears to have a stout union with the jugal below the orbit so that the jugal forms more than half of the lateral orbital border. The skull roof of MCZ 1438 (PI. 4) is incomplete, but it ap- pears to follow the general pattern seen in MCZ 2200 and 1371. In contrast to its form in the other skulls, the lacrimal here reaches to, or nearly to, the external naris. The only other apparent difference is in the slightly larger dimensions of the 168 BULLETIN : MUSEUM OF COMPARATn^E ZOOLOGY individual bones, which agree with the general porportions of the skull. As stated earlier, the skull roof of MCZ 2518 consists of a number of isolated fragments. In so far as it can be traced, the dermal roof pattern is like that of the holotype. The sculpture of the dermal roof bones consists of ridges and pits, or grooves, radiating from a center of ossification in each bone. Notable differences between the sculpture pattern in Neldasaurus and T rimer orhachis are obviously correlated with differences in the shape of individual bones in the two animals. As any figure shows (Bystrow, 1938, fig. 11), there is no linear sculpture in Trimer orhachis, whereas in Neldasaurus there is a suggestion of linear sculpture on adjacent regions of the frontals and nasals and on the frontals and parietals. The sculpture also tends to become linear in portions of other long bones in Nelda- saurus such as the lacrimal and jugal. The Sensory Canal System The channels of a sensory canal system are well defined on the surface of the dermal roof bones in Neldasaurus (Figs. 1 and 2). In all the specimens the pattern of these channels is re- markably constant, and, except for the anterior rostral portions of the supraorbital and infraorbital channels, the pattern is clear. The pattern of the sensory canal system generally resembles that of Trimer orhachis (Case, 1935, fig. 5 A) but differs in some particulars. In Neldasaurus the supraorbital groove crosses the postfrontal, prefrontal, and nasal but does not enter the lacrimal as in Trimer orhachis. The anterior portion of the infraorbital groove in Neldasaurus runs from the jugal along the suture be- tween the maxilla and lacrimal, then continues for some distance on the lacrimal. The anterior portion of the infraorbital groove in Trimerorhachis is confined to the maxilla except for a short loop onto the lacrimal. The position of the temporal groove in Neldasaurus is esssen- tially as in Trimerorhachis, though it is slightly more medial in Neldasaurus. A jugal sulcus of the sensory canal system, separated by a short gap from the infraorbital groove on the jugal, crosses the jugal-squamosal suture near its midpoint and has the shape of a medially convex curve on the central region of the squamosal, swinging laterally near the posterior margin CHASE : NELDASAUEUS WRIGHTAE 169 to end at the posterolateral corner of the bone. It does not ap- pear to have passed across the quadratojugal, but over the pos- terior border of the squamosal, just behind the end of the qua- dratojugal, to its connection with the lower jaw system. This groove is unknown in T rimer orhachis. The courses traced by the sensory canal grooves of MCZ 1371, 1438 and 2518, where they can be seen, are the same as those on the holotype. Palate and Palatoquadrate (Figure 3 and Plate 2) In all of the skull specimens the lower jaws lay in more or less normal closed position against the ventral surface of the skull, thus obscuring the lateral palatal surface. Fortunately, it was possible to expose this region by removing the right lower jaw ramus and 2.5 cm of the anterior end of the left jaw ramus from the ventral surface of the holotype skull. The description and reconstruction of the palate and the ventral skull surface, as shown in Figure 3, were mainly based on this specimen. Restored portions are hatched in the figure. The general form and proportions of the central and posterior portions of the palate in Neldasaurus, including the presence of large interpterygoid vacuities, resemble conditions in Trimero- rhachis. However, the region of the palate in front of the inter- pterygoid vacuities in Neldasaurus, reflecting the proportions of the elongate snout, is more extensive. The anterior end of the palate is incomplete, but the evidence suggests that the premaxilla had a moderate palatal exposure. The ventral exposure of the maxilla, throughout most of its length, represents little more than the width of the tooth row. The maxilla accounts for nearly two-thirds of the lateral mar- gin ; its posterior end briefly enters the anterior portion of the lateral rim of the subtemporal fossa. The jugal apparently enters for a short distance into the ventral margin of the skull posterior to the maxilla, but most of the lateral border of the fossa is formed by the quadratojugal. Determination of the extent of the palatal exposure of the maxilla adjacent to the internal naris was difficult, and some un- certainty exists. Since the palatine appears to form most if not all of the lateral border of the choana, the maxilla could enter 170 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Fig. 3. Reconstruction of skull of Neldasaurus wrightae in palatal as- pect, X -75. Eestored areas are hatched. Abbreviations: APF, anterior palatal fenestra; BO, basioccipital; CH, choana; EC, ectopterygoid; EO, exoccipital; J, jugal; MX, maxilla; PAL, palatine; PMX, premaxilla; PS, parasphenoid ; PT, pterygoid; Q, quadrate; QJ, quadratojugal ; SE, sphen- ethmoid ; VO, vomer. CHASE : NELDASAURUS WRIGHTAE 171 its border (if at all) only at the outer anterior corner. Exclu- sion of the maxilla from the lateral rim of the choana by a union of the vomer and the palatine, which appears to be the case here, is to my knowledge reported in only one other form, Eohrachyops townendi (Watson, 1956). The suture between the maxilla and premaxilla can not be seen in ventral aspect. Never- theless, on the basis of its location on the lateral skull margin, it probably lies close to the level of the anterior member of the vomerine tusk pair. The marginal teeth are small, crowded and numerous. By counting the teeth in areas where they can be seen, it was pos- sible to estimate a total number of 93 maxillary teeth. The an- terior teeth are less than 1 mm in diameter, and in the anterior portion of the maxilla there are slightly less than 10 teeth per cm, while towards the posterior end, where the teeth are still smaller, there are approximately 11 per cm. The length of the tooth-bearing margin of the premaxilla, as reconstructed, is approximately 2.5 cm. The narrow palatal exposure of this bone suggests small tooth size in this region also and, allowing for some decrease in the number of teeth per unit area, the pre- maxilla should have held at least 15 teeth. In some regions of the maxilla it was noted that teeth alternated with replacement sockets, though no consistent pattern could be seen. Although the anterior portion of the palate in MCZ 2200 is damaged, there is evidence of the presence of two sets of open- ings, the anterior palatine fenestrae and the choanae. Immedi- ately anterior, and slightly medial, to the anterior vomerine tusk on each side there is an opening in the palate. In the matrix deposited in this region, on both sides, there are remnants of the upper ends of symphysial tusks belonging to the lower jaw. It appears that anterior palatal fenestrae were present on the suture between the premaxillae and the vomers in Neldasaurus. The matrix-filled space containing the upper end of the tusk is confluent with a gap in the anterior part of the floor of the ex- ternal naris in MCZ 2200. The condition of this region in MCZ 1371 and 1438 is not clear but appears to be similar in the former at least. A comparable situation exists in Eupelor {Buettneria) . The choanae are located halfway between the anterior end of the interpterygoid vacuities and the end of the snout. They are suboval in outline, with a length of approximately 1.5 cm and a greatest width of .6 cm. Anteriorly, they are constricted by a lateral curvature of the medial border. The lateral margin of 172 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY the choana is formed in the main, and perhaps entirely, b}^ the palatine, the anterior and medial margins by the vomer. Al- though the palate was crushed against the ventral surface of the skull roof in this region of the type specimen, it is apparent that the vomer formed a stout rim on the medial and anterior borders of the opening. The anterior ends of the choanae are 1 cm behind the level of the posterior rims of the external nares. Therefore, to reach the internal naris from the outside, air had to pass backward through a short passageway between the skull roof and the palate — a passageway surely enclosed in a cartilaginous nasal capsule. The central palatal surface anterior to the interpterygoid vacuities is formed by the vomers. These bones, conforming to the contours of the snout, are approximately four times as long as they are wide. Anteriorly, the vomer is bounded by the pre- maxilla, and forms the posterior rim of the anterior palatal fenestra ; laterally, it joins the maxilla and the palatine, and forms the medial and posterior boundaries of the choana. The vomer meets the palatine midway on the posterior rim of the choana. The suture between the two bones appears to run diago- nally backward and outward, medial to the palatine tooth-row. Posteriorly, the vomer forms the anterior margin of the inter- pterygoid vacuity, but the two vomers are separated in the mid- line, posteriorly, by the wedge-shaped anterior end of the cultri- form process of the parasphenoid. At the point where the vomers meet medially the end of the cultriform process appears to pass dorsal to them. The posterolateral sutural relations of the vomers are obscured by numerous cracks and the poor preservation of the palatal surface. However, it appears that the vomer formed the antero- lateral border of the interpterygoid vacuity, retaining a primi- tive contact with the pterygoid medially near the anterior end of that opening. The palatine is bounded laterally by the maxilla and medially by the vomer. Posteriorly, the palatine meets the ectopterygoid. This suture was hard to find but appears to run diagonally for- ward across the palatal toothrow immediately in front of an en- larged tooth pair and 2.5 cm behind a pair of palatine tusks. If this is the correct position of the suture, the ectopterygoid bears a pair of tusks at its anterior end. It is of interest to note that in T rimer orhachis, as described by Case (1915), the ectopterygoid is a short bone, and the palatine supports not only an anterior CHASE : NELDASAURUS WRIGHTAE 173 fang pair but a second pair of enlarged teeth as well. No en- larged tusks occur on the ectopterygoid in that form. A review of the literature failed to uncover any different description of the palatine-ectopterygoid suture in Trimerorhachis. Thus, if the present interpretation is correct, Neldasaurus is more primitive than TrimerorhacJiis in this respect. Assuming that the position of the suture is correct, the ectopterygoid is a long bone, about equal in length to the pala- tine. ]\Iedially, it is bounded by the palatal ramus of the ptery- goid and laterally by the maxilla. Posteriorly, where it enters the anterior border of the subtemporal fossa, it is very narrow. The palatine and ectopterygoid are thickened under the pala- tal tooth row, and join the maxilla in a firm articulation on the medial surface of that bone. The palatine is stouter than the ectopterygoid. The union of these lateral palatal bones with the maxilla does not reach the ventral edge of the latter with the result that, in ventral aspect, there is a shallow longitudinal groove along the lateral edge of the palate just inside the skull margin. In primitive fashion, the vomer, the palatine and, apparently, the ectopterygoid each bears a pair of tusks. Those of the palatine are the largest and those of the ectopterygoid are the smallest. It was not possible to determine whether or not any small teeth were present on the vomers between the choanae. However, both the palatine and ectopterygoid tusks are accompanied by a single row of small teeth. On the left side of the palate in the type specimen the vo- merine tusk pair is represented by an empty socket with a tusk behind it. On the right the conditions are reversed. The vo- merine tusk is .5 cm in diameter near the base and is some- what over 1 cm long. The right palatine bone also supports a tusk pair at its an- terior end, the posterior tusk represented by an empty socket. The palatine tusk preserved is larger than that of the vomer, with a diameter at the base of .6 cm and an approximate length of 1.2 cm. Behind the tusk pair on the palatine is a close-set row of 8 small teeth. The more anterior ones are oval in out- line, their long axes perpendicular to the tooth row. The pos- terior ones are round in section. Larger than the maxillary teeth, the palatine teeth average 2 mm in diameter. The third pair of tusks is located at the anterior end of the ectopterygoid. The anterior member of the pair has a basal diameter of 3 mm, 174 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY and the posterior a diameter of 4 mm. Behind these, extending nearly to the end of the ectopterygoid, is a row of 21 teeth, de- creasing slightly in size towards the posterior end of the row. Their diameter at the base is about 1 mm. The pterygoid of the type specimen, though damaged, is nearly complete. There are also portions of pterygoid bones in MCZ 2406 and 2407. The pterygoid consists of a horizontal palatal ramus, a stout central body and a vertical quadrate ramus. The curved, me- dial margin of the palatal ramus forms the lateral and postero- lateral borders of the interpterygoid vacuity. The bone in gen- eral is fairly stout but is thin at its medial edge. The palatal ramus has a narrow anterior process, bounded laterally by the ectopterygoid, which articulates with the vomer anteriorly. The posterior portion of the palatal ramus is expanded horizontally. The outer edge of the pterygoid is here thickened and turned ventrally to form a flange about 1 mm deep which forms the anterior portion of the medial border of the subtemporal fossa; the remainder of the border is supplied by the central body and the quadrate ramus of the pterygoid. The basal articulation was apparently movable (Fig. 4, A). The thickened, horizontal body of the pterygoid opposite the ..j^yOy- B Fig. 4. Neldasaurus wrightae. A, reconstruction of the basal articular region of the skull of MCZ 2200 in ventral view, X -75. (Evidence for the presence of the teeth shown on the right pterygoid based on MCZ 2406 and 2407). B, projection of a portion of the medial surface of the left palato- quadrate of MCZ 2200, X -75. Abbreviations: BO, basioccipital ; CE, conical recess; EO, exoecipital; ET, "excavatio tympanica"; PS, parasphenoid ; PT, pterygoid; RC, pocket for rectus capitus muscle ? CHASE : NELDASAURUS WRIGHTAE 175 basal articular region of the basisphenoid has an internal process that presumably articulated with the basipterygoid process of the braincase. The internal process, seen best on the right in MCZ 2200, curves dorsally from the level of the central body of the pterygoid so that its tip is nearly .5 cm above the lower surface of the bone. An ascending process rises from the medial edge of the pterygoid just behind the tip of the internal process. Its height above the internal process is about .5 cm, but it rises rapidly posterior to this point to a height of 1.7 — 1.8 cm opposite the basipterygoid process of the braincase. The concave pos- terior face of the internal process of the pterygoid forms the anterior and ventral walls of a socket or "conical recess" for the reception of the basipterygoid process of the braincase. The back wall of the "conical recess" is formed by an essen- tially vertical ridge that rises from the dorsal surface of the pterygoid, just inside the medial edge of the bone, about 1 cm behind the tip of the internal process. This leaves a very limited socket for the reception of the basipterygoid process. Behind the ridge there is a vertical groove, about 5 mm wide, which separates the ridge from the anterior end of the vertical quadrate ramus (Fig. 4, B). The latter, rising above a small medial projection of the pterygoid, forms a second ridge posterior to the groove. The posterior face of the ridge thus produced at the root of the quadrate ramus, and the medial face of the quadrate ramus ad- jacent to it, share a concavity just above the ventral surface of the pterygoid. It is obvious that this deeply excavated area rep- resents the "excavatio tympanica" of Bystrow and Efremov (1940). It appears that Neldasaurus is unusual in that the pos- terior wall of the "conical recess" is separated by a groove from the ascending ridge at the base of the quadrate ramus which forms the anterior wall of the tympanic excavation. The quadrate ramus of the pterygoid is approximately 4 cm long and curves outward slightly as it proceeds from its root to an attachment with the quadrate at the outer posterior corner of the skull. Just behind the root, the quadrate ramus, as noted, becomes more or less vertical in the skull, apparently ascending dorsally and somewhat medially. Near the level of, or just an- terior to, the otic notch the vertical flange of the quadrate ramus overlaps on the medial side a vertically descending flange of the squamosal. The posteromedial face of the quadrate ramus ap- pears to have had an irregular surface, marked by longitudinally oriented ridges and grooves. There is a prominent, more or 176 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY less horizontal, elongate groove in the quadrate ramus, just be- low its connection with the descending flange of the squamosal. Although post-mortem flattening of this region has introduced some uncertainty in dorso ventral relationships, it seems highly probable that this groove is homologous with the groove in the posteromedial face of the quadrate ramus of T rimer orhachis, which apparently serves as a floor for the tympanic cavity (Watson, 1956). Little can be seen of the quadrate bone. On the medial side of the left cheek in the holotype, however, an anterior extension of the quadrate, which diminishes in size anteriorly, meets the squamosal dorsally and the quadrate ramus of the pterygoid ventromedially. There is no evidence of an ossified epipterygoid in Nelda- saurus. A review of the Trimerorhachis materials in the collec- tion of the Museum of Comparative Zoology confirmed earlier reports that there is no ossified epipterj-goid in that form either. The obstinate character of the matrix left no indication of palatal teeth in areas other than those described in MCZ 2200, but the fragments of the pterygoid bones in MCZ 2406 and 2407 show that the horizontal flange and part of the palatal ramus are covered by a densely-packed shagreen of small teeth except at the medial edge of the bone. Parasphenoid and Braincase The braincase is fairly complete in ventral aspect in the holo- type specimen. The dorsal surface is crushed against the under- side of the skull roof and little can be seen of the lateral surface. The only other braincase material consists of a badly weathered partial parasphenoid with MCZ 1438 and a small portion of the left side of an isolated parasphenoid in MCZ 2518. The de- scription of the parasphenoid and the braincase is based on the holotype. The proportions of the braincase are remarkably similar to those of TrimerorJiachis (Case, 1935; Watson, 1956). The sphe- nethmoid region is narrow, but behind its articulation with the pterygoids the braincase is expanded and more or less flattened. As is usual, the ventral surface of the braincase is sheathed by the parasphenoid. The parasphenoid narrows gradually in front of the basi- pterj^goid process of the braincase, giving rise to an anteriorly CHASE : NELDASAURUS WRIGHTAE 177 directed cultriform process, which forms the medial border of each interpterygoid vacuity. The cultriform process is about 6.6 cm long, fairly broad at each end, but narrow in the central region. The anterior end of the process is wider than in Tri- merorhachis, but it does not project as far forward into the anterior palatal surface as it does in such forms as Sanrerpeton or the Triassic metoposaurs. The anterior and posterior portions of the ventral surface of the cultriform process are nearly flat; the central portion is ventrally convex in cross section, the lateral edge being approxi- mately 2 mm above the level of the ventral surface. The surface of the cultriform process is marked by fine longitudinal ridges and grooves, which also appear to some extent on the main body of the parasphenoid. There is no evidence of teeth on the ventral surface of the parasphenoid, though the rigorous preparation required to remove the matrix could have obscured them. The main body of the parasphenoid is broad and essentially flat in ventral view (Fig. 4, A). Behind the root of the cultri- form process the parasphenoid expands laterally, the central region between the laterally expanded ''wings" being moder- ately concave. Laterally, the parasphenoid sheaths the anterior, ventral and posterolateral faces of the basipterygoid process. The "core" of the process, ossified as part of the basisphenoid in many labyrinthodonts, was here apparently cartilaginous. It would have been exposed at the tip of the process above the parasphenoid, where it could have entered the "conical recess" of the pterygoid to meet the cartilaginous epipterygoid. The anterior margin of the parasphenoid covering the basipterygoid process is directed diagonally outward and backward in direct apposition to the posterior margin of the internal process of the pterygoid. As in Trimcrorhachis, but in no other Permian temnospondyl, the articular process of the parasphenoid is set off from the main body of the bone bj^ a groove which runs di- agonally outw^ard and backward, widening and increasing in depth distally. The anteroventral rim of the articular process of the parasphenoid issues from the side of the braincase on the same level as the ventral surface of the main body of the bone. Behind the basipterygoid processes, the parasphenoid expands laterally to a width of approximately 5 cm. The margin of the bone then curves sharply inward, matching posteriorly the width of the basioccipital which is 2.2 cm across the base. The pos- terior border of the braincase has been damaged, but it appears 178 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY that the parasphenoid reaches, or nearly reaches, the posterior ventral rim of the basioccipital, although it is extremely thin at this point. A lateral extension of the parasphenoid turns upward on either side around the basioccipital, gaining a narrow exposure on the lateral face of the braincase at the level of the condyle. Anteriorly this lateral exposure increases, attaining a height of nearly 5 mm ventral to the fenestra ovalis, the ventral rim of which was apparently formed by the parasphenoid. The dorsal and lateral margins are incomplete. In T rimer orliachis the lateral concave face of the parasphenoid behind the basiptery- goid process contains a deep pocket ventral to the fenestra ovalis, which Watson suggests could have received a rectus capitus muscle. Although this region in Nelclasaurus is not com- plete, the details that can be seen suggest that a similar, though less pronounced, fossa for muscle insertion was present. Careful scrutiny failed to disclose the presence of any open- ings for the palatine or internal carotid arteries in the ventral or lateral surfaces of the parasphenoid. However, these openings are often small and are not consistently demonstrable even in a well known form like Trimerorhachis. Failure to find them in Neldasaurus can probably be attributed to their small size or to the rigorous treatment necessarily employed in removing the matrix from the surface of the bone. Some features of the dorsal surface of the parasphenoid can be determined from exposed portions of the lateral edges, and from the face of a break which runs diagonally across the main body of the braincase. The outer end of a broad groove in the dorsal surface of the parasphenoid can be seen on the lateral surface of the braincase above the laterally expanded "wing." The groove is 6 mm wide at its lateral end and is directed an- teromedially. It appears to be comparable to a similar groove on the upper surface of the parasphenoid in Trimerorhachis and, as Watson (1956) suggests, probably sheathed the lower medial end of the prootic. Immediately anterior to the basipterygoid process, the lateral edge of the parasphenoid is thin, but both faces of a break in this region exhibit a thickened mass of bone. Although the form of this thickened mass is partially obscured by matrix, it appears to be part of a strongly developed transverse ridge of bone on the upper surface of the parasphenoid between the roots of the basipterygoid processes. The ridge could occupy the position to CHASE : NELDASAURUS WRIGHTAE 179 be expected of the posterior rim of a depression for the pituitary. Reduction of ossification in the braincase is well advanced. The basisphenoid and otic capsule were apparently cartilaginous, since neither is preserved in the holotype. There is, however, evidence of an ossified sphenethmoid. A flat bone about 1 cm high appears to occupy the space between the cultriform process and the skull roof from the anterior end of the interpterygoid vacuity back to a point not far in front of the level of the parie- tal foramen, a distance of about 4 cm. This bony sheet, bounding laterally the space above the cultriform process, undoubtedly represents an ossified sphenethmoid of the right side. There is no evidence that there was any lateral expansion of the bone like that seen in Edops and Eryops. Unfortunately, the stapes is not preserved in any specimen. Occiput (Figure 5) The occipital region of the holotype, though damaged, is nearly complete and can be restored. An isolated partial condyle among the materials associated with MCZ 1371 also offers some information, but the description is based mainly on the holotype. The occipital region of MCZ 2200 has been subjected to post- mortem flattening, resulting in some dislocation of its parts. The dorsal portions of the exoccipitals are missing, and only the left side of the condyle is reasonably intact. The incomplete nature of the exoccipitals hindered an attempt to get an accurate picture of the proportions of the occiput. Fig. 5. Neldasaurus wrightae (MCZ 2200 and MCZ 1371). Eeconstruc- tion of the skull in occipital aspect, X -75 approx. Abbreviations : BO, basi- oeeipital; FM, foramen magnum; POP, paraoccipital process; PS, para- sphenoid; PT, pterygoid; PTF, posttemporal fossa. Other abbreviations as in Figure 1. Eestored features are in dotted lines. 180 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY However, on the basis of the general similarity of the structure and proportions of the occipital region of Neldasaurus and T rimer orhachis, the height of the dorsal portions of the exocci- pitals of Neldasaurus was estimated by comparison with Wat- son's figures for T rimer oj'hachis. Accordingly, the height of the occiput of Neldasaurus was approximately 3.9 cm. As noted earlier, the backward slope of the occiput was probably not pronounced in life. The occipital condyle is a single, subcircular structure with a concave posterior face. Its lower division is formed by the basi- oecipital, best seen in MCZ 1371. The basioccipital is wedge- shaped in lateral view and extends forward ventrally into the floor of the brainease for a distance of approximately 1.5 cm. The dorsal surface of this anterior extension is unfinished. As said earlier, the parasphenoid covers the ventral surface of the basioccipital to, or nearly to, its posterior border. As in Tri- merorhachis, the ventral surface of the basioccipital has longi- tudinal ridges which were in contact with the upper surface of the parasphenoid. The exoccipitals, resting directly on the basioccipital, form the dorsolateral portions of the condjde and complete its concave posterior face. The position of the suture between the basiocci- pital and exoccipital can not be determined with certainty but appears to be about half-way between the base and the top of the condyle. Dorsally, the condylar portions of the exoccipitals are well separated, but they approach each other in the midline ventrall}". Above the center of the condyle, which contains a notochordal pit, the exoccipitals are only separated by a narrow groove. Above the condyle the exoccipital had a dorsal process, represented in the specimens by its basal portion only, which formed the lateral wall of the foramen magnum and articulated dorsally with the inferior surface of the descending flange of the postparietal. Anteriorly, a projection of the basal portion of the exoccipital of each side extends forward in the floor of the brainease above the anterior extension of the basioccipital, but details are obscure. Probably, as in T rimer orhachis, it provided a floor for the medullary region of the brainease. These an- terior extensions of the exoccipitals do not quite meet in the midline. The structure of this region, in so far as it can be determined, corresponds generally to that described for Tri- mer orhachis by Watson. CHASE : NELDASAURUS WRIGIITAE 181 Anterior and somewhat lateral to the base of the dorsal exten- sion of the exoccipital, there is a small process of the exoccipital which forms the anterior wall of the vagal foramen. The pos- terior wall of the foramen is formed by the base of the ascending process of the bone. Just posterior to the vagal foramen the lateral wall of the exoccipital is pierced by a singe foramen for nerve XII. The inner opening of this foramen is in the medial surface of the bone, anterior and slightly ventral to its lateral exit, as seen in MCZ 1371. The ventral surface of the condyle is broad and more or less flat. The width of the condyle in MCZ 2200 is 2.2. em and its height is 1.7 cm, while the isolated condyle of MCZ 1371, with a width of 2.3 cm and a height of 1.3 cm, is even lower. These proportions seem to reflect a general trend towards flattening in Neldasaurus. The lateral sides of the condyle rise abruptly from the more or less flattened base. In lateral aspect they are considerably narrower than the sides of the condyle in Trimero- rhachis. Taken together, these characteristics of the condyle resemble the proportions of the intercentra of Neldasaurus, in which slender, upright lateral processes arise from the edges of a broad ventral base. The occipital exposure of the postparietal has a slight back- ward slope, which is interpreted as being normal. There is no supraoccipital ossification. The ventral articular surface of the postparietal flange slants laterally and ventrally at an angle of 45 degrees to the horizontal, ending laterally in a blunt point. The medial portion of this flange presumably articulated with the upper end of the exoccipital; the laterally projecting outer corner may have rested on the base of a cartilaginous paroccipi- tal bar. There is some evidence that a cartilaginous paroccipital bar extended between the exoccipital and the tabular. The upper end of the bar could have been received in the fossa in the ven- tral surface of the tabular already described. The lower end of the bar presumably met the anterior surface of the exoccipital ventromedially. There is no indication that the tabular or the exoccipital invaded the paroccipital bar, the latter thus forming the whole of the lower boundary of the posttemporal fossa. The remainder of the fossa, as in T rimer orkacliis, is formed by the postparietal and the tabular. 182 bulletin : museum of comparative zoology Lower Jaw (Figure 6) All of the lower jaw specimens have suffered some post-mortem damage and distortion, especially in the articular region. The most complete jaws, MCZ 2200, 1371 and the anterior por- tions of 1438, are crushed against the ventral surface of the skull so that much of their dorsal and medial surfaces is obscured. Fragments of lower jaws are also present in MCZ 2516, 1381, 1463, 2406, 2404, and 2518. The right jaw of the holotype specimen is well preserved throughout its prearticular extent, the articular region being hidden. The left jaw is nearly complete from front to back in ventral aspect, the articular region, though twisted, retaining its connection with the prearticular ramus. The infradentary portions of both jaws have been flattened against the palate, los- ing their original relation to the vertical plane of the dentary. The left lower jaw of MCZ 1371 (Fig. 6, A) has undergone some distortion and damage, but the preservation of the surface permits a more accurate tracing of the suture pattern than either jaw in MCZ 2200. The anterior tip is missing, but the contours of the preserved portion permit an accurate restoration. Also, though it is incomplete, the articular region is more readily vis- ible in this specimen. The reconstruction of the lower jaw shown in Figure 6, B is a composite, based primarily on MCZ 2200 and 1371. The rela- tion of the articular region to the ventral rim of the jaw is based on this portion of a small jaw in MCZ 1381, but this relationship is, admittedly, subject to some uncertainty in the larger specimens where the articular region has shared in the general flattening of the ramus. Aside from a few details, the lower jaw of Neldasaurus closely resembles the lower jaw of TrimerorJiachis. Viewed from below, the anterior end of the jaw is bluntly rounded, the end of each ramus swinging abruptly towards the symphysis from its lateral border, thus paralleling the blunt contours of the snout. The jaw is very shallow anteriorly, its strongly curved outer surface directed as much ventrally as it is laterally. Posteriorly, it be- comes progressively deeper, and the convexity of the lateral surface, though continued, is less pronounced. The ventral rim of the jaw was apparently medial to the plane of the dorsal rim CHASE : NELDASAURUS WRIGHTAE 183 in life. The medial surface, forming a sharp angle with the lat- eral surface ventrally, is essentially vertical. Posteriorly, the ventral margin of the jaw curves sharply upward to the articu- lar region. There is a modest retroarticular process which ex- tends for a distance of 6 mm behind the posterior rim of the glenoid fossa on the medial side of the jaw. Two cm behind the anterior end of the jaw the dorsal edge of the dentary rises to a point 3 mm above its anterior level; 3 cm behind this point a second, though less abrupt, increase in height occurs. From here, the dorsal margin of the jaw rises gradually to the articu- lar region. CHT SANG AN 6 SANG S MAND SACC Fig. 6. Lower jaw of Neldasaurus wrightae X -50. A, ventral view of left ramus of MCZ 1371. B, external view of left ramus restored (based mostly on MCZ 2200 and MCZ 1371). Abbreviations: AMF, anterior Meckelian fenestra; ANG, angular; CHT, foramen for chorda tympani; D, dentary; PA, preartieular ; PMF, posterior Meckelian fenestra; SACC, accessory sulcus; SANG, surangular; SD, dentary sulcus; SMAND, mandibular sul- cus; SP, splenial; SPP, posterior splenial; VT, fenestra for reception of vomerine tusk. The bones of the outer surface of the jaw are sculptured. The medial surface and the ventral rim are smooth. Two narrow, elongate Meckelian fossae are present just above the ventral margin on the inner surface of the jaw. There is a foramen for the chorda tympani just under the edge of the glenoid fossa at the root of the retroarticular process on the medial surface of the jaw. 184 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY In the discussion of the palatal dentition, a fenestra for the reception of a vomerine tusk in the lower jaw was mentioned. In both lower jaws of MCZ 2200 and 1438 and the single jaw of 1371 there is a hole that penetrates the jaw from top to bot- tom to accommodate the tip of a vomerine tusk. That this is not the result of post-mortem flattening of the skull is shown by the absence of breakage of the surface bones and the finished rim of the opening. The opening of the fenestra in the ventral surface of the jaw is suboval in outline, the long dimension be- ing directed anteromedially. It is approximately 4 mm wide and 5 mm long, and is located on the suture between the dentary and the splenial at, or slightly behind, the level of the posterior limit of the symphysis. In MCZ 2200 a second fenestra is present in the anterior end of the postsplenial for the accommodation of a palatine tusk. This second fenestra, however, is not evident in MCZ 1371 or 1438. Colosteus scutellatus and Ei'petosauriis rodiafus (Romer, 1930) have a groove in the outer surface of the lower jaw for the reception of a large premaxillary fang. Some Crocodylidae have developed fossae which penetrate the upper surface of the anterior end of the snout for the lower jaw fangs and, of course, anterior palatal openings are present in a large number of tem- nospondylous labyrinthodonts. But perforation of the lower jaw for the accommodation of an upper jaw tusk appears to be unique in Neldasaurus. The absence of the posterior fenestra in two of the skulls may reflect individual variation in the depth of the jaw in relation to the length of the tusks, or perhaps in older animals large tusks may gradually wear a smooth opening through the lower jaw. The symphysis is formed by the dentary anteriorly, and the splenial (presplenial) forms the posterior half. Ventrally, it is approximately 1.5 cm long; dorsally it is 2.2 cm long. A medial expansion of the splenial near its anterior end meets a similar expansion from the opposite side, thus lengthening the extent of the symphysis dorsally. The posterior border of the medial process so formed lies 1 cm above the ventral rim of the jaw. A somewhat similar condition appears to obtain in Trimero- rhachis, though developed to a lesser degree. The dentary occupies most of the upper portion of the lateral surface of the jaw ramus and, as already shown, forms most of CHASE : NELDASAURUS WRIGHTAE 185 the anterior end of the jaw. Except for some uncertainty con- cerning the posterior extent of the dentary, its relation to the other dermal elements is similar to that of other rhachitomes. It is bounded ventrally by the splenial, postsplenial and angu- lar from front to back, and overlaps each of these elements to a variable extent. The dentary narrows abruptly at the back, its ventral edge ascending diagonally upward and backward. Al- though its relation to the surangular is not clear, an anterior process of the surangular appears to be interjected between the upturned posterior end of the dentary and the dorsal border of the angular. Much of the dentary is sculptured, but the dorsal portion along the tooth row is quite smooth except for the pres- ence of fine, longitudinal striations. The splenial, as stated, forms the posterior portion of the sym- physis, its anterior end meeting the dentary in a serrate suture. On the outer surface of the jaw it has a long diagonal external suture with the postsplenial, which slants downward and back- ward, and internally reaches the anterior Meckelian fenestra, forming the anterior rim of that opening. The postsplenial extends posteriorly to the posterior Meckelian fenestra, forming the anterior third of the border of that open- ing. Posteriorly, it has a ventrally directed, diagonal suture with the angular on the lateral surface of the jaw. The post- splenial turns over the ventral rim of the jaw, appearing at the level of the posterior Meckelian fenestra to have a height well over 1 cm on the medial surface. The large angular forms the posteroventral rim and most of the lateral surface of the jaw below and anterior to the articu- lation. Posteriorly and dorsally the angular is bounded by the surangular, anteriorly and dorsally by the dentary. It has a rather long anterior extension, confined to the lateral face of the ramus, which projects between the postsplenial and the den- tary in a gradually narrowed process. Ventrally, a suture be- tween the angular and the prearticular is located on, or just medial to, the ventral rim of the jaw. The surangular is not complete in any of the specimens. However, on the medial surface of the articular region of the jaw of MCZ 1371 a rather large crack slants upward and back- ward from the end of the angular to the lip of the glenoid fossa. In the absence of other evidence, and because of its position, this "crack" is tentatively identified as the suture between the surangular and the prearticular. Thus, the surangular would 186 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY form the posterior border of the glenoid fossa. Immediately posterior to the "suture," and, as noted previously, just under the edge of the glenoid fossa, is a foramen for the chorda tym- pani. Its dorsal position and the fact that it is nearly, if not completely, surrounded by the surangular contrasts to the sit- uation in T rimer orhachis. In that form, the surangular, angular and prearticular all contribute to the borders of the foramen, according to Case's figures, and its position is more ventral. The foramen for the chorda tympani in MCZ 2200 has a lower posi- tion than that observed in MCZ 1371, suggesting that its loca- tion is subject to individual variation. The limits of the articular bone can not be determined and a dorsal view of the glenoid fossa is seen only in MCZ 1371. Un- fortunately, it has been damaged. The fossa consists of a con- cave depression passing diagonally inward from the outer pos- terior corner of the jaw. A slight ridge separates an outer ar- ticular facet from a longer facet on the inner side of the fossa. The full extent of the prearticular can not be determined, nor can its relation with the coronoids be seen in surface view. In typical fashion it forms the medial wall of the adductor fossa above the angular and postsplenial. Its anterior end extends for some distance in front of the posterior Meckelian fenestra. The medial edge of the adductor fossa is, as typically, lower than the lateral edge. The fossa appears to end anteriorly at the level of the posterior rim of the posterior Meckelian fenestra, having a length of approximately 4 cm. Unfortunately, the coronoid series is not visible in any of the specimens. The only source of information is that supplied by cross sections of the right lower jaw of MCZ 2200 (Fig. 7). A section through the anterior coronoid, Section J^, shows three small teeth in a transverse series. Presumably these teeth rep- resent the condition continued throughout the rest of the bone. Traces of similar small teeth were also found in sections showing the other coronoid elements in Neldasaurus. Section Ji was made approximately 3.5 cm behind the tip of the jaw and 2.3 cm in front of the anterior end of the anterior Meckelian foramen. It should, therefore, pass through the den- tary, the splenial and the anterior coronoid. As can be seen in the figure, some distortion is evident, but the three elements can be identified. A comparison of this section with a section taken through the anterior end of the jaw of T rimer orhachis (Broom, 1913, fig. CHASE : NELDASAUBUS WRIGHTAE 187 9D) demonstrates a basic similarity in structure, though the jaw of Neldasaurus is relatively more narrow. Section J2 was made approximately 1 cm in front of the pos- terior Meckelian foramen. This section shows very clearly a medial extension of the dentary below the tooth row and above the Meckelian space. The lower portion of the jaw has been folded medially to the vertical plane of the dentary. If this were drawn back to a reasonably normal position, the height of the jaw at this level would be approximately 1.8 cm and the ventral rim would lie a considerable distance medial to the lat- eral surface of the upper border of the dentary. PAL? ANG SPP J2 Fig. 7. Transverse sections of the right lower jaw and lateral palatal border of MCZ 2200, X 1- Ji, Section 3.5 cm behind the anterior end of the ramus; J2, Section 1 cm in front of the posterior Meckelian foramen. Abbreviations: Ci, anterior coronoid; lAC, inferior alveolar canal; MSP, Meckelian space; PD, palatal debris; V?, vomer. Other abbreviations as in previous figures. The sensory canal system of the lower jaw appears to be rep- resented by three lateral line grooves — a mandibular, a den- tary, and an accessory. The mandibular sulcus originates on the posterior margin of the jaw, lateral to the retroartieular process. Its course is readily traced in MCZ 1371 and 2200, where it pro- ceeds along the ventrolateral edge of the jaw, just above the ventral rim, to the posterior border of the symphysis. The dentary sensory groove arises from the mandibular groove below the articular region, and runs forward along the upper margin of the angular. Anterior to the midpoint of the angu- lar it can not be traced with certainty. There is some evidence in MCZ 1463 and in the small jaw of 1381 of the presence of an accessory sulcus, dorsal to the dentary sulcus, at the posterior end of the jaw. Due to poor preservation it can not be positively identified in either 1371 or 2200. 188 BULLETIN : MUSEUM OE COMPARATIVE ZOOLOGY By piecing together the information that can be obtained from MCZ 2200 and 1371, it is possible to give a fairly complete account of the lower jaw dentition. Neldasaurus possesses a pair of prominent symphysial tusks on either ramus, each with a diameter at the base of about 5 mm and a height of at least 1 cm. The teeth of the dentary series in MCZ 2200 and 1371 agree generally in size and number in comparable sections of the jaws of the two specimens, and it appears that the lower jaw of Nel- dasaurus contained 60 or so teeth. The teeth are more crowded than they are in Trimerorhachis (where there are about 48 teeth in the lower jaw), a feature which reflects the high tooth count of the maxillary series already described. An attempt to recognize a consistent pattern of tooth replace- ment was unsuccessful. There is a gradual reduction in tooth size from front to back, but with some suggestion of regional enlargement not far be- hind the anterior end of the dentary in MCZ 2200. The teeth in MCZ 1371, though all have lost their tips, are better preserved than in either jaw of MCZ 2200. Again, there is regional en- largement of the teeth somewhat behind the anterior end of the tooth row. Near the anterior end of the dentary the teeth have a diameter at the base of 1.75 mm. The enlarged teeth that fol- low have a diameter of 2.5-3.0 mm. In the lower jaw^ tooth series here described, teeth 18-22 are enlarged, 21 and 22 being the largest. Comparison of the position of this series with the palatal dentition shows that these teeth would be located between the vomerine and palatine tusks, though closer to the latter. Be- hind this series a second, though less pronounced, enlargement occurs in teeth 33-38. Vertebral Column (Figures 8 and 9) In addition to numerous isolated elements, mostly centra and intercentra, three specimens show associated vertebral material in more or less continuous series. The vertebral column of MCZ 2200 is distorted and much of the detail is obscure. However, it is possible to trace a reason- ably continuous presacral column in which there were approxi- mately 34 vertebrae. MCZ 2518 includes a number of blocks con- taining presacral vertebrae and several isolated elements, in- cluding 33 intercentra. Twenty-five vertebrae can be accounted CHASE : NELDASAURUS WRIGHTAE 189 for in MCZ 1371. It seems probable, therefore, considering the possibility of loss of some elements, that Neldasaurus had ap- proximately 34 presacral vertebrae. The estimated number for Trimcrorhachis is 31 (Williston, 1915; Case, 1935). Unfortunately, no atlas vertebra was found and there are only three elements possibly belonging to the caudal series. On the basis of the relative positions of the vertebral elements in blocks, variations in size of the elements, and variations in the height of the rib facets on centra and intercentra, the pre- sacral vertebral columns of MCZ 2200, 1371 and 2518 were re- constructed. ^ The similarity of the vertebrae of all the Nelda- saurus specimens makes a detailed description of each specimen unnecessary. B '• w O '^ -id _ 'H ^ ^ \ ^ £ g 's^ CS bfio or, - jo; 9 ^ H. 3 o ,i2 Q>o t: - -^^.'-S-t ii: o'^ 5 ,>^ S i ^- -r = ? = ^ ^ '~ 5 — r^ r, — x" " .;i o ■'- s ■^' - r^ .rt r^ " - -^ ~ .^ '^ ~:° --"ri^c;-!- o-*-''t:s;— -il^ ft:®":; o ^^ be =H ^ _ H .2 3 5 & --2 O" 6 -£ X H s h; ^ Cm .2 :£ ^ U 0) 5, =ss§g •sss^BS^pis^o.'Ba: — ^ 'i.^ ^^ X ca CS 1 >-, -M [S o .2 1 'rt ■^ *i_ Cw Sc o ■+-' 4-- o o 13 Is S cr. < = ® - i •> o j- i =tH J -r - >-. .- "*"* ^ -^H ^ •^ x c (7^ "^^ cr o n2 '^ if • •. o cu ■^ ." >-. i ^ ^ ^• ^ =^ 53 QT. -> -*^ o X w ^, t^ S X ^ -1 "^ t: -^ X — ^ '"-n :^ ^ F- r ~ < i^ ^~* -*-* X H c^ c ft o ^^ ^ s* t^ ■^ ^ o r^ =4-1 y; 1^ r^ "T o O . — fet i/ c >-l o fc ."^ 1j o ^ > "3 o o o ~ ex OJ ■Jl ^ O o "ft J_^ =t-i ' o >-^ •r-t o , =t-l -— -*-- " o c ft X X "-4-i o — o ■+1 5 - o be "« — s o o .~ o S ^' ^ r: O > -*— ' >■ M p. ■^ ^ >> f-^ S »^ ^ o r* a> o ■4-' t£ OC 'i; CJ • ^ t'. ^ o •^ ^ Si o "^ =;-l -J—' -i^ >". '^ r 5 O X — o - !=! ft ® ^ ^ ^^ +^ 'o K ■*- be •-; >> X ft ^1 X '^ ;i X p - o 2 - . - M .2 -— r" - ._ O-S t-H-^flH^^^bi ^ ^ .2 I i ^ ^ OJ a O; 0; o se O cS o Gj ^ j; be ~ O X ^ cS >■ X =4-1 X o o xf O 3 ft o o '5 o — '" ^ S bJ3 o ■r 'Z ^ j 3 ItJ^ ^^x = g H ° 3-3-. = -^ =^^ I * "" 11 I i^ s ^ " .- 3 ^- i S J = 1 1 :s'-^^ -^r tt +- O -^^^ z:_2 Jl , o- — ' c:/ ■-'^ x> i;-T o jj ■- Z '■''- Z ^'^ i:-^ -?fe2 2.^ --St B -S o ^ _ _= r-- ? p +- r, = _ •- - . o X V i' C o _■ o o = u rt a; C r— ^ '^ 'i^ rs o +i ^ :« y. B -l-i rt K j- X o ^i^ =4-1 o i^ !>•; A ^ O ft H-i pC +- 220 bulletix : museum of comparative zoology Trimerorhachoid Origins and Relationships The origin of the trimerorhaehoids is problematical. Some characters of this group are advanced and of a sort typically met with in Triassic labyrinthodonts. On the other hand, a number of primitive features in the trimerorhaehoids, notably single condyle, movable basal articulation, and retention of inter- temporal, are not found in labyrinthodonts above the edopsoid level and hence suggest an origin of the trimerorhaehoids from an early member of the edopsoid group. Among the earliest edopsoids are such Pennsylvanian forms as Gaudrya from Nyfany, the dendrerpetontids from Joggins, and the English edopsoid-like labyrinthodont, Eugyrinns. A trend in dermal roof pattern in edo]isoids, characterized by withdrawal of the lacrimal from the anterior orbital border, is already well established in (laudrya and the dendrerpetontids. which would bar them from being directly ancestral to the trimerorhaehoids. Eugyrinus shows a number of characters that might be ex- pected in a trimerorhachoid ancestor, such as: (a) short skull. (b) broad snout, (c) short face, (d) small otic notch, (e) skull table moderately expanded, (f) apparently single condyle, (g) parietal foramen close to orbits, (h) primitive dermal roof pat- tern, including intertemporal, (i) traces of lateral line grooves on dermal bones of skull roof, (j) open palate, but with ptery- goids still extending anteriorly to nearly meet cultriform jiroe- esses on the anteromedial edges of the interpterygoid vacuities, (k) apparently movable basal articulation, (1) lower jaw with a modest retroarticular process and coroiioid jirocess, the last named structure occurring in many, though not all trimero- rhaehoids. Only in one specific point is Eugyri)ius too specialized to be a trimerorhachoid ancestor — the fact that, unlike conditions in any trimerorhachoid, the quadratojugal enters into the jaw articulation. However, an edopsoid resembling Eugyrinus in all features except this one specalization Avould be a reasonable trimerorhachoid ancestor. The possible relationship of trimerorhaehoids to the Triassic brachyopids has been treated by Watson, Nilsson and Romer. Watson (1956) proposes Eohrachyops and Dvinosaurus (1919. 1956) as ancestors of the Triassic brachyopids; Dvinoscn())is, according to that author, represents a more or less intermediate staue in a line stemming from an ancestral form close to CHASE : NELDASAimrS WRIGIITAE 221 Eobravhyops in the Lower Permian. Nilsson (1937) agrees in general that Dvinosaurus. tliough not directly ancestral, may be close to the ancestry of brachyopids. Romer (1947) denied relationship of Dvinosaurus to the brachyopids, considering it closer to Saurerpeton and Trivicrorhachis. and suggested the metoposaurs as brachyopid ancestors. In its parabolic skull outline and anteriorly placed orbits, Dvinosaurus superficially resembles the brachyopids. However, the pattern of dermal bones in the skull roof of Dvinosaurus, including reduction of the postparietal and the topography of the postorbital region, is in contrast to conditions in the brachy- opids where the postparietals are not notably reduced and the postfrontal-supratemporal contact, as Romer points out, is more reasonably derived from the more "normal" pattern of typical rhachitomes. Further, the lacrimal contact with the orbit and naris in Dvinosaurus differs from the brachyopid condition, in which the lacrimal is typically reduced and withdrawn from the orbit border. The skull table of Dvinosaurus is shorter than in the brachyopids. AVatson "writes off" such features as a mov- able basal articulation and tlic greater anterior extent of the pterygoids in Dvinosaurus as merely the retention of primitive characters. However, the retention of such a remarkably primi- tive feature as a movable basal articulation at such a late date is a condition one would hardly expect if we were, in Dvinosaurus, confronted by an immediate brachyopid ancestor. Other palatal structures of Drinosaitrus are also features hardly to be expected in an innnediate ancestor of the brachy- opids. In Dvinosaurus, the vomers are expanded to meet the ptei-ygoids on the lateral border of the interpterygoid vacuities, so that the palatine, which in Dvinosaurus is a remarka])ly small bone, takes no part in the border of the interpterygoid vacuity, whereas in brachyopids the palatine forms the anterior border of that opening. Perhaps significant is the presence of anterior palatal fenestrae in Dvinosaio-iis — none are recorded in the brachyopids. Some differences between Dvinosaurus and the brachyopids may reflect a more primitive level of organization in Dvinosaurus and its obvious neoteny. However, the number of morphological differences that exist between Dvinosaurus and the brachyopids rule against a close relationship. Eohrachijops is more like the bracliyopids than is Dvinosaurus in such features as: less broadly rounded snout; nares, though 222 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY widely separated, closer together than in Dvinosaurus; orbits of more "normal" proportions; very shallow cheek below orbit; pattern of postorbital region ; lacrimal, though still in contact with orbit, greatly reduced; longer skull table than in Dvino- saurus: entrance of palatine into the margin of interpterygoid vacuity. On the other hand, such specialized characters in Eohrachyops as tiny tabulars, great reduction of sensory canal grooves, ab- breviated maxilla, peculiar pattern of palatal bones adjacent to choanae, and position of quadrate articulation further forward than in some brachyopids, tend to remove it from a directly an- cestral position. In i)hu"ing Eohrachyops close to the ancestry of the brachy- opids, Watson listed five characters in which Eohrachyops and Dvinosaurus resemble the brachyopids and "differ from all other labyrinthodonts" (1956, p. 365). This, which would appear to be a clinching argument, loses strength on closer examination. Two of the five characters, it is true, are found only in Eohrachy- ops, I)ri)iosau)-us and brachyopids. These are: (1) a distinct space, formerly occupied by a cartilaginous ridge on the posterior surface of the quadrate, separates the hinder border of the ptery- goid from those of the squamosal and ({uadratojugal ; (2) the outer surface of the squamosal and (juadratojugal passes round onto the posterior face of the quadrate and there forms a later- ally concave, nearly vertical surface. However, the other three charact<'rs cited by Watson are not exclusive to Eohrachyops, Dvinosaurus and the brachyopids. but are present in other trimerorhachoids (Sauro'peton, Acrop- Jous), and two of the three — projecting occiput (m?toposaurs. plagiosaurs) and ventral position of the quadrate condyle (edop- soids, eryopsoids) — may also be found in other labyrinthodont groups. If Eohrachyops is a Permian representative of the brachyopids, special cliaracters and typically advanced features in Eohrachij- ops would certainly debar Dvinosaurus from a position inter- mediate between Eohrachyops and later brachyopid-;. LITERATUKE CITED Amalitsky, Y. p. iy:i-±. On the Dvinosainidac, a family of labyrinthodonts from the Permian of Xorth Russia. Ann. :\rag. Xat. Hist., (9)13: 50-64. CHASE : NELDASAURUS WBIGHTAE 223 Broom, E. 1913. Studies on the Permian teiimosiiondylous stesoeeiilialiaiis of North America. Bull. Anier. Mus. Nat. Hist., 32: r)63-59(i. Bystrow, a. p. 1935. Morphologische Uiitersuclmugen der Deekkuoiheu des Schiideb der Wirbeltiere. I. Mitteilung. Schiidel der Stegoeephalen. Acta Zool., 16: 65-141. 1938. IJrinosaurus als ueotenisclie Form der Stegoeephalen. Acta Zool., 19: 209 295. Bystrow, A. P. and J. A. Ekremov 1940. BentJiosucJtu.'i suslil-ini Efr. — a lal)yrinthodont from the Eo- triassie of Sharjenga River. Trav. Inst. Pal. Acad. S;i. TTRSS, 10: 1-152. Carroll, R. L. 1964. Early evolution of the dissorojiliid aiii]i]iiliiaus. Bull. Mus. ("omp. Zool., 131(7): 161-250. Case, E. C. 1911. Revision of the Aiiiidiiliia and Pisces of the Permian of Xortli America. Puld. Carnegie Inst. Washington, Xo. ]4(i: 1-179. 1915. The Permo-Carboniferous red beds of North Anu'rica and their vertebrate fauna. Publ. Carnegie Just. Washington, No. 207: 1-176. 1922. New reptiles and stegocephalians from the Ujijier Triassic of western Texas. Publ. Carnegie Inst. Washington, No. 321: 1-84. 1935. Description of a collection of associated skeletons of Trimcro- rhachis. Contrib. Mus. Pal. Univ. Michigan, 4: 227-274. Chase, J. N. 1963. The labyrinthodont dentition. Breviora, Mus. Comp. Zool., No. 187: 1-13. Colbert, E. H. 1955. Scales in the Permian aniidiiliian Trlrm V(irha<-his. Anier. Mus. Novit., No. 174(1: 1-17. Cope, E. I). 1878. Descriptions of extinct Batracliia and Reptilia from the I'ermian formation of Texas. Proc. Amer. Philos. Soc, 17: 505-530. Dunbar, Carl O. et al. 1960. Correlation of the Peiinian formations of North America. Bull. Geol. Soc. Amer., 46: 1763-1806. Efremov, J. A. 1932. On the Permo-Triassic labyrinthodonts from USSR. I. The labyrinthodonts of the Campylian beds of mountain Gieat Bogdo. II. On the morphology of Bvinosanrns. Trav. Inst. Pal. Acad. Sci. URSS, 1 : .17-67. 1940. Preliminary descrijjtion of the new Permian and Triassic Tetra- poda from USSR. Trav. Tn.st. Pal. Acad. Sci. URSS, 10(2): 1-140. 224 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Fritsch, a. 1901. Fauna der Gaskohle mid der Kalksteine der Perniforniation Bohmeiis. Vol 1. Prague. HOTTON, N. Ill 1959. Acroplous vorax, a new and unusual labyrinthodont amphibian from the Kansas Permian. Jour. Paleont., 33(1): 161-178. MORAN, W. E. 1952. Fossil vertebrates of the tri-state area. Art. 1. Location and stratigraphy of known occurrences of fossil tetrapods in the upper Pennsylvanian and Permian of Pennsylvania, West Virginia and Ohio. Ann. Carnegie Mus., 33: 1-4-4. NiLSSON, T. 1937. Bin Plagiosauride aus dem Ehat Sehonens. Beitriige zur Kenntnis der Organisation der Stegocephalengruppe Brachyo- poidei. Acta Univ. Lund, 34(2) : 1-75. Olson, E. C. 1948. A preliminary report on vertebrates from the Permian Vale Formation. Jour. Geol., 56: 186-198. 1951a. Vertebrates from the (Jhoza Formation, Permian of Texas. Jour. Geol., 59: 178-181. 1951b. Fauna of the Upper Vale and Choza : 1-5. Fieldiana, Geol., 10: 89-128. 1951c. Fauna of the Upper Vale and Choza: 6. Fieldiana, Geol., 10: 147-166. 1955. Fauna of the Vale and Choza: 10. Fieldiana, Geol., 10(21): 225-274. 1962. Late Permian terrestrial vertelirates, U.S.A. and U.S.S.R. Trans. Amer. Philos. Soc, (N.S.; 52(2) : 1-224. Olson, E. C. and J. R. Beerboweb 1953. The San Angelo formation, Permian of Texas, and its verte- brates. Jour. Geol., 61 : 389-423. Parrington, F. R. 1949. A theory of the relations of lateral lines to dermal bones. Proc Zool. Soc. London, 119(1) : 65-78. ROMER, A. S. 1930. The I'ennsylvanian tetrapods of Linton, Ohio. Bull. Amer. iMus. Nat. Hist., 59: 77-147. 1935. Early history of Texas rcdiieds vertel)rates. Bull. Geol. Soc. Amer., 46: 1597-1658. 1945. The late Carboniferous vertelirate fauna of Kounova (Bohemia) compared with that of the Texas redlieds. Amer. Jour. Sci., 243: 417-442. 1947. Review of the Labyriuthodontia. Bull. ^Fus. Comp. Zool., 99(1) : 1-368. 1952. Fossil vertebrates of the tri-state area. Art. 2. Late Pennsyl- vanian and early Permian vertebrates of the Pittsburgh-West Virginia region. Ann. Carnegie Mus., 33: 47-110. CHASE : NEIiDARATTRrS WRIftHTAE 225 1958. The Texas I^erniiaii rodbods and tlieir vertebrate fauna. In, Studies on Fossil Vertebrates, WestoU, T. S., editor. Univ. London, Athlone Press: l-17-17n. RoMER, A. S. AND E. V. Witter 194:^. Edops, a primitive rliaeliitoniuus amphibian from the Texas red- beds. Jour. Geol., 50: 925-960. Steen, M. 1931. The British Museum eolleetion of Amphibia from the Middle Goal Measures of Linton, Ohio. Proe. Zool. Soc. London, 1930 (1931): 849-891. SUSHKIN, P. P. 1936. Notes on the pre-Jurassie, tetrapods from USSE. III. Dvino- saurus Amalitzki, a perennibranchiate stegocepha'ian from the Upper Permian of the North Dvina. Trav. Inst. Pal. Aead. Sci. UESS, 5: 43-91. TiLTON, J. L. 1926. Permian vertebrates from West Virginia. Bull. Geol. Soc. Amer., 37: 385-396. Watson, D. M. S. 1919. The structure, evolution and origin of the Amphibia. The "orders" Ehachitomi and Stereospondyli. Philos. Trans. Eoy. Roc. London, (B), 209: 1-73. 1921. On Eiigyrinus wildi (A. Smith Woodward), a branchiosniir from Lancashire coal-measures. Geol. Mag., 58: 70-74. 1940. The origin of frogs. Trans. Eoy. Soc. Edinburgh, 60: 195-231. 1956. The lirachyopid labyrinthodonts. Brit. :\rus. CXat. Hist.), Geol. Bull., 2(8): 318-391. Westoll. T. S. 1943. The origin of tetrapods. Biol. Rev., 18: 78-98. Whipple, E. W. and E. C. Case 1930. Discovery of Permo-Carboniferous vertebrates in the Dunkard formation of West Virginia. Jour. Washington Acad. Sci., 20(14) : 370-372. Williston, S. W. 1915. Trimerorhachis, a Permian temnospondyl amphibian. Jour. Geol., 23: 246-255. 1916. The skeleton of TrimrrorJiachis. Joui-. Geol., 24: 291-297. Woodward, A. S. 1891. On a microsaurian (Ilylonomous ivildi, sp. nov.) from the Lan- cashire coal field. Geol.]\rag., (3) 8: 211-212. BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY ^-^■■P^' ■' PLATE 1 Xeldasaurus icrightae, n. gen., ii. sp. Holotype skull in dorsal view. MCZ 2200, Block A, X -66 approx. CTIASE : XELDASAT'RfS WRTGTITAE PLATE 2 Xrhld.sauras lorightae, n. gen., ii. sp. Ilolotype skull in palatal view. MCZ 2200, Block A, X -66 approx. BULLETIX : MUSEUM OF COMPARATIVE ZOOLOGY PLATE 3 N eld u. sail ni.s icriglitue, slab showing dorsal vertebrae in series and a par- tially articulated left forelimb. MCZ 1371. Block A, X 1 approx. Bulletin of the Museum of Comparative Zoology HARVAKD UNIVEKSITY Vol. 133, No. 4 THE GENERA OP THE CHILOCORINI (COLEOPTERA, COCCINELLIDAE) By Edward A. Chapin Museum of Comparative Zoology CAMBEIDGE, MASS., U.S.A. PRINTED FOR THE MUSEUM September 10, 1965 Publications Issued by or in Connection WITH THE MUSEUM OF COMPARATIVE ZOOLOGY HARVAPvD UNIVERSITY Bulletin (octavo) 1863 — The current volume is Vol. 133. Breviora (octavo) 1952 — No. 227 is current. Memoirs (quarto) 1864-1938 — Publication terminated with Vol. 55. Johnsonia (quarto) 1941 — A publication of the Department of Mollusks. Vol. 4, no. 43 is current. Occasional Papers of the Department op Mollusks (octavo) 1945 — Vol. 2, no. 31 is current. Proceedings of the New England Zoological Club (octavo) 1899-1948 — Published in connection with the Museum. Publication terminated with Vol. 24. The continuing publications are issued at irregular intervals in num- bers which may be purchased separately. Prices and lists may be obtained from the Publications Office of the Museum of Comparative Zoology, Cambridge 38, Massachusetts. Peters' Check-List of Birds of the World. Volume 1 is out of print ; volumes 2 to 7, 9, 10 and 15 are currently available. Phylogeny and Evolution of Crustacea. Proceedings of a confer- ence held at Cambridge, Mass., March 6-8, 1962. H. B. Whittington and W. D. I. Rolfe, editors. Cambridge, Mass., 192 pp., 80 figs., 1963. Cloth bound, $6.75. Fishes of the Gulf of ]\raine, by Henry B. Bigelow and William C. Schroeder. AVashington, viii - 577 pp., 1953. Photo-offset reprint, $6.50. Publications of the Boston Society of Natural History The remaining stock of the scientific periodicals of the Boston Society of Natural History has been transferred to the Museum of Comparative Zoology for distribution. Prices for individual numbers may be had upon request. Bulletin of the Museum of Comparative Zoology HAEVAED UNIVEESITY Vol. 133, No. 4 THE GENERA OP THE CHILOCORINI (COLEOPTERA, COCCINELLIDAE) By Edward A. Chapin Museum of Comparative Zoology CAMBEIDGE, MASS., U.S.A. FEINTED FOE THE MUSEUM September, 1965 Bull. -Mils. Coinp. Zool., Hiuvnnl Univ., 133 (4): 227-271, Sept. 1965 Xo. 4. — The Genera of the Chilocorini {Coleopi< va, Coccinellidae) By Edward A. Chapin INTRODUCTION It is now ninety-one years since the latest attempt tu evaluate the described genera of Coccinellidae of the world was made by G. R. Crotch (A Revision of the coleopterous family Coccinel- lidae, London, 311 pages, 1874). Since 1874, there has appeared a bibliographic catalog of the species of this family by R. Kors- chefsky (Coleopterorum Catalogus, Junk-Schenkling, Partes 118, 120, Coccinellidae, 659 pages, 1931-1932). A critical review of the genera and species of this family is long overdue but has now become virtually impossible for any one worker to ac- complish. So it seems best to select some small, well circum- scribed group within the famil.y as a starting point. I have chosen the Chilocorini, a well known and easily recognized tribe. This tribe is a very compact group of genera distributed generally throughout the world. In the Korchefsky catalog twenty-four generic or subgeneric names are listed. One of these, Clanis, is a preoccupied name that has already been re- placed. A second, Notolipernes, had been placed as a s;snionym of Telsimia and removed from this tribe at the time that the catalog was going through press. Corystes Mulsant was de- scribed in the Hyperaspini, and its placement has been under discussion since Chapuis {in Lacordaire, 1876, Genera Coleopt., 12:244, 249) referred it to the Chilocorini. The genus is here returned to the Hyperaspini where it apparently belongs. I have found no explanation for the inclusion of Elpis Mulsant in the Chilocorini and, after an examination of the type species, E. dolens Mulsant, have returned it to the Coccinellini to a position near Menochilus Timberlake. There remain twenty names that are properly included in this tribe. Of these, I have been able to examine specimens, usually of the type species, of seventeen. In the course of my work I have found it necessary to propose two new genera. As far as possible the characterizations of the genera have been taken from the respective tjqDe species. 230 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY I wish to express my thanks to Dr. P. J. Darlington, Jr. for encouragement and assistance in securing necessary material for study, to Dr. R. D. Pope for specimens of Elpis dolens Mulsant, Endochilus plagiatus Sicard, and Endochilus styx Sicard, to Mr. E. B. Britton and Mr. G. F. Gross for specimens of three species of Orcus. My thanks are also due to Mrs. Morna MacLeod for preparing the illustrations from my tracings and to Mrs. Neil Barth for typing the manuscript. Specimens for study have been dissected, the pertinent parts cleared of muscular tissue in cold KOH 10 per cent solution, and passed through the alcohols and oil of wintergreen (methyl salicylate) into Canada balsam. The drawings were made by tracing the outlines projected onto paper at various magnifica- tions through a Bioscope projection apparatus. There follows a discussion of the parts studied. STRUCTURE AND TERMINOLOGY Antenna. The number of segments varies from ten to seven. The basal segment may be more or less cylindrical or it may be produced laterally at its apex in a rounded or subconical lobe (HaJmus, Priasus, Orcus). The apical segment may be very short and embedded in the penultimate segment or it may be as long as or longer than the penultimate segment {Anisorcus, Orcus, Chilocorus, Egius, Priasus). Mandibles. In place of the supplementary tooth (absent in all Chilocorini), there is a short knife-like ridge on the ventral aspect running backward a short distance from the apex. In general the mandibles follow a pattern throughout the Chilo- corini. In three cases {Axion, Egius, Priasus) the apices are obtusely triangular, without the hooked tip of the others. In Halmus, the right, and to a lesser extent the left, mandible has a more or less broad triangular internal tooth. Rudiment of this tooth can be seen on the right mandible of Anisorcus. Maxilla. 1 have not attempted to describe the differences which exist in the maxilla proper, but have confined my atten- tion to the palpus. The terminal segment, described as securi- form for the family as a whole, varies from that shape in several instances. In Anisorcus, Egius, and Phaenochilus the segment is cylindro-acuminate while in Chilocorus and Priasus the sides are almost parallel. In Halmus the segment is barrel-shaped Avitli oblique apex. CHAPIN : GENERA OF THE CHILOCORINI 231 Liijulu. Ill general this sclerite is soft, quadrate, with an- terior angk^s rounded. It bears the labial palpi which show some variation. The terminal segment may be shorter than the preceding, or longer, as in Harpasus and Axion. It may be slender, cylindro-acuminate as in E.iochomus or short, stout, and conical as in Halmus. The number of visible abdominal sternites has been considered as one of the prime characters in coccinellid taxonomy. How- ever, in the Chilocorini, most of the genera have six visible sternites in the male and five in the female. This is because the fifth in the male is truncate or broadly emarginate, thereby dis- closing the sixth sternite. In the genus Orcus, the fifth is as broadly rounded in the male as in the female, thereby con- cealing the sixth in both sexes. On the other hand, in the genera Halmus, Anisorcus, and Egiius, the fifth is truncate or emargi- nate in the female as in the male, thereby disclosing the sixth in both sexes. Except for these slight difi'erences, the fifth sternite is unmodified. The first visible sternite carries the metacoxal arcs — thin ridges arising at the sides of the inter- coxal process of the sternite and extending out. In the Chiloco- rini we find the same types that have been long used in the classifications of the Scymnini. Thus, in Exochomus, Brunms (male), and Brumoidcs the arcs are complete or closed, as in Fullus. In Axiom, Brumus (female), Cladis {Clanis renamed), Curinus, Harpasus, and Zagrcus, the ares are incomplete but directed forward at their extremities as in Scymnus. These genera are characterized by the presence of tibial spurs. The arcs are incomplete and directed outward at their extremities or are fused with the posterior margin of the sternite in those genera without tibial spurs. Afiisorcus, Orcus, and Priasus have arcs directed outward, as in Nephus, while Chilocorus, Egius, and Halmus have arcs fused with the margin of the sternite as in Diomus. Legs. The femora are unmodified throughout the group, ex- cept for a shallow groove for the reception of the tibiae. The tibiae show some modifications. In Brumus and Brumoidcs, the legs are unusually slender, and in Brumoidcs especially, the tibiae are many times longer than broad. Three genera, Chilo- corus, Egius, and Phacnochilus have developed a stout, tri- angular tooth on the outer margin of the basal half of the tibia. 232 BULLETIN: MUSEUM OP COMPARATIVE ZOOLOGY Claws. Ill general the form of the tarsal claws is characteristic of the genus. In this tribe there are three genera with com- pletely simple claws {Brunius, Brumoides, and Priasus) ; claws with a very small triangular tooth basally or slightly postmedian are present in Endochilus and Parapriasus (in part) ; and claws with a heavy triangular or subquadrate tooth are in the other genera. Usually the tA'pe of claw can be determined from a dry specimen but it may be necessary to mount a claw in balsam in order to perceive a minute tooth as in Endochilus or in some species of Parapriasus. In the figures which accompany this paper, all illustrations of claws are drawn to the same scale. Male genitalia. The median lobe, except in one genus, is tri- angular in dorsal view. It may be perceptilily asymmetrical as in Axion, Chilocorus, Curinus, Egius and Exochomus. It may he less than twice as long as broad, in Brumoides and Claelis, or as much as seven times longer than broad as in Egius and Exo- cJwniiis. In Egius, the lobe is longer than the parameres, in other genera equal {Chilocorus) or shorter. The exception is the genus Halmus, in Avhich case the lobe is parallel sided, longer than the parameres, the tip upturned and in side view concave in the basal two-thirds and again in the apical third. The parameres are slender throughout as in Halmus and Priasus or are moderately broad and spatulate and may be slightly or strongly constricted in basal half. In the genus Anisorcus they are very broad and concave, without any basal constriction. The trahes is simple, slightly curved, and slightly thickened at its free extremity. It is unusually thin in Axion. The sipho does not vary much from genus to genus. Moderately stout and curved, with the siphonal capsule more or less similar in all genera but Anisorcus, and the apex variously modified, the apex has more value for specific than for generic identifications. In the illustrations, the aedeagus and sipho are always drawn to the same scale. Female gcmtalia. The recepiaculum seminis is of a type not to be found, to the best of my knowledge, elsewhere in the Coccinellidae. The body is stout, very much so in the l)asal part, the ramus is entirely absent, the nodulus usually absent but represented in some genera {Egius, Zagreus) by a small button- like dome, and the cornu small, rounded, and bent at right angles to the body. In four genera (Anisoreus, Chilocorus, Phaenochilus, and Egius) most of the cornu is replaced by an appendix, definitely separate from the body. In the genus CHAPIN: GENERA OF THE CHILOCORINI 233 Halmus the body is more slender and appears nearly coceinel- line in its form. The accessory gland is moderate to very long, sometimes, as in Axion, several times as long as the recepta- culum. The sperm duct is extraordinary as among Coccinel- lidae in my experience. The duct consists of two parts : one of small diameter arises at the head of the bursa copulatrix, at the base of the infundibulum, and connects with the second part of greater diameter, which leads to the receptaculum. These parts are roughly equal in length. Two genera deviate from the norm. In Anisorcus, the first part seems to be an ex- tension of the bursa ; it is of greater diameter than usual but is very thin-walled and definitely shorter than the second part. In Halmus the first part is longer than the second and almost as great in diameter. The infundibulum is present in all but four genera and is very variable in shape and size. In Egius, Phaenochilus, and Chilocorus it is replaced by a fleshy pro- tuberance, unsclerotized, on the head of the bursa; in Anisor- cus it is completely lacking. The bursa copulatrix in all genera is a simple thin-walled sac with no modifications, such as are seen in certain species of Hippodamia. The hemisternites (di- vided eighth sternite) are usually subtriangular, more or less elongate, each with a small button-like stylus at apex. In the genera Endochilus, Anisorcus, Egius, Phaenochilus, and Chi- locorus, the styli are absent while in Halmus they are cylindri- cal and longer than wide. In the key which follows it will be noticed that half of the genera studied lack tibial spurs, all but two of which {Chilocorus and Egius) are restricted to the Old World, while the other lialf of the genera possess tibial spurs and all but two {Brumus and Brumoides) are confined to the New World. Of these four exceptions, one of each pair is nearly world- wide, the other closely related to it. This suggests that Bru- moides developed first in the Ncav World, then spread into most of the Old World, with Brumus as an Old World offshoot, and that Chilocorus was originally an Old World genus which invaded the New World, with Egius as its New World segre- gate. 234 BULLETIN : MUSEUM OF COMPAKATIVE ZOOLOGY SYSTEMATIC SECTION Tribe CHILOCORINI Costa Costa, 1849, Fauna Eegno Napoli, 1:9. Coccinellidae of small to large size, 2.0-8.0 mm, form oval to nearly circular, moderately to strongly convex, upper sur- face glabrous or pubescent. Antenna of less than eleven seg- ments, short, terminal segments forming a fulsiform club, its origin concealed beneath the genal extension of the clypeus which enters or passes immediately beneath the eye. Mandible without supplementary apical or subapical tooth but with a sharp ridge which runs back a short distance from apex on the ventral face. Prosternal lobe without carinae. Abdomen usually with six visible sternites in male, five in female, some- times five in both sexes or six in both sexes. Legs usually normal, occasionallj' with modified tibiae, tibiae with or with- out terminal spurs, tarsi four-segmented, tarsal claws simple, or swollen at base, or with basal tooth. Male genitalia : median lobe of aedeagus usually elongate triangular, compressed, often asymmetrical near apex, parameres shorter or longer than the median lobe, occasionally modified at apices. Female genitalia : receptaculum seminis very stout, somewhat bent, ramus absent, nodulus occasionally developed but never prominent, cornu a small terminal portion, sometimes with an appendix, accessory gland usually very large. Sperm duct very long, of two parts. The part leading into the receptaculum, usually half the total length, is a moderately coarse tube, the part leading from the bursa a thinner, usually much thinner, tube. Infundibulum present or absent; if present the sperm duct arises at its base. Bursa copulatrix a large, simple sac. Hemisternites usually elongate, sometimes short, subtriangular, rounded at apices, styli small, usually button-shaped with two to four long setae, some- times absent. KEY TO GENEBA OF TEIBE 1. Tibial spurs present on legs II and III, metacoxal ares as in Scymnus or Piilliis, antenna 10, 9, or 8-segniented 2 Tibial spurs absent on all legs, metacoxal arcs as in Xcphus or Diomus, antenna 9, 8, or 7-segmented 9 2. Tarsal claw simple, sometimes slightly thickened at base 3 Tarsal claw with tooth in basal half, which is sometimes quadrate, sometimes acute, often very small 4 3. Antenna 10-segmented, Old World Brumus Mulsant Antenna 8-segmeuted, world-wide Brumoides n. gen. 4. Antenna 10-scgmented 5 CHAPIN : GENERA OF THE CHILOCORINI 235 Antenna 9-segmented Hnrpasus Mulsant Antenna 8-segmented Zagreus Mulsant 5. Elytra! epipleura f oveolate for reception of tips of femora 6 Elytral epipleura not foveolate 8 6. Elytra without retlexed margins, with marginal bead, size small to moderate, up to 6 mm 7 Elytral margins feebly but distinctly retlexed, with or without marginal bead, size very large, up to 8 mm . . Axion Mulsant 7. Size larger, up to 6 mm, elytra blue, parameres not appendieulate Curinus Mulsant Size smaller, up to 3.75 mm, elytra black or bluisli l>lack, with red basal spot, parameres appendieulate Arawana Leng 8. Metacoxal ares complete or virtually so . Exocliomus Eedtenbacher Metacoxal arcs widely incomplete Cladis Mulsant 9. Eeceptaculum seminis without apical appendix, infundibulum usually present (absent in Endochilus) 10 Eeceptaculum seminis with an apical appendix, infundibulum absent 14 10. Antenna 9-segmented 11 Antenna with less than 9 segments 12 11. Tarsal claw without basal tooth but somewhat swollen at base, elytral margin feebly retlexed in apical half Priasus Mulsant Tarsal claw with sharp quadrate basal tooth, elytral margin strongly retlexed throughout Parapriasus n. gen. 12. Upper surface wholly or in large part pubescent, infundibulum absent Endochilus Weise Upper surface glabrous, anterior angles of pronotum sometimes sparsely hairy, infundibulum present 13 13. Antenna 7-segmented Hahnus Mulsant Antenna 8-segmented Orciis ]\lulsant 14. Antenna 7-segmented, all tibiae without tooth on outer margin in basal half Anisorcus Crotch Antenna 8-segmented, tibiae II and III, sometimes I, with tri- angular tooth on outer margin at extremity of tarsal groove 15 15. Terminal segment of maxillary palp one and one-half times as long as wide, its apex strongly oblique, elytral margin feebly retlexed Chilocorus Leach Terminal segment of maxillary palp two or three times as long as wide, apex acuminate 16 16. Elytra strongly alutaceous, the lateral margins not retlexed, terminal segment of maxillary palp about twice as long as wide, cylindro-acuminate, tooth on tarsal claw weak, Xew World Egius Mulsant Elytra not alutaceous, the lateral margins strongly retlexed, ter- minal segment of maxillary palp three times as long as wide, gradually narrowed to a blunt point, tooth on tarsal claw very strong, nearly as long as apical portion of claw. Old World Phaenochihis Weise 236 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY BrUMUS Mulsant Brumus Mulsant, 1850, Species Trimeres Seeuripalpes, p. 492; Crotch, 1874, Eevision of the Coceinellidae, p. 195; Weise, 1885, Best. -Tab. Europ. Coleopt., Coeciiielliden, II, ed. 2, p. 5. Type species. Cocoinella octosignata Gebler, through synonymy with C. desertorum Gebler, by subsequent designation of Crotch, 1874. Chilocorini with form broadly oval, moderately convex, upper surface glabrous. Antenna ten-segmented ; first segment stout, less than twice as long as wide, constricted at base ; second almost as wide as first, slightly longer than wide ; third to fifth mutually similar in form, the fifth shorter ; sixth and seventh mutually similar, each wider than long ; eighth and ninth large, eighth almost as long and ninth longer than sixth and seventh together ; tenth small, more or less embedded in the oblique apex of the ninth. Mandibles stout, somewhat angulate on the outer margin. Terminal segment of maxillary palp rather elongate, apex strongly oblique, lacinia with a row of three shorter spines along the edge of the row of long setae. Terminal segment of labial palp cylindro-acuminate, rounded Fig. 1. Brumus octosignatus (Gebler). In this figure and all subsequent figures the lettering is as follows : A — antenna, B — mandibles, C — maxilla, B — ligula, E — legs, F — claw, G — first abdominal sternite, S — aedeagus, I — sipho, J — sclerotized portions of female genitalia. CHAPIN : GENERA OF THE CHILOCORINI 237 at apex. Prosteriial lobe moderately wide, truncate at apex. Elytra with lateral margins very slightly reflexed, finely beaded, epi- pleura not foveolate for reception of the femoral apices. Abdo- men with six visible sternites in male, five in female. Meta- coxal arcs almost complete in male, widely incomplete in female. Legs slender, femur of leg III a little stouter than the others, tibial spurs present, tarsal claws unusually long and slender, without basal tooth or swelling. Male genitalia : median lobe lanceolate, slightly wider at middle of length, four and one-half times as long as greatest width ; parameres moderately wide at base and apex, constricted near middle of length, slightly longer than the median lobe ; trabes rather stout, as long as the main parts of the aedeagus ; sipho rather long and slender, of nearly uniform diameter through most of its length, apex blunt, with a series of oblique lines or grooves shortly before apex. Female genitalia : receptaculum seminis stout, cornu rather small, bent and rounded ; accessory gland large ; sperm duct shorter than usual, the parts equal in length, the slender part very slender; infundibulum present, rather longer than usual ; hemisternites short and blunt, triangular, apices rounded, styli apical, button-shaped, with two or three long setae each. This genus was established by Mulsant to include two species, Coccinella desertoriim Gebler (=C. octosignata Gebler) and Coccinella sutiiraJis Fabricius. I have removed the second species to serve as the type of a new genus, Brumoides. Brumus octosignatus (Gebler) is the only species of Brumus known to me. Brumoides new genus Brumiis, — Leng, 1908, Journ. New York Ent. Soc, 16:34; Casey, 1908, Can. Ent., 40:409. Type species. Coccinella suturalis Fabricius, by present designation. Chilocorini with form oval, moderately convex, upper surface glabrous. Antenna eight -segmented ; first segment short and stout, slightly bent; second, as wide as first at base, gradually tapering to half that width at apex; third, fourth, and fifth similar and nearly equal, all slightly longer than wide ; sixth one and one-half times longer than wide, of nearly the same diameter as fifth ; seventh about equal in length to sixth, apex obliquely truncate ; eighth small, partly embedded in apex of seventh. Mandibles not heavily developed, outer margin strongly convex in apical two-thirds, concave in basal third. Terminal 238 BULLETIN: MUSEUM OP^ COMPARATIVE ZOOLOGY segment of maxillary palp securiform with apex strongly oblique, lacinia with row of three short, stout spines along the edge of the row of longer setae. Labial palp rather long and slender, the terminal segment cylindro-acuminate, the apex rounded. Prosternal lobe narrow, truncate. Elytral margin very finely reflexed, with bead, epipleura not foveolate for the reception of the femoral apices. Abdomen with six visible sternites in male, five in female. Metacoxal ares complete. Legs slender, femora not noticeably inflated, tibiae slender, tibial spurs present, tarsi rather long, tarsal claws moderately long, slightl^y thickened at base but Avithout angular basal tooth. Male genitalia : median lobe triangular, less than twice as long as wide at base ; para- meres thin, spatula-shaped, hardly at all constricted in basal half, nearly twice as long as median lobe ; trabes rather slender, equal in length to the main parts of the aedeagus; sipho mod- erately slender, of nearly even diameter in basal half, tapering slightly in apical half, apex twisted and terminating in a short finger-like process. Female genitalia : receptaculum seminis much as in Brumus; sperm duct moderately long, the thin portion not much more slender than the thick portion ; infundi- bulum present, small ; hemisternites moderately long, parallel- sided to just before apices where they taper to blunt points. Fig. 2. Bnnnoifles suturalis (Fabricius). CHAPIN : GENERA OP THE CHILOCORINI 239 styli small, button-shap(Ml. willi two or three long setae from each. The following species have been studied and belong to Brii- moides -. Coccmella suturalis Fabricius, Exochonius foudrasii Mulsant, E. hoegei Gorham, E. dcscrtorum Casej^ E. j)arvi- collis Casey, E. histrio Fall, Exochonius {Brumus) davisi Leng, E. (B.) ncvadensis Leng, Brumus septentrionis Weise. HaePASUS Mulsant Orcus {Harpasus) Mulsant, 1850, Species Trimeres Seeuripalpes, p. 473. Harpasus, — Crotch, 1874, Revision of the Coccinellidae, p. 190 (as synonym of Curinus Mulsant). Type species. Orcus {Ear2)a.svs) pallidilahris Mulsant, by subsequent designation of Crotch 1874. The ensuing description has been taken from specimens of Orcus (Harpasus) zonatus Mulsant, one of the originally in- cluded species. Chilocorini with form very broadly oval, nearly circular, moderately convex, upper surface glabrous. Antenna nine- segmented; first segment short and stout, narrowed at base; second as long but not quite as wide as first ; third about as long as width at apex which is twice width at base ; fourth to sixth mutually similar in shape, the fourth half as long as third and slightly wider ; fifth and sixth each a little longer and wider than the preceding segment ; seventh almost as long as fifth and sixth combined and still wider; eighth more than twice as long as seventh, sides nearly parallel ; ninth small, conical. Mandibles with outer margin feebly concave in basal half, then sharply convex to apex. Terminal segment of maxillary palp elongate, the sides nearly parallel, the apex strongly oblique, lacinia with row of four short, stout spines along the edge of the row of longer setae. Terminal segment of labial palp cylindro-acuminate, rounded at apex. Prosternal lobe rather broad, its lateral margins strongly grooved, the groove then following the margin of the coxal cavity. Elytral margin nar- rowly reflexed, finely beaded, epipleura foveolate for the recep- tion of the femoral apices. Abdomen with six visible sternites in male, five in female. Metacoxal arcs widely incomplete. Legs slender, the femora not much enlarged, tibiae simple, tarsi normal, tarsal claws bent, basal tooth large, quadrate. Male genitalia : median lobe lanceolate, greatest width in basal third, four times as long as wide ; parameres spoon-shaped, strongly 240 BULLETIN: MUSEUM OF COMPAEATIVE ZOOLOGY Fig. 3. Harpasus zonatus Mulsant. constricted in basal half, longer than the median lobe by its greatest width ; trabes rather stout, a little longer than the median lobe; sipho stout, of uniform diameter throughout most of its length, apex slightly twisted. Female genitalia : recepta- eulum seminis stout, cornu small; just before the cornu the body is girdled by a slightly prominent band ; sperm duct long, the thick portion relatively not as thick as usual, the thin por- tion extremely thin ; infundibulum present ; hemisternites tri- angular, rounded at the apices, styli small, button-shaped, with two or three long setae each. Harpasus zonatus Mulsant is the only species belonging to this genus that I have been able to study. From descriptions, it appears to be quite distinct from the type species, H. palli- dilahris Mulsant and an examination of that species may show that the two are not congeneric. ZaGREUS Mulsant ExocJiomiis (Zagreus) Mulsaut, 1850, Species Trimeres Securipalpes, p. 488. Type species. Exochomus (Zagreus) bimaculos^^■s Mulsant, by present designation. CHAPIN : GENERA OF THE CHILOCORINI 241 Chiloeorini with form very broadly oval to nearly circular, moderately convex, upper surface glabrous. Antenna eight- segmented ; first segment stout, slightly bent, half again as long as wide ; second as stout at base as first, tapering to apex, as long as wide; third and fourth nearly equal in length, each wider at apex than at base, fourth wider than third; fifth as long as fourth but wider; sixth as long as wide; seventh about as long as first, with apex strongly oblique ; eighth short, conical, em- bedded in seventh. Mandible moderately stout, its outer margin broadly curved. Terminal segment of maxillary palp subsecuri- form with apex strongly obliciue, lacinia with oblique row of seven slender spines on outer face. Terminal segment of labial palp cylindro-acuminate, a little more than twice as long as wide, truncate at apex. Prosternal lobe narrow. Abdomen with six visible sternites in male (in a related species), five in female. Fig. 4. Zagreus himaculosus Mulsant. Metacoxal arcs incomplete, ending about half way to base of segment, posterior margin of fifth sternite broadly rounded in female. Legs with tibial spurs, femora moderately stout, tibiae slender, shallowly excavate near apices, the margin of excavation 242 BULLETIN: MUSEUM OP COMPARATIVE ZOOLOGY edged with a row of slender bristles. Tarsal claws wdth stout, subquadrate basal tooth on each. Male genitalia not studied. Female genitalia : receptaculum seminis with nodulus distinct from body, but not prominent, eornu curved, rounded at apex, without appendix ; accessory gland about twice as long as receptaculum ; sperm duct long, the thick portion much thicker than the very thin portion ; infundibulum present ; hemister- nites short and stout, triangular, about twice as long as wade at base, apices broadly rounded, styli short, button-shaped, with long setae. The following species have been determined as belonging to the genus Zagreus. Exochomus (Zagreus) himaculosus Mulsant, E. (Z.) cinctipennis ]\Iulsant, Exocliomus jordani Mulsant, E. adelae Crotch, E. histillafus Weise, E. decempunctatus Weise, E. guttatus Weise, E. suhcoendeus AVeise, and E. ritchiei Si- card. AxiON Mulsant Exocliomus {Axion) Mulsant, 1850, Species Trimeres Securipalpes, p. 477. Axion, — Crotch, 1874, Eevison of the Coccinellidae, p. 191; Gorham, 1892, Biol. Centr.-Amer., Ins., Coleop., 7:176; Casey, 1899, Journ. New York Ent. Soc, 7:105; Leng, 1908, Journ. New York Ent. Soc, 16:34. Type apecies. Coccinella tripustulata DeGeer, by subsequent designation of Crotch, 1874. Chilocorini with form subcircular, size large to very large, strongly convex, upper surface glabrous. Antenna ten-seg- mented; first segment almost straight, twice as long as narrow median portion ; second barrel-shaped, very little longer than wide ; third through sixth similar in shape, narrow at base and widening from base to apex ; third as long as wide, fourth and fifth each shorter than the preceding segment, sixth slightly .longer than fifth ; seventh through tenth forming a compact oval club, seventh and eighth equal in length, ninth longer with the apex oblique, tenth small, deeply embedded in ninth, conical. Mandible very stout, subangulate near middle of outer margin. Terminal segment of maxillary palp wider toward apex than at base, apex strongly oblique, lacinia with an irregular row of seven or eight stout spines on the outer side. Terminal segment of labial palp cylindro-acuminate, two and one-half times longer than wide. Prosternal lobe narrow, truncate at apex. Pronotum without fine marginal line across base. Elytra with lateral margins slightly reflexed, epipleura with very small foveae for reception of the femoral apices. Abdomen M-ith six CHAPIN: GENERA OF THE CHILOCORINI 243 Fig. 5. Axion tripustulatum (DeGeer). visible sternites in male, five in female, fifth sternite of male broadly emarginate. Metacoxal arcs broadly incomplete. Femora not noticeably stout, tibiae moderate, feebly excavated at apices for reception of first segments of tarsi, tarsal claw with very strong quadrate, plate-like tooth in basal half. Male genitalia: median lobe long, slender, slightly asymmetrical in apical third ; parameres noticeably longer than median lobe, paddle-shaped and constricted at basal third; trabes very slender, about as long as a paramere; sipho slender, siphonal capsule well de- veloped, apex of sipho blunt, slightly twisted. Female geni- talia: receptaculum seminis much as in Curinus; sperm duct proportionately shorter than in Curinus and differing in that the thick portion is longer than the thin portion ; inf undibulum present, inverted Y-shaped; hemisternites parallel nearly to apices, then rounded triangular, styli button-shaped, each with three or four long setae. This purely North American genus is in need of revision. Eight trivial names have been associated with the generic name ^Axion ; four of these were considered as valid specific names in the world catalog. 244 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY CUEINUS Mulsant Orciis (Ctirinus) Mulsant, 1850, Species Trinieres Securipalpes, p. 472. Curinus, — Crotch, 1874, Eevision of the Coccinellidae, p. 190 ; Gorham, 1892, Biol. Centr.-Anier., Ins., Coleop., 7:176; Koischefsky, 1932, Coleopt. Catalogus (Jimk), pars 120, p. 252. Type species. Orcus (Curinus) coeruleus Mulsant, monobasic. Chilocorini with form nearly circular, strongly convex, upper surface glabrous. Antenna ten-segmented ; first segment stout, slightly curved, slightly longer than wide ; second barrel-shaped, one and one-half times longer than wide and nearly as wide as the first; third half as wide at base as second, nearly twice as wide at apex as at base ; fourth nearly equilateral but slightly wider apically; fifth through tenth forming a compact fusiform club ; fifth slightly wider than long ; sixth and seventh almost equal in length, each wider at apex than at base; eighth longer and wider than seventh, maximum width of club at apex of eighth; ninth about twice as long as eighth, sides gradually tapering to apex which is oblique ; tenth short and conical, Fig. 6. Curinus coeruleus Mulsant. CHAPIN : GENERA OF THE CHILOCORINI 245 slightly Avider than long. Mandible stout, subangulate at mid- dle of outer margin. Terminal segment of maxillary palp with lateral margins subparallel and apical margin strongly oblique, laciuia with irregular row of five stout spines on outer face. Terminal segment of labial palj) elongate cylindrical but with outer margin slightly curved, about three times longer than wide. Prosternal lobe flat, moderately broad. Elytral margins slightly reflexed, with fine marginal bead, epipleura foveolate for the reception of the femoral apices. Abdomen with six visible sternites in male, five in female, apical sternite of male emar- ginate or notched, of female simple and rounded. Metacoxal arcs on first sternite incomplete, failing to reach base of ster- nite. Legs with femora moderately stout, tibiae near apices shallowly excavate, the margins of the excavation edged with rows of slender bristles, tarsal claws stout with subquadrate basal tooth. ]\Iale genitalia : median lobe slightly asymmetrical ; parameres longer than the median lobe, constricted near basal third ; trabes slender, about as long as paramere ; sipho mod- erately stout, of even diameter nearly to apex, where the tube undergoes a quarter turn, the apex truncate. Female genitalia : receptaculum seminis more or less globular, basal portion with- out well-defined nodulus or ramus, cornu short, small, and bent ; accessory gland long and slender ; sperm duct very long, the thin part about twice as long as the thick part ; infundibulum present, inverted Y-shaped; hemisternites elongate triangular, styli button-shaped, with three or four long setae from each. With the removal of Curinus peleus (Mulsant) to Exochomus, and the reestablishment of Harpasus, with the transfer of its species from Curinus to it, Curinus is left with the type species and one very doubtfully included species from Chile, Curinus ruizi Brethes. Brethes figures the antenna as having eleven segments, which if true, makes that species unique in the tribe. I think it more likely that there is an error in the drawing and that the species belongs in Harpasus. ArAWAXA Leng Exochomus {Arawaiia) Leng, 1908, Journ. New York Ent. Soc, 16:34, 38; Casey, 1908, Can. Ent., 40:409. Type species. Exochomus arisoniciis Casey, by original designation of Leng. Chilocorini with form broadly oval, strongly convex, upper surface glabrous. Antenna ten-segmented; first segment short and stout, bent at almost a right angle; second equally stout, 246 BULLETIN: MUSEUM OP COMPARATIVE ZOOLOGY as long as wide ; third strongh' wedge-shaped, much narrower at base than at apex, about as long as width at apex ; fourth to seventh similar in shape ; fourth and fifth nearly equal, sixth and seventh each slightly wider and longer than the preceding segment ; eighth one-half longer but hardly wider than seventh, ninth twice as long as eighth and as wide, its apex strongly oblique; tenth small, embedded in apex of ninth. Mandibles stout, somewhat angulate at middle of outer margin. Terminal segment of maxillary palp twice as long as wide, apex strongly oblique, lacinia with irregular row of six stout spines along the edge of a row of long setae. Terminal segment of labial palp slender cylindro-acuminate, rounded at apex. Prosternal lobe moderately broad, truncate. Elytral margin not reflexed, very finely beaded, epipleura foveolate for the reception of the femoral apices. Abdomen Avith five visible sternites in both sexes. Metacoxal arcs nearly complete. Legs moderately stout, femur of leg I a little stouter than the others, tibia I with the outer margin expanded into a thin keel, tibiae II and III with spurs, tarsal claws stout, strongly hooked, with triangular tooth in basal half. Male genitalia : median lobe lanceolate, four times longer than its greatest width at basal third ; parameres noticeably longer than median lobe, strongly constricted near Fig. 7. Araicana arizonica (Casey). CHAPIN : GENERA OF THE CHILOCORINI 247 base, each with a finger-like process at its apex; trabes slender, one and one-half times as long as the median lobe ; sipho long in proportion to the aedeagus, of uniform diameter throughout most of its length, siphonal capsule small, apical portion of sipho slightly expanded and twisted. Female genitalia : reeep- taculum seminis much as in the genus Curinus; sperm duct with the thicker portion considerably longer than the extremely thin portion ; inf undibulum present, inverted Y-shaped ; hemi- sternites about twice as long as wide, generally tapering to the blunt apices, styli apical, button-shaped, each with two or three long setae. In addition to the type species, I am including two other species in the genus Arawmia, Exochomus scapularis Gorham from Central America and Exochomus cuhensis Dimmock from Cuba, this last with some hesitation. E. cuhensis is included because, while the fore tibiae do not have the thin keel but are simply broadly curved on their outer margin, the male genitalia are quite similar to those of the other two species and do have the finger-like processes at the apices of the parameres. I con- sider the structure of the genitalia of more importance than that of the fore tibiae. All three species are similar in general appearance. Exochomus Redtenbache Exochomus Redtenbacher, 1843, Tentauieii dispositiouis Coleopterorum pseudotrimerorum, p. 11 ; Eeprint of aljove in Gerniar, 1844, Zeitseh. f. Ent., .5:121; Mulsaiit, 1850, Species Trimeres Securipalpes, p. 476; Crotch, 1873, Trans. Amer. Ent. Soc, 4:376; 1874, Revision of the Coecinellidae, p. 192; Barovsky, 1922, Ann. Mus. Zool. Russie, 23:293, figs, la, lb, Ic; Korschefsky, 1932, Coleopt. Catalogus (Junk), Coe- cinellidae, Pars 120, p. 252. In 1922, Barovsky proposed a division of ExocJiomus into three subgenera, based on the structure of the tarsal claws, as follows : I. Unguiculi dente valido basali instructi Subg. Exochomus, s.str. II. Unguiculi dente indisiincto basali instructi Subg. Parexochomus new III. Unguiculi dente subacutangulo post medium instructi . . Subg. Anexochomus new 'o 248 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY ExocHOMUs (ExocHOMus) Redteiibacher ExocJiomus (Exochomus) — Barovsky, 1922, Ann. Mus. Zool. Eussie, 23:291, fig. la. Type species. Coecinella 4-pustulata Limie (quadrip-ustulata auct.), sub- sequent designation of Korschefsky, 1932. Two previous type designa- tions for this genus had been made, both by Crotch and both invalid, since the species designated were not among those originally included. Chilocorini with form broadly oval to almost circular, mod- erately convex, upper surface glabrous or pubescent. Antenna ten-segmented ; first segment slightly curved, a little less than twice as long as wide ; second barrel-shaped, as long as wide and same diameter as first ; third obconical, half as long and half as wide as second; fourth, fifth, and sixth nearly equal, each slightly wider apically than at base; seventh similar in shape to sixth but about half as long as wide, and forming with the eighth, ninth, and tenth a slender fusiform club ; eighth almost as long as wide, at base equal in width to apex of seventh, greatest width of club at apex of eighth ; ninth as long as eighth, conical with apex strongly oblique; tenth small, conical deeply embedded in ninth. Mandible heavy, angulate at middle of outer face. Terminal segment of maxillary palp subsecuri- form with apex strongly oblique, lacinia with row of five long, slender (not short, stout) spines on outer face. Terminal seg- ment of labial palp cylindrical, twice as long as wide. Pro- sternal lobe narrow, truncate at apex, the anterior coxae almost contiguous. Pronotum very finely margined across base, lateral margins slightly reflexed. Elytral margin strongly beaded, epipleura not foveolate for the reception of the femoral apices. Abdomen with six visible sternites in male, five in female. Meta- eoxal arcs complete or virtually so. Legs with moderately stout femora, tibiae slender, tarsal claws strong, with subquadrate basal tooth on each. Male genitalia : median lobe long, slender, parallel in basal two-thirds, slightly asymmetrical in apical third; parameres about one-eighth longer than the median lobe, slender, somewhat constricted in basal half, spoon-shaped in apical half; trabes slender, almost as long as main body of the aedeagus; sipho moderately slender, of even diameter through- out most of its length, siphonal capsule rather heavy, apex of sipho twisted. Female genitalia: receptaculum seminis a stout body without clearly defined nodulus or ramus, cornu more slender than body, stronglj- curved; accessory gland slightly longer than receptaculum ; sperm duct long, the parts about CHAPIN : GENERA OF THE CHILOCORINI 249 Fig. 8. Exochomus quadripustulatus (Linne). equal in length, the thick part noticeably thicker than the thin part ; infundibulum present, tapering slightly from a broad Imse and slightly curved; hemisternites elongate, subparallel, about three times as long as wide at base, apices acutely rounded, styli button-shaped, each with two or three long setae. The following species have been verified as belonging to Exochomus sensii stricto: Coccinella quadripustulata Linne, C. aethiops Bland, C. flavipcs Thunberg, C. marginipennis Leconte, C. nigripcnnis Erichson, Orcus pelews Mulsant, Exochomus childreni Mulsant, E. calif or nicus Casey, E. contristatus Mul- sant, E. jamaicensis Sicard, E. latiusculus Casey, E. lituratus Gorham, E. metaUicus Korchefsky, E. mormonicus Casey, and E. uropygidialis Mulsant. Eighteen species, originally described in Exochomus, have been removed from this genus. Two are here placed in Arawmia, seven in Zagreus, and seven in Brumoides. Two species, Exo- chomus tricoJoratus Gorham and E. championi Gorham are re- moved from the Chilocorini and should be placed in the Coc- cinellini near Cycloneda. 250 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY ExocHOMus (Parexochomus) Barovsky Exocliomus {Parexochomus) Barovsky, 1922, Ann. Mns. Zool. Eussie, 23:292, fig. lb. Type species. Exochomus puhcscens Kiister, by present designation. I have been unable to examine any specimens conforming to Barovsky 's description. Barovsky assigns the following palae- arctic species to this subgenus : E. (P.) puhcscens Kiister, E. (P.) anchorifcr Allard, E. (P.) semenovi Weise, and E. (P.) kirgi- zorum Barovsky. Exochomus (Anexochomus) Barovsky Exochomus (Anexocliomus) Barovsky, 1922, Ann. Mus. Zool. Eussie, 23:292, fig. Ic. Type species. Exocliomus undul-atus Weise, by present designation. A single female specimen from Pusan, Korea, in the collec- tion of the Museum of Comparative Zoology appears to be a color variant of E. (A.) mongol Barovsky. It satisfies the orig- inal description in every respect except that the subapical spot on the elytron is missing. Its tarsal claws conform to figure Ic of the original description. A study of the gross anatomy shows no character of importance that would warrant a change of status for Anexochomus. Barovsky assigns the following species to this subgenus : E. (A.) un(lulatusWehe,E. (A.) kiritshenkoi Barovsky, and E. (A.) mongol Barovsky, all from the palaeare- tic region. -e^ ClAJ)IS Mulsant Exochomus (Clanis) Mulsant, 1850, Species Trimeres Securipalpes, p. 479 (name preoccupied). Exochomus (Cladis) Mulsant, 1850, idem, Appendix, p. 1033 (new name for Clanis Mulsant not Hiibuer 1819). Cladis, — Crotch, 1874, Eevision of the Coccinellidae, p. 192; Weise, 1883, Wien. Ent. Zeitseh., 2:67; Gorham, 1894, Biol. Centr.-Amer., Ins., Coleopt., 7:179. Type species. Coccinella nitidula Fabricius, through synonymy with Ex- ochomus (Clanis) uva Mulsant, by subsequent designation of Crotch, 1874. Chilocorini with form broadly oval, moderately convex, upper surface glabrous. Antenna ten-segmented ; first segment slightly bent, about one and one-half times longer than wide; second equal in width to first, almost equilateral; third as long as sec- ond but only half as wide ; fourth nearly as wide as third. CHAPIN : GENERA OF THE CHILOCORINI 251 equilateral; fifth to eighth rather similar, each wider and shorter than the preceding ; ninth large, equal to the seventh and eighth conihined; tenth small, more or less embedded in the oblique apex of the ninth. Mandible with the outer margin moderately curved, with no unusual modifications. Terminal segment of maxillary palp broad and short, apex stronglj- oblique, lacinia with two or three stout spines along the edge of the row of long setae. Terminal segment of labial palp cylindro-acuminate, rounded at apex. Prosternal lobe narrow, truncate at apex. Elytra with lateral margin not reflexed, epi- pleura not foveolate for reception of the femoral apices. Abdo- men with six visible sternites in male, five in female. Metacoxal Fig. 9. CJadis nitidula (Fabrit-ius). ares widely incomplete. Legs with femora not unusually stout, tibiae simple, rather slender, shallowly excavate near apices, tarsal claws strongly curved, each with a large, quadrate basal tooth. Male genitalia : median lobe acutely triangular, two and one-half times longer than wide at base; parameres very 252 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY broad in apical half, strongly constricted in basal half, notice- ably longer than the median lobe ; trabes moderately stont, curved, about as long as a paramere from basal attachment to apex ; sipho moderately stout, of nearly uniform diameter throughout, apex somewhat twisted, blunt. Female genitalia : receptaculum seminis much as in Curinus or ExocJwmus, acces- sor}^ gland moderate in size ; sperm duct moderately long, the thick portion slightly longer than the thin portion; infundi- bulum present, inverted Y-shaped; hemisternites moderately heavy, elongate triangular, apices rounded, styli minute, button- shaped, each with two or three long setae. Mulsant segregated two supposedly new species under the name Clanis, which name had been previously used by Hiibner. At present it is considered that Mulsant 's "species" are not distinct and are synonyms of the much earlier Fabrician species, which occurs on various of the West Indian islands. It has also been reported from Buenos Aires. PrIASUS :\lulsant Orcus (PriasHs) Mulsant, 18.50, Species Trimeres Securipalpes, j). 4(37; Crotch, 1874, Eevision of the Coceinellidae, p. 188. Type species. Coccinclla bilunulata Boisduval, by suljscqueiit designation of Crotch, 1874. Chilocorini with form nearly circular, strongly convex, upper surface glabrous. Antenna nine-segmented ; first segment slightly bent, strongly constricted at middle, apically produced laterally in a small, rounded lobe ; second stout, barrel-shaped, nearly equilateral ; third, fourth, and fifth trapezoidal, decreasing in length and slightly increasing in width at apices; sixth longer than fifth and slightly wider ; seventh and eighth each longer than the preceding, the eighth slightly produced laterally at apex ; ninth longer than eighth, subcylindrical, apex slightly oblique. Mandible stout, the outer margin not strongly curved, apex unusually blunt. Terminal segment of maxillary palp with outer margin convex, inner margin concave, the sides nearly parallel, apex oblique ; lacinia with regular row of four stout spines on outer face. Terminal segment of labial palp cylindro-acuminate, truncate at apex. Prosternal lobe rather narrow, slightly dilated behind, apex rounded. Pronotum with a strong marginal line. Elytra with lateral margins only slightly reflexed, epipleura not foveolate for the reception of the femoral apices. Abdomen with six visible sternites in male, CHAPIN : GENERA OF THE CHILOCORINI 253 five in female; metaeoxal arcs incomplete, running almost into the posterior margin of the sternite in female, curving toward anterior margin for a short distance in male. Legs stout, an- terior femora about twice as long as wide, tibiae simple, slightly expanded apically, tarsal claws short and stout, swollen at base, without basal tooth. Male genitalia : median lobe elon- gate triangular ; parameres straight, parallel-sided, rounded at apices, noticeably longer than median lobe ; trabes as long as a paramere with its basal attachment; sipho slender, of even diameter except near the siphonal capsule where it is somewhat thicker, apex twisted and with a short, sharp apical process. Fig. 10. Priasus hihniuJatus (Boisduval). Female genitalia : receptaculum seminis stout bulbous, nodulus hardly differentiated, ramus absent, cornu short, stout, and rounded ; accessory gland about four times longer than the receptaculum; sperm duct rather short, the thick portion equal in length to the thin portion and about three-fourths as long as the accessory gland; infundibulum present; hemisternites elongate, the sides almost parallel to near apices Avhich are abruptly narrowed and rounded, styli button-shaped, hardly prominent, each with two or three long setae. 254 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Mulsant placed three species, all Australian, under this name. I have been able to study all three and have found that, Avhile the type species has toothless claAvs, the other two have claws which are equipped with a basal tooth on each. In one case, Coc- cinella aiistralasiac Boisduval, the tooth is large and conspicu- ous, in the other, C. nummularis Boisduval, the tooth is much smaller and is somewhat transparent and difficult to see. These two species have been referred to the following genus. ParAPEIASUS new genus Type species. Coccinella australasiae Boisduval, by present designation. Chilocorini with form nearly circular, strongly convex, upper surface glabrous. Antenna nine-segmented; first segment a little longer than second, bent, constricted at basal third, apically widened : second barrel-shaped, wider at base than at apex, nearly equilateral ; third to seventh similar in shape, third longer than wide, fourth shorter than third, fifth shorter than fourth or sixth, seventh as long as sixth but slightly wider; eighth longer than wide ; ninth longer than eighth and tapering to a blunt point. Mandible stout, apex short and acute, outer margin subangulate at middle of length. Terminal segment of maxillary palp subsecurif orm, apex oblique ; lacinia with a patch of three or four short and stout bristles, in addition to the row of long setae. Terminal segment of labial palp cylindro- acuminate, rounded at apex. Prosternal lobe moderate, convex, margined laterally, the bead prolonged part way around the coxal cavity. Elytral margin rather strongly reflexed, without marginal bead ; epipleura not foveolate for the reception of the femoral apices. Abdomen with six visible sternites in male, five in female, the fifth sternite broadly truncate in male, broadly rounded in female. Metacoxal arcs broadly open, the line running from the intercoxal process out and back almost to the posterior margin of the sternite, then curving forward for a short distance. Legs moderately stout, tibiae slightly widened toward apices, tibia I with a row of small denticles along outer margin, tarsal claw stout, with a subquadrate, trans- lucent tooth in basal half. Male genitalia : median lobe about as long as the parameres, slender, elongate triangular but with the lateral margins nearly parallel in basal two-thirds, apical third tapering to a slender blunt apex ; parameres slender, not noticeably constricted, apices broadly rounded ; trabes slender, slightly curved, a little shorter than the median lobe ; sipho CHAPIN: GENERA OF THE CHILOCORINI 255 Fig. 11. Parapriasus australasiae (Boisduval). rather similar to that of Priasus. Female genitalia : receptacn- lum seminis similar in shape to that of Priasus but less stout, especially the cornu; accessory gland about twice as long as receptaculum ; sperm duct with thick portion a little longer than the very thin portion ; infundibulum a simple, bar-like structure; hemisternites long, slender, and tapering to blunt apices, styli terminal, button shaped, each giving rise to two or three long setae. Two species have been seen which belong to this genus, Coc- cinella australasiae Boisduval and C. nummularis Boisduval, both from Australia. In the latter the tooth on the tarsal claw- is less well developed and less easily seen. EnDOCHILUS Weise EndoChilus Weise, 1898, Deutsch. But. Zeitsch., p. 119, pi. 1, fig. 10, 1910, Verb. Naturf. Ver. Briinn, 48:52; Sicard, 1929, Ann. Mag. Nat. Hist., (10) 4:518. 256 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Type species. Endochilus cavifrons Weise, by subsequent designation of Korschefsky, 1932. As the type species is not available to me, I have prepared the generic description and illustrations from specimens of E. plagiafiis Sicard. Chilocorini with form nearly circular, not strongly convex, margins broadly but not abruptly reflexed, upper surface wholly or partly set wdth fine pubescence. Antenna eight-segmented ; first segment constricted at middle, twice as long as width at constriction ; second elongate globose, slightly shorter and wider than first; third, fourth, and fifth wedge-shaped, each slightly shorter and wider than the preceding, third and fourth together about as long as the second ; sixth longer and wider than fourth and fifth together ; seventh slightly longer than sixth but hardly wider, the margins nearly parallel ; eighth conical, as long as sixth, its apex rounded. Mandible with outer margin evenly curved. Maxillary palp very stout, the terminal segment elongate-conical, four times as long as the penultimate segment ; lacinia with a row of four short, stout spines on outer face. Labial palp stout, the terminal segment conical, a little longer than the penultimate segment. Prosternal lobe broad, flat, not margined. Elytral margin subexplanate, slightly reflexed, with- out bead ; epipleura weakly f oveolate for the reception of the femoral apices. Abdomen with five visible sternites in either sex, in both the fifth sternite is broadly rounded. Metacoxal arcs run from the sides of the rather broad intercoxal process almost straight toward the posterior angles of the sternite, where the lines are slightly recurved. Legs slender, femur I a little stouter than the others, tibiae simple, tarsal claws gently and evenly curved, each with a small, obtuse, triangular tooth at base. Male genitalia : median lobe lanceolate, its greatest width at middle of length, apex acuminate ; parameres slightly longer than the median lobe, their bases concave and partly embracing the lobe ; trabes slender, about as long as median lobe, gradu- ally broadened from apical third to apex; sipho long and slen- der, of nearly uniform diameter throughout most of its length, siphonal capsule small, apical portion slightly enlarged just back of the acuminate apex. Female genitalia : receptaculum seminis bent sausage-shaped, strongly constricted at basal fourth, nodu- lus not prominent, cornn stout, without apical appendix, acces- sory gland small, globular ; sperm duct short, the thick portion shorter than the length of the receptaculum, the thin portion CHAPIN : GENERA OF THE CHILOCORINI 257 Fig. 12. Endochilus -plagiatus Sieard. shorter than the thick portion ; infnndibulum absent ; hemister- nites elongate, sides nearly parallel, about five times longer than average width, apices rounded, styli absent, represented by two long, apically placed, setae. A genus of a few, below average size, species, mostly from West Africa south of the Gulf of Guinea. The genus is im- mediately distinguished from other chilocorine genera by the peculiar conformation of the head. The genal lobes and the elypeus form the transverse frontal margin of the head, with a forward extension below the eyes equal to half the long diam- eter of the eye. This margin passes below the eye and joins the temporal areas of the head. Thus there is no invasion of the eye itself. HaLMUS Mulsant Orcus (Ealmus) Mulsant, 1850, Species Trimeres Securipalpes, p. 471. Ealmus, — Weise, 1923, Arkiv Zool., 15(12): 134. Type species. Coccinella chalyhea Boisduval, the genus being originally monobasic. Chilocorini with form nearly circular, moderately convex, upper surface glabrous. Antenna seven-segmented; first seg- ment bent, constricted in basal half, apically produced laterally 258 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY in a conical lobe ; second nearly equilateral, barrel-shaped ; third slender, twice as long as wide, the apex oblique ; fourth and fifth similar in shape and proportions, the fifth half again as large as the fourth ; sixth as wide at base as fifth, beyond middle it tapers to half its basal diameter ; seventh conical, slightly longer than its diameter at base. Mandibles rather delicate, the outer margins strongly curved, apices acute ; just distad of the median notch the inner margin is cut away so that it appears to be tridentate. Terminal segment of maxillary palp broadly oval, the apical margin oblique ; lacinia with an irregular row of six or seven stout spines on outer face. Terminal segment of labial palp short, stout, and conical. Prosternal lobe moderately broad, truncate at apex. Pronotum without basal marginal line. Elytra with lateral margins reflexed, epipleura not foveolate for the reception of the femoral apices. Abdomen similar in both sexes, with six visible sternites, the fifth sternite in both sexes truncate. Metacoxal arcs incomplete, running parallel and close to the posterior margin of the sternite. Legs with femora not noticeably stout, tibiae simple, tarsal claws strongly curved, with broad, quadrate, plate-like tooth in basal half. Male genitalia : median lobe long, narrow, parallel-sided nearly to apex where it is rounded, as seen in dorso ventral ^dew ; in Fig. 1.3. Eahnus chalybetis (Boisduval). CHAPIN : GENERA OF THE CHILOCORINI 259 lateral view it is somewhat slipper-shaped, the apex or "toe" turned up and with a triangular projection in the dorsal margin at apical third; parameres very slender, slightly shorter than the median lobe ; trabes a little longer than body of aedeagus ; sipho slender, of even diameter throughout, apex somewhat twisted and with a short process. Female genitalia : receptaculum seminis curved, sausage-shaped, nodulus and ramus not defined, cornu separated from the main body by a slight constriction and broadly rounded apically, accessory- gland longer than the receptaculum ; sperm duct comparatively short, the thick sec- tion shorter than the thin section but of almost the same diameter except at the junction point ; infundibulum present ; hemister- uites short, and broadly triangular, styli cylindrical, nearly twice as long as the diameter at base, each with two long setae. Halnius chalybeus is a median sized species, 3.0-3.7 mm long, clear blue to blackish-blue above except that in the male the head and flanks of the pronotum are yellowish-white. This is one of the few genera that shows a sexual difference in coloration. The species is further distinguished by the unusual form of the genitalia of both sexes. Weise was apparently the first to raise Mulsant's subgenus Halmus to full generic status. He did this because of the un- usual shape of the genae. Here the genal lobe, which extends out over the antennal insertion and on into the eye, is abruptly narrowed at the inner margin of the eye and is continued into the eye as a fine line. This character, apparently unique among the Chilocorini, when taken together with the other character- istics which deviate from the norm, indicates that the genus stands in a rather isolated position in the tribe. For the present, I am referring Orcus ovalis Blackburn to this genus. It agrees with Hahnus chalyheus in its antennae, mandi- bles, ligula with labial palps, legs, metacoxal arcs, visible sixth sternite in the female, and receptaculum seminis. It differs in the broadly securiform maxillary palps, the male genitalia (which are simple as in most Chilocorini), and most importantly in the form of the clypeus and genal lobes. The clypeus is not more or less deeply emarginate without a marginal bead, as in most other Chilocorini, but is straight across the front, without a trace of an emargination, and with a strong marginal bead. The genal lobes are as in most other Chilocorini, not suddenly contracted at the inner margin of the eye. "When more of the Australasian "Orcii.^" are thoroughly studied, it may seem best to establish 0. ovalis as the type of a distinct genus. 260 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY OrCUS Mulsant Orcus Mulsaut, 1850, Species Trimeres Seeuripalpes, p. 465; Crotch, 1874, Eevision of the Coccinelliclae, p. 188 ; Weise, 1923, Arkiv Zool., 15(12) :133. Type species. Orcus jantliinus Mulsant, by subsequent designation of Crotch, 1874. Cliilocorini with form nearly circular, moderately convex, upper surface glabrous. Antenna eight-segmented ; first seg- ment slightly curved, less than twice as long as wide, constricted at basal third, apex with a small, rounded lateral lobe ; second barrel-shaped, a little longer than wide; third obconical, much narrower at base than apex of second, less than twice as long as wide ; fourth through seventh mutually similar in shape but becoming progressively longer and wider ; seventh somewhat oblique at apex ; eighth nearly twice as long as wide, rounded acuminate. ]\Iandible heavy, subangulate at middle of outer margin. Terminal segment of maxillary palp stout, subsecuri- form, the apex strongly oblique; lacinia with three or four short, stout spines on outer face. Terminal segment of labial palp cylindrical, slightly tapering, twice as long as wide at Fig. 14. Orcus janthinus Mulsant. CHAPIN: GENERA OF THE CHILOCORINI 261 base. Prosternal lobe moderately broad, truncate at apex. Pro- notuni finely margined across median third of base. Elytra with lateral margins slightly reflexed, epipleura weakly foveo- late for the reception of the femoral apices. Abdomen similar in both sexes, each with five visible sternites, the fifth evenly rounded. Metacoxal arcs broadly open, the line running par- allel to the posterior margin of the sternite, nearly to the lateral margin. Legs w-ith femora moderately stout ; tibiae excavate in apical half or more for the reception of the tarsi ; tarsal claws strongly curved, each with a large quadrate, plate-like tooth in basal half. Male genitalia : median lobe long, lanceolate, not visibly asymmetrical apically ; parameres slender, straight, not broadened in apical half, slightly longer than median lobe; trabes longer than median lobe ; sipho rather slender, terminat- ing in a slender, acute spine. Female genitalia : receptaculum seminis a stout body with nodulus indicated by a slight prom- inence, ramus not defined, cornu strongly curved and broadly rounded at apex, accessory gland much longer than recepta- culum; sperm duct with parts nearly equal in length, the thick section about three times the diameter of the thin section; in- fundibulum present; hemisternites slender, three times longer than wide at base, styli button-shaped, each with two or three long setae. Orcus janthinus is a large species, 5.0-5.7 mm, uniformly steel-blue above with the exception of the reddish-yellow labrum. Beneath, the meso- and metathoraces and legs are castaneous, the antennae and abdomen reddish-yellow. The series studied is from Bogor, Java, taken in February, 1954, by Mr. 0. D. Deputy. I have seen no other species of this genus. AnISORCUS Crotch Anisorcus Crotch, 1874, Eevision of the Coccinellidae, p. 190; Weise, 1902, Termes. Fiizetek, 25:508. Type species. Anisorcus fryi Crotch, liy original designation. Chilocorini with form broadly oval, moderately convex, upper surface glabrous. Antenna seven-segmented ; first segment short, bent, hardly constricted; second barrel-shaped, of same diam- eter and as long as the first; third obconical, shorter than and much narrower than second; fourth and fifth mutually similar in shape, the fifth slightly wider and longer than the fourth but shorter than the third; sixth similar in shape to the second and slightly longer; seventh nearly twice as long but not quite 262 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY as wide as sixth, cyliiidro-aeuminate, the apex rounded. Mandi- bles moderately stout, the inner margin of the right mandible with a small tooth just in front of the notch, outer margins broadly rounded, apices acute. Terminal segment of maxillary palp cylindrical, tapering slightly toward apex which is oblique but not strongly so; lacinia with a patch of nine or ten stout spines on outer side. Terminal segment of labial palp cylindro- acuminate, twice as long as wide at base. Prosternal lobe mod- erately broad, truncate at apex. Pronotum without fine marginal line across base. Elytra with lateral margins slightly reflexed, very finely beaded ; epipleura shallowly foveolate for the recep- tion of the femoral apices. Abdomen alike in both sexes, each with six visible sternites, the fifth sternite in each is truncate. Metacoxal arcs nearly straight from intercoxal process nearly to posterior margin of sternite where they turn abruptly and follow margin nearly to lateral margin of sternite. Legs with femora not noticeably stout ; the tibiae rather broad and short ; tarsal claws each with a strong tooth in basal half. Male geni- talia : median lobe elongate triangular, not noticeably asym- metrical apically ; parameres very broad, not constricted in Fig. 15. Anisorcus fryi Crotch. CHAPIN : GENERA OF THE CHILOCORINI 263 basal half, one-third longer than median lobe ; trabes slender, as long as main portion of the aedeagns ; siplio rather slender, the siphonal capsule poorly developed, apex of sipho slightly recurved and covered by a membraneous hood. Female genitalia : receptaculum seminis with a very stout body, nodulus and ramus not defined, cornu very short and ends in a flat, rounded ap- pendix, accessory gland only a little longer than receptaculum ; sperm duct in two parts, the thick section is of even diameter throughout and only slightly thicker than the thin section at or near the junction of the two sections ; the thin section, nor- mally thin throughout its length, remains of small diameter for but a short distance, then gradually enlarges its diameter until it joins the bursa copulatrix; infundibulum absent; hemister- nites very elongate triangular, rounded at tips, styli apparently absent. Anisorcus fryi is a species of below medium size, 3.0-3.2 mm. piceous black above except for head, pronotum, and narrow lateral margins of the elytra which are reddish-yellow. The under parts are uniformly reddish-yellow. The structure of the female genitalia apparently indicates a relationship with Cliilo- corus and Phaenochilus. The series before me comes from Suva, Fiji Is., taken in March 1960 by Mr. N. L. li. Krauss. Two other species were included in this genus at the time of descrip- tion ; I have seen neither of them. ChILOCORUS Leach Chilocoriis Leach in Brewster, 1815, Edinburgh Encyclopedia, 9:116; Eed- tenbacher, 1843, Tcntamen dispositionis Coleopterorum pseudotrimer- orum, p. 11; Eeprint of above in Germar, 1844, Zeitsch. f. Ent., 5:118; Mulsant, 1850, Species Trimeres Seeuripalpes, p. 452; Crotch, 1874, Eevision of the Coccinellidae, p. 183; Gorham, 1892, Biol. Centr.-Amer. Ins., Coleopt., 7:175. Type species. Cocci7iella cacti Linne, monobasic. Chilocorini wath form broadly oval, strongly convex, upper surface glabrous except for a few hairs on the flanks of the pronotum. Antenna eight-segmented ; first segment elongate, subcylindrical, slightly bent, more than twice as long as wdde at widest part ; second nearly equilateral, dome-shaped, greatest width near base, rounded in apical third; third wedge-shaped, apex twice as wide as base and wider than length ; fourth nearly quadrate, one-third wider than long ; fifth through eighth form- ing a fusiform club, fifth and sixth each equal in length to fourth. 264 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY each wider at apex than at base, seventh slightly longer and wider than sixth, eighth conical, half again as long as seventh but slightly narrower, slightly excavate on one side near apex. Mandible stout, apex slender, outer margin subangulate near middle. Terminal segment of maxillary palp with lateral mar- gins nearly parallel, apical margin strongly oblique, lacinia with a patch of eight stout spines on outer face. Terminal segment of labial palp elongate subconical, slightly more than twice as long as wide at base. Prosternal lobe flat, moderately wide. Fig. 16. Chilocorus cacti (Linne). Elytral margins not reflexed, finely beaded; epipleura shal- lowly foveolate for the reception of the femoral apices. Abdo- ment with six visible sternites in male, five in female ; metacoxal arcs in form of a quarter circle, merging with the posterior margin of the sternite ; margin of fifth sternite transverse, straight, of sixth slightly emarginate in male, margin of fifth evenly rounded in female. Legs with stout femora, tibiae with a triangular tooth on each at basal third, tarsal claws with small, quadrate tooth on each at base. Male genitalia : median CHAPIN : GENERA OF THE CHILOCORINI 265 lobe very slightly asymmetrical, elongate triangular with sides gently curved; parameres hardly longer than median lobe, mod- erately stout ; trabes slender, longer than the main parts of the aedeagus ; sipho moderately stout, of even diameter through- out most of its length, near apex it is twisted through a quarter turn, apex truncate. Female genitalia : reeeptaculum seminis a stout body without dilferentiation into nodulus and ramus, eornu very short and bent, with a falciform appendix at apex, accessor}^ gland very long and slender ; sperm duct long, in the usual two sections, the thick section only slightly thicker than the thin section ; infundibulum absent, replaced by a fleshy, unpigmented protuberance ; hemisternites rather broadly triangular. Chilocorus Leach is very closely related to Egius Mulsant, a monobasic West Indian genus and also to Phacnochilns AVeise of the Malayan region. The absence of tibial spurs and the worldwide distribution of the species of Chilocorus suggests that the Old World is the original home of the genus. Chilocorus is a large genus, with many species in the Palaearctic and Ethiopian regions, fewer in the Indo-malayan, still fewer in the Nearctic, Neotropical and Australian regions. A thorough study of the species would probably result in the splitting of the genus. Egius Mulsant Egius Mulsant, 1850, Species Trimeres Securipalpes, p. 464. Chilocorus, — Crotch, 1874, Eevision of the Coccinellidae, p. 188 (part). Type species. Egius platycepJialus Mulsant, monobasic. Chilocorini with form nearly circular, flattened convex, upper surface glabrous. Antenna eight-segmented; first segment short and stout, nearly equilateral; second longer than first but of equal width, slightly narrower at apex than at base ; third less than half as long as second, twice as wide at apex than at base ; fourth similar in shape to third but longer and wider ; fifth a little shorter but about as wide as fourth; sixth similar in shape to fifth but a little wider ; seventh longer than wide, a little longer than sixth; eighth longer than seventh, tapering to a rounded apex. Mandibles stout, apices rather blunt, outer margin feebly concave in basal half, nearly straight in apical half. Terminal segment of maxillary palp long and narrow, tapering, apex slightly oblique ; lacinia with a row of six or seven short, stout spines parallel to the row of longer setae. Terminal segment of labial palp elongate conical, rounded at apex. Prosternal lobe 266 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY moderate in width, rounded at apex, the margins with a strong; bead. Elytral margin not reflexed, with a fine marginal bead in basal third, epipleura not foveolate for the reception of the femoral apices. Abdomen with six visible sternites in both sexes ; metacoxal ares joining the posterior margin of the sternite near the lateral margin. Legs rather stout; femora not inflated; tibiae with a triangular tooth on outer margin at middle of length ; tarsi broad, tarsal claws each with a feeble basal tooth. Male genitalia : median lobe elongate triangular, slightly asymmetrical Fig. 17. Egius plaiyceplialus Mulsant. near apex, side margins nearly parallel in basal half ; parameres flattened, not constricted in basal half, slightly shorter than median lobe ; trabes moderately stout, equal in length to median lobe ; sipho rather slender, the apex twisted. Female genitalia : reeeptaculum seminis and accessory gland similar to those parts in Chilocorus; sperm duct with slender section very thin, total length of duct proportionately longer than in Chilocorus; in- fundibulum absent, replaced by a fleshy protuberance on the CHAPIN : GENERA OF THE CHILOCORINI 267 bursa copulatrix ; hemisternites short, rather broad, somewhat ereseent-shaped, styli apparently absent. The type and only species of this genus is found on Cuba. While anatomically very close to Chilocorus, its superficial ap- pearance is very different from that of any of the recorded species of Chilocorus. The very strongly alutaceous surface of the head and elytra, and the almost iridescent metallic greenish- purple color of the elytra set it apart immediately. PhAEXOCHILUS Weise Phaenochilus Weise, 1895, Ann. Soe. Ent. Belgique, 39:135; 1913, Philip- pine Journ. Sci., 8, D:241. Type species. Phaenochilus punctifrons Weise, by subsequent designation of Korschefsky, 1932. Chilocorini with form nearly circular, strongly convex, upper surface glabrous. Antenna eight segmented ; first segment short, stout, curved; second longer and thicker than first, much nar- rower at apex than at base ; third nearly twice as long as wide at base, apex half again as wide as base ; fourth to sixth mutually similar in shape, fourth and fifth nearly equal, seventh slightly Fig. 18. Phaenochilus punctifrons Weise. 268 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY longer and wider than fifth ; eighth twice as long as sixth, taper- ing to a blunt point. Mandible somewhat longer than broad, outer margin not strongly curved, apex rather elongate and slender. Maxillary palp long and slender, the terminal segment very long, about three times longer than wide, tapering to a blunt point ; lacinia wdth a patch of about seven stout spines on outer face. Terminal segment of labial palp slender acuminate, rounded at apex. Prosternal lobe moderately wide, convex, rounded at apex, the lateral margins beaded, the bead continued partly around the anterior coxal cavities. Margins of elytra strongly reflexed, with extremely fine marginal bead ; epipleura shallowly f oveolate for the reception of the femoral apices. Abdomen with six visible sternites in male, the fifth sternite broadlj' truncate, five in female, the fifth broadly rounded. Metacoxal arcs run outward and back nearly to the posterior margin of the sternite, thence following the margin nearlj^ to the lateral margin of the abdo- men. Legs rather short ; femora moderately stout ; tibiae slender, those of legs II and III with triangular tooth at the top of the tarsal groove, that on leg III stronger than that on leg II ; terminal segment of tarsus stout, claws stout, the apical half at nearly a right angle to the basal half, the basal tooth elongate, parallel to and nearly as long as the apical half of the claw. Male genitalia : median lobe equal in length to the parameres, widest at apical third, side margins straight in basal two-thirds, apical third triangular, apex acutely pointed ; parameres paddle- shaped, not noticeably constricted basally, apices bluntly rounded ; trabes slender, longer than the median lobe ; sipho rather slender, of even diameter in first four-fifths, apical fifth much more slender and ending in a harpoon-shaped j)oint. Female genitalia : receptaculum seminis very stout, much like that of Chilocorus, appendix of cornu strongly developed, ac- cessory gland somewhat longer than long diameter of the re- ceptaculum; .sperm duct with thick portion more than twice as long as the very thin portion ; infundibulum absent, replaced by a fleshy tab at apex of bursa copulatrix ; hemisternites moder- ately long, triangular, styli not prominent, appearing as pits near apices of these sclerites, each pit furnished with two or three long setae. This small genus of three species, P. punctifrons Weise, P. ruficollis Weise, and P. monostigma Weise, is at home in the Indomalayan region from Java to Mindanao in the Philippines. Specimens of all three species are in my hands. The genus may CHAPIN : GENERA OK THE CHILOCORINI 269 be immediately recognized by the long, slender maxillary palps and the broadly reflexed elytral margins. TriCHORCUS Blackburn Trichorcus Blackburn, 1892, Trans. Proe. Koy. Soe. South Australia, 15:73. Type species. Trichorcus cinctus Blackburn, monobasic. The original diagnosis is quoted in full below. "Trichorcus (gen. nov. Coccinellidarum) Ab Oreo differt corpore baud metallico, subopaco, dense pubescenti. I can find no structural character to distinguish the genus from Orcus, but it seems scarcely possible to place in that genus a non-metallic subopaque species densely clothed with pubes- cence." From the description of the type species little of importance can be learned except that the individuals of this species are a trifle wider than long. It will be necessary to study specimens of this species in order to disclose its relationships. CHIL0C0RU8 (TrICHOCORUS) Sicard Chilocorus {Trichocorns) Sicard, 1921, Bull. Soc. Portugaise Sci. Nat., 8:213; Mader, 1954, Explor. Pare Nat. Albert, Mission de Witte, Lief. 80, Coccinellidae, pars III, p. SO. Type species. Chilocorus pilosus Sicard, by present designation. It is not possible to place this taxon in its proper position in the tribe from the information contained in the above references. Two species were included in the original description and these ditfer in the formation of the anterior tibiae. The type species is said to have angulate but not spinose tibiae, a very small scutellum, upper surface clothed with somewhat long, gray pubescence, elytral margins not at all reflexed, basal margin of pronotum sinuate either side of the scutellum, and other charac- ters mostly relating to color. Unfortunately, through a misreading of the original descrip- tion, Dr. Mader has introduced a false character into his key. Sicard states that the pubescence is "dirigee en avant" on the pronotum only, on the elytra it is "diriges {sic!) en arriere." CORYSTES Mulsant Cnrystcs Mulsant, 1850, Species Trimeres Securipalpes, p. 506; Crotch, 187-1, Kevision of the Coccinellidae, p. 208; Chapuis, in Lacordaire, 1876, Gen. Coleopt., 12:24-4, 249; Gorham, 1894, Biol. Centr.-Amer., Ins.. 270 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Coleopt., 7:182; Weise, 1899, Deutseh. Ent. Zeitscli., p. 269, 1903, ibid., p. 208; 1904, ibid., p. 358; 1927, Arkiv f. Zool., 18A(4), no. 34, pp. 12-13. Type species. Corystes hypocrita Mulsant, monobasic. Mulsant, 1850, placed his new genus Corystes in the third branch, Thalassaires, of his liyperaspiens. He took note of the fact that the structure of the anterior part of the head was dif- ferent from that of his Chilocoriens in that in Corystes there is a break in continuity between the clypeus and genal lobes and no break in other Chilocoriens. Crotch, 1874, agrees wdth Mulsant in placing Corystes near Thalassa but brings these genera to the beginning of his Hyper- aspides, directly following his group, Chilocorides. Chapuis, 1876, removed the genus from his Hyperaspites and placed it as a transitional form in the Chilocorites. However, he noted, as Mulsant had before, that the structure of the head is not typically chilocorine and that otherwise the characters of the genus are largely hyperaspine. Gorham, 1894, added little to the clarification of the problem. He accepted Chapuis' findings and continued Corystes in the Chilocorini. Weise, 1899, working with specimens which he believed at the time were Corystes hypocrita, returned the genus to the Hyper- aspini. In 1903, he added a second species to the genus, noting that the structure of the head of the new species was somewhat different from that of the original species. In 1904, Weise again asserts that the genus is undoubtedly hyperaspine. In 1927, in a posthumous paper, Weise reverses himself on the ground that the species dealt with in 1899, 1903, and 1904, were misidentified. In this paper, he sets up a new genus, Dia- zonema, type I), fallax Weise {Corystes hypocrita Weise, not Mulsant, renamed), in the Hyperaspini and redescribes Corystes hypocrita Mulsant from specimens from San Fermin, Bolivia. He now places Corystes in the Chilocorini but notes the difference between it and the rest of the Chilocorini in the structure of the clypeus and genal lobes. I have before me a short series of a species from Colombia which I am convinced represents an undescribed species of Corystes. There is no question in mj- mind that the species does belong in the Hyperaspini near Thalassa. The antenna with the antepenultimate segment very long, the mandible with a second tooth very near the apex, the male genitalia with a de- pressed, strongly asymmetrical median lobe, and otlier characters CHAPIN : GENERA OF THE CHILOCOBINI 271 are all foreign to the Chilocorini but common to all Hyperaspini known to me. ElPIS Mulsant Elpis Mulsant, 1850, Species Triineres Securipalpes, p. 449; 1866, Mono- graphie ties Coecinellides, pp. 283, 291; Crotch, 1874, Revision of the Coceinellidae, p. 181; Chapuis in Lacordaire, 1876, Gen. Coleopt., 12:191, 193; Sicard, 1909, Ann. Soc. Ent. France, 78:64, 82; Kors- ehefsky, 1932, Coleopt. Catalogus (Junk), pars 120, Coceinellidae, p. 248; Mader, 1941, Explor. Pare Nat. Albert, Mission de Witte, fasc. 34, p. 185; 1954, op. cit., fasc. 80, pp. 117, 118. Type species. Elpis dolens Mulsant, monobasic. The above are all the references to Elpis that I have been able to find in the literature. All of the authors, except Korschefsky, agree in placing this genus in the Coccinellini, near Menochilus. I have examined a specimen of the type species and it is obvious that the genus does belong in the Coccinellini. Korschefsky gave no explanation for placing the genus in the Chilocorini. (Eeeeived 8 December 1964.) Bulletin of the Museum of Comparative Zoology HAEVARD UNIVERSITY Vol. 133, No. 5 COMMENTS ON SOME RECENT CHANGES IN THE CLASSIFICATION OF THE CIIDAB (COLEOPTERA) By John F. Lawrence CAMBRIDGE, MASS., U.S.A. PRINTED FOR THE MUSEUM October 20, 1965 Publications Issued by or in Connection WITH THE MUSEUM OF COMPARATIVE ZOOLOGY HARVARD UNIVERSITY Bulletin (octavo) 1863 — The current volume is Vol. 133. Bbeviora (octavo) 1952 — No. 230 is current. Memoirs (quarto) 1864-1938 — Publication terminated with Vol. 55. Johnsonia (quarto) 1941 — A publication of the Department of Mollusks. Vol. 4, no. 43 is current. Occasional Papers of the Department of Mollusks (octavo) 1945 — Vol. 2, no. 31 is current. Proceedings of the New England Zoological Club (octavo) 1899-1948 — Published in connection with the Museum. Publication terminated with Vol. 24. The continuing publications are issued at irregular intervals in num- bers which may be purchased separately. Prices and lists may be obtained from the Publications Office of the Museum of Comparative Zoology, Cambridge 38, Massachusetts. Peters' Check-List of Birds of the World. Volume 1 is out of print; volumes 2 to 7, 9, 10 and 15 are currently available. Phylogeny and Evolution of Crustacea. Proceedings of a confer- ence held at Cambridge, Mass., March 6-8, 1962. H. B. Whittington and W. D. I. Rolfe, editors. Cambridge, Mass., 192 pp., 80 figs., 1963. Cloth bound, $6.75. Fishes of the Gulf of Maine, by Henry B. Bigelow and "William C. Schroeder. Washington, viii - 577 pp., 1953. Photo-offset reprint, ^6.50. Publications of the Boston Society of Natural History The remaining stock of the scientific periodicals of the Boston Society of Natural History has been transferred to the Museum of Comparative Zoology for distribution. Prices for individual numbers may be had upon request. Bulletin of the Museum of Comparative Zoology HAEVARD UNIVERSITY Vol. 133, No. 5 COMMENTS ON SOME RECENT CHANGES IN THE CLASSIFICATION OF THE CIIDAE (COLEOPTERA) By John F. Lawrence CAMBRIDGE, MASS., U.S.A. PRINTED FOR THE MUSEUM October, 1965 Bull. Mus. Comp. Zool., Harvard Univ., 133(5) : 273-293, October, 1965. No. 5 — Comments on Some Recent Changes in the Classification of the Ciidae (Coleoptera) By John F. Lawrence Musum of Comparative Zoology, Harvard University, Cambridge, Mass. The family Ciidae includes at present 40 genera and about 550 species. The group has been badly neglected in the past, and although a few excellent local monographs have appeared, the last treatment of the world fauna was published in 1848. The European fauna has probably been more extensively studied than that of any other area, but most of the workers have not been specialists and have tended to be provincial in their ap- proach, completely ignoring, for instance, the contributions to the Holarctic fauna made by Japanese and Americans. Lohse (1964) produced a short paper in which some of the European generic concepts were reexamined and 3 new generic group names proposed. This recent contribution is certainly welcome, since it represents the first step in clarifying the relationships within this difficult group. In the present discussion, I will review Lohse 's proposals and reevaluate his concepts in the light of my own work on the North American Ciidae. Since the histories of the various European generic names have been discussed pre- viously, it may seem redundant to include them here ; I think this is justified, however, considering that even Lohse overlooked or misinterpreted some of these historical facts. 1. ENTYPVS, RHOPALODONTUS, AND 8ULCACIS Although the genera Entypus and Rhopalodontus represent distinct taxa whose species are only distantly related, the nomen- clatural histories of the two names are so interwoven that it seems practical to discuss them together. Ropalodontus Mellie (justifiably emended to Rhopalodontus by Gaubil, 1849) was first proposed in 1847 for the species Cis perfomtus Gyllenhal, the main diagnostic character being the expanded, rounded, and externally spinulose protibial apices. In the same year, Redten- bacher described Entypus, which included only the species Cis affinis Gyllenhal (misidentified as Apate fronticornis Panzer), characterized by having 9-segmented antennae, with the 3rd seg- ment longer than the following 3 together, and spinulose protibial 276 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY apices. In the following year, Mellie included C. affinis in his genus Ennearthron, which had been described some months prior to the publication of Redtenbacher's work, and placed Entypus in synonymy. Mellie included the true A. fronticornis in the genus Cis, primarily because of its 10-segmented antennae. Thomson (1863) considered Ehopalodontus and Orophius Red- tenbacher (=Octotemnus Mellie) to be so different from other members of the family that he included them in a separate tribe, the Orophiina. Being a careful worker, he found several other characters which consistently distinguished the group, such as the conical, projecting, and subcontiguous procoxae, short pro- sternum, deep antennal sulci, and subequal 3rd and 4th antennal segments. In the same paper, Thomson noted the close similarity between Ennearthron affine (Gyllenhal) and Cis fronticornis (Panzer), and in spite of the difference in antennal segmenta- tion, he placed them both in Entypus. Mellie 's name Ennear- thron could not be applied, because a very different species, Cis cornutns Gyllenhal, had been designated as its type by Demarest (1860). Kiesenwetter (1877) included Entypus as a distinct subgenus of Cis and added to it a third species, Ennearthron ivagai Wankowicz, which had been described as having spinulose protibial apices. He considered the Orophiina to be a separate family, the Orophyidae, to which he added Xylographus Mellie. A more extreme view was taken by Seidlitz (1872), who placed these genera in a tribe of the family Anobiidae. In spite of the efforts of Thomson and Kiesenwetter, most European entomologists chose to ignore these two concepts and to rearrange the above species according to more superficial char- acters, such as the number and relative lengths of antennal seg- ments. Thus, Jacquelin Du Val (1861) placed C fronticornis in the genus Rhopalodontus in spite of its totally different pro- coxal structure, and returned affinis to Ennearthron, because of its 9-segmented antennae. In 1915, Peyerimhoff placed Cis hi- cornis Mellie in Rhopalodontus on the basis of antennal and pro- tibial characters only, and Nobuchi (1960a) described 2 more species, R. japonicus and R. tokunagai, which also appear to be in the fronticornis group. Until quite recently, then, the genus Rhopalodontus contained two distinct and distantly related groups of species : those resembling Ennearthron affine and be- longing properly to Entypus of Redtenbacher and Thomson, and those resembling R. perforatus. The situation was clarified by Lohse (1964), who had restudied the European species carefully enough to recognize the true LAWRENCE: CLASSIFICATION OF CIIDAE 277 relationships. By reviving the concept of Redtenbacher and Tliomson and uniting E. affinc, E. fronticornis, and R. hicornis into a single genus, he not only called attention to the apparent affinities of these three species, but also eliminated discordant elements from both Ennearthron and Rhopalodontus. In addi- tion to the above, Lohse added Cis bidentulus Rosenhauer to the group. Because of the variation in antennal segments, he pro- posed the subgenus Entypocis (with C. bidentulus as its type) for those species with 10-segmented antennae, leaving Entypus affinis the only member of the nominate subgenus. Ennearthron wagai Wankowicz (1869) (mis-cited in Lohse as E. wagai Wanka) was placed in a new genus, Wagaicis, because of its very narrow and laminate prosternal process. I agree with Lohse 's revival of Entypus and his inclusion of the 4 European species. In my own studies of North American Ciidae, I have encountered two more species, Cis curtidus Casey and Sulcacis lengi Dury, which also belong to this group. I have compared these with specimens determined as and fitting the descriptions of E. fronticornis, E. affinis, and E. bidentulus. C. curtulus is very similar to E. bidentulus, w^hile S. lengi is unique in having 9-segmented antennae, a somewhat shortened proster- num, and no sexual modifications on the head of the male. Since lengi is here selected as the type of Stdcacis Dury (1917), this genus becomes a junior synonym of Entypus. I would tentatively add R. japonicus and R. tokunagai, both from northern -Japan, to this genus, on the basis of descriptions and illustrations. I do not think the proposed division into subgenera is necessary, especially if it is based on the number of antennal segments. There is one final matter that appears to have been overlooked by all of the w^orkers in the Ciidae. This is the fact that the name Entypus was not available when used by Redtenbacher, having already been applied to a pompilid wasp (Hjanenoptera) by Dahlbom (1843). Entypus is a primary junior homonym and must be replaced. Of the two remaining available names which have been applied to the group, Sulcacis Dury is the oldest and is here considered as a replacement name. Although Sulcacis lengi Dury is the most atypical member of the group, it becomes the type species because of the priority of the generic name. I do not think that S. lengi is distinct enough to be placed in a separate genus ; if this species were to be removed from the group, however, then Lohse 's Entypocis is the next oldest available generic name and E. bidentulus would be considered the type. 278 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY A synonymy and redefinition of Sulcacis is given below. SULCACIS Dury Sulcacis Dury, 1917, Jour. Cincinnati Soc. Nat. Hist., 22(2) :20; Leng, 1920:247. Type, by present designation, Sulcacis lengi Dury, 1917:21. Eniypus Eedtenbaelier, 1847, Faun. Austr., (3):350 {non Dahlbom, 1843:35); Thomson, 1863:193; Sahlberg, 1926:78; Lohse, 1964:118. Type, by monotypy, Cis affinis Gyllenhal, 1827:628 (misidentified as Apate froniicornis Panzer). Cis {Eniypus), Kiesenwetter, 1877:190; Eeitter, 1878:21. Entypus (Eniypocis) Lohse, 1964, Ent. Blatter, 60(2) :121. Type, by orig- inal designation, Cis hideyitiilus Eosenhauer, 1847:58. Cis (in part), Gyllenhal, 1827:624; Mellie, 1847:109; Mellie, 1848:236 Lacordaire, 1857:551; Jacquelin Du Val, 1861:237; Seidlitz, 1872:44 Abeille de Perrin, 1874:19; Eeitter, 1878:27; Seidlitz, 1891:282 Casey, 1898:78; Sehilsky, 1900 :37E; Eeitter, 1902:47; Dalla Torre, 1911:5; Dury, 1917:5. Ennearthron (in part), Mellie, 1848:360; Lacordaire, 1857:552; Jacquelin Du Val, 1861:238; Abeille de Perrin, 1874:80; Eeitter, 1878:30; Sehilsky, 1900 :37B; Eeitter, 1902:59; Dalla Torre, 1911:23; Nobuchi, 1960a: 41. Ehopalodontus (in part), Jacquelin Du Val, 1861:238; Abeille de Perrin, 1874:76; Eeitter, 1878:30; Sehilsky, 1900 :37D; Eeitter, 1902:57; Dalla Torre, 1911:21; Peyerimhoff, 1915:26; Nobuchi, 1960a: 39. Form oblong to elongate, strongly convex and cylindrical ; ves- titure short, usually dual, consisting of both erect and inclined bristles. Head moderately declined, partly covered by pronotum ; frontoclypeal ridge of male usually with 2 small teeth or tuber- cles ; antennal fossa relatively deep. Antenna 9- or 10-segmented, with 3-segmented club; segments III and IV subequal, or III longer than next 2 combined ; maxillary palp narrow and elongate. Pronotum somewhat constricted anteriorly, narrowly margined laterally; anterior edge usually simple in both sexes. Elytra with relatively coarse, indistinctly dual, uniform or seriate punctation. Prosternum slightly tumid, concave laterally, slightly shorter than iutercoxal process which is narrow but not laminate. Protibia expanded at apex, outer apical angle forming a rounded process bearing several small spines. Metasternum convex, the suture short. Meso- and metatibiae slightly expanded and spinulose at apex. Sternite III of male with median pubes- cent fovea. A small genus with species throughout the Holarctic region. Intermediate in its characters between Cis and Eridoulus, on the one hand, and Ceracis, Wacjaicis, and Malacocis on the other, dif- fering from the former by the spinulose protibial apices and from LAWRENCE : CLASSIFICATION OF CIIDAE 279 the latter by the broader intercoxal process. The species which have been studied biolo HaDRAULE Thomson Hadraule Thomson, 1863, Skand. Col., 5:182; Sahlberg, 1926:71; Lohse, 1964:119. Type, by monotypy, Cis elongaiulus Gyllenhal, 1827:627. Cis {Hadraule), Seidlitz, 1872:44; Kiesenwetter, 1877:188; Reitter, 1878:24; Seidlitz, 1891:281; Sehilsky, ]900:37F (in part); Reitter, 1902:47 (in part); Dalla Torre, 1911:5; Jacobson, 1915:953 (in part). Tityocis Peyerimhoff, 1918, Bull. Soc. Ent. France, 1918:141. Type, by monotypy, Pityocis coarctatus Peyerimhoff, 1918:142 ( = Cis elongatulus Gyllenhal). [See Peyerimhoff, 1933.] Ennearthron {Knahlia) Roubal, 1936, Festsehr. 60 Geburtst. Embrik Strand, 1:53. Type, by monotypy, Cis elongatulus Gyllenhal, 1827:627. Hadraida Leng, 1920:246, incorrect subsequent spelling. Eridaulus Thomson Eridaulus Thomson, 1863, Skand. Col., 5:191; Seidlitz, 1872:45; Sahlberg, 1926:79. Originally included species: Anohium nitidum Fabricius, 1792:238 and Cis jacquemarti Mellie, 1848:328. Type not designated. Cis {Eridaulus), Kiesenwetter, 1877:191; Reitter, 1878:21, 24; Seidlitz, 1891:281; Sehilsky, 1900 :37F; Reitter, 1902:48; Dalla Torre, 1911:5; Roubal, 1912:29; Roubal, 1937:39; Chujo, 1940:132; Miyatake, 1954:49; Nakane and Nobuchi, 1955:49; Nobuchi, 1955:56; Nobuchi, 1960b :65. Xestocis Casey, 1898, Jour. New York Ent. Soc, 6(2) :85; Dalla Torre, 1911:20; Dury, 1917:15 (in part); Leng, 1920:247; Hatch, 1962:231. Type, by present designation, Xestocis Icvettei Casey, 1898:85. NEW SYNONYMY. Anohium, Ptinus, Cis auctt. Form oblong, strongly convex ; vestiture of short, fine hairs, longer recumbent hairs, or short, stout setae. Head moderately declined, partly covered by pronotum ; f rontoclj'peal ridge of male produced on each side forming 2 flattened plates which are rounded to triangular ; genal ridge strongly elevated and carinate, forming relatively deep antennal fossa. Antennae 10- segmented, with a 3-segmented club. Pronotum strongly convex, variously margined laterally, anterior angles rounded to pro- duced and acute ; anterior edge in male simple or produced and cmarginate. Elytra with dual punctation, consisting of larger nude macropunctures, and smaller setiferous micropunctures, the punctures uniform or seriate. Presternum relatively short, strongly tumid, concave laterally and carinate mesially ; inter- coxal process relatively narrow but not laminate, subacute at LAWRENCE : CLASSIFICATION OF CIIDAE 283 apex. Protibia expanded at apex, outer apical angle produced and dentate. Sternite III of male with median pubescent fovea or patch. A large genus with a primarily Holarctic distribution, but extending south at least in the Pacific region. Closely related to Cis and Strigocis, but distinguished by the strongly carinate pro- sternum, dual elytral punctation, and produced and dentate pro- tibial apices. The genus differs from Hadraule, Orthocis, and Dulichocis in general body form and by the dentate protibial apices, as well as by the carinate prosternum. Most of the species in this genus feed on the larger woody fungi, such as Fomes and Ganoderma. 3. ORTHOCIS, MELLIEICIS, AND DOLICHOCIS Casey (1898) proposed the genus Orthocis for the two Ameri- can species, Cis puncfatus Mellie and Orthocis aterrima Casey, which were distinguished from the species of Cis by the "... more parallel form of the body . . . glabrous surface, margined elytral suture, and . . . simple apex of the anterior tibiae." Dury (1917) described a third species, 0. longida, from the east- ern U.S. and Kraus (1908) added two species, 0. huesanus and 0. pidcher, from Florida. Another species, 0. platcnsis Brethes (1922), was described from Argentina. In the European literature, it has long been recognized that Cis alni Gyllenhal, and several related forms, such as C. perrisi Abeille de Perrin and C. coluhcr Abeille de Perrin, form a distinct group, characterized by the elongate body form, short pubescence, simple protibial apices, and lack of distinct sexual modifications on the head or prothorax of the male. Lohse (1964) proposed for these species the subgenus Mcllieicis and selected Cis alni as its type. Although the exact limits of the subgenus were not given in Lohse 's paper, it was indicated that all species of Cis with rounded or truncate protibial apices should be included. Having studied the European literature and ex- amined specimens in the American collections, I have come to the conclusion that both Cis punctatns Mellie sensu Casey and Orthocis aterrima Casey are very closely related to the Palearctic species in the Cis alni group. The name Orthocis, then, should apply to this group of species, and McUicicis should be considered a junior synonym. In 1908, Kraus described 3 species of Ennearthron, E. annida- tum, E. transversatum, and E. pallidum, from the southeastern 284 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY United States and Cuba, which differ from the species of Orthocis only in the possession of 9-segmented antennae. I think it is only sensible to expand the definition of Orthocis to include these species as well, instead of leaving them in a genus which appears to be a conglomerate of unrelated forms. If the limits of the group are thus expanded, it becomes necessary to consider several other species which might be included. In 1919, Dury described the genus Dolichocis with the single species D. manitoha Dury, the main characteristics being the 9-segmented antennae, narrow, cylindrical form, vestiture of short bristles, and rounded pro- tibial apices. The species differs from the American species of Orthocis, not only in its antennal segmentation, but by its stouter pubescence, anteriorly constricted prothorax, and the presence of 2 clypeal tubercles and a pubescent fovea on the head of the male. Hatch (1962) described a second species, D. incUstinctus, from western North America, which is very similar to and obviously congeneric with E7inearthron laricinmn (Mel- lie) of the European fauna. Other Palearctic species which might be in the same group are Ennearthron yuasai Chujo, E. pruino- siilum (Perris), and E. poriae Nakane and Nobuchi, the last two of which resemble Orthocis more than Dolichocis in the more quadrate prothorax with broader lateral margins and in the absence of sexual modifications on the head of the male. The problem is made more complex by the presence of 3 Palearctic species, Cis festivus (Panzer), Cis pygmaeus (Marsham), and Cis rhododactylus (Marsham), and one North American species, Cis angustus Hatch, which share certain character- istics with both Cis and Orthocis. Nyholni (1953) noted that the first three all have truncate or rounded protibial apices, and I have observed the same character in C. angustus. Yet all of these have sexual ornaments on the head of the male, and they differ in general appearance from typical Ortho- cis. The setting of generic limits, if this is at all justified, presents a difficult problem, which can be solved only after a more detailed investigation of all species involved. In the pres- ent discussion, I will make a few suggestions based on a rather superficial study of a large number of forms from various parts of the world and a more intensive study of American representatives. In the North American fauna, both Orthocis and Dolichocis seem to be well defined both morphologically and biologically. All of the species of Orthocis are elongate and parallel, with a LAWRENCE: CLASSIFICATION OF CIIDAE 285 vestiture of very short and fine hairs, a fairly smooth and shining surface, margined elytral suture, flat or slightly tumid prosternum, rounded protibial apices, and an absence of sexual ornaments on the head of the male. There is a tendency in some species (0. longulus Dury) towards an extreme attenuation of the body, and several subtropical forms are bicolored and may have 9-segmented antennae. Field observations indicate that at least some species occur under bark, in decaying branches and vines, and apparently not in the tougher fruiting bodies of most Polyporaceae. The genus is more common in the southern part of the continent, and a number of unnamed species have been seen from Central and South America. The two species of Boli- chocis, on the other hand, dift'er in having the prothorax con- stricted anteriorly, the vestiture consisting of short, stouter bristles, and the head modified in the male. Both species occur in the northern part of the continent, and, like most Eridaulus, they occur mainly on the fruiting bodies of woody fungi. In the European fauna, generic distinctions are not so easily made. Cis coluber Abeille de Perrin, C. reflexicollis Abeille de Perrin, and C. juglandis Reitter appear to be closely related to 0. alni, but they all have a stouter pubescence. Members of the Cis festivus group differ both in vestiture and in the presence of clypeal tubercles in the male. Of the species with 9-segmented antennae, E. laricinum undoubtedly belongs to Dolichocis, but E. priiinosulum appears to be closer to Orthocis. One character which has been overlooked by most European workers is the margined elytral suture in species of Orthocis. At the apex of each elytron, there is a distinct raised margin which curves laterad, leaving a narrow flat area between it and the edge of the suture. I have observed this in 0. alni, as well as in the American species. This character, in combination with others mentioned above, might serve to distinguish consistently the species of Orthocis from those of Cis and Dolichocis. Several other features which I have observed only in the North American Orthocis are the elongate prementum, elongate maxillary stipes with a flattened laterally placed laeinia, and the deeply emar- ginate, mesially unpigmented, 8th sternite of the male. An in- vestigation of these and other more cryptic characters in the European species may shed some light on their true relation- ships. Orthocis, like Eridaulus, is so similar to Cis that the question arises whether a generic distinction is justified or not. I think 286 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY that the same criteria can be used here as were applied in the case of Eridaulus. The size of the group is quite large, especially in view of the number of unnamed species which I have seen, and the distribution is probably cosmopolitan, with the majority of species occurring in tropical and subtropical regions. Several unique trends occur within the genus, and it appears that the group may have given rise to some of the peculiar endemic forms on the Hawaiian Islands and in other Pacific areas. The latter resemble GrfJiocis in the minute vestiture, flattened prosternum, simple protibial apices, and lack of male ornaments, but they vary considerably in body form, sculpture, and coloration. The trend towards extreme attenuation has already been mentioned. The biological characteristics of the genus, although briefly mentioned above, deserve further comment. Most of the species of Ciidae feed on the sporophores of wood-rotting fungi, espe- cially the more fibrous or woody fruiting bodies of the Poly- poraeeae and Hydnaceae. These species have evolved various morphological and physiological adaptations to enable them to cope with a substrate which not only presents a physical barrier to feeding but usually contains little moisture and a high per- centage of pure chitin, unavailable to insects not possessing a symbiotic intestinal flora. In addition, there has been a tendency for various species to specialize in certain groups of host fungi (Paviour-Smith, 1960). The species of Ortliocis, whose habits have been recorded, appear to be general fungus feeders, oc- curring under bark, in dead branches and vines, in more humid situations, and in association with mycelia and fruiting bodies of a wide variety of wood-rotting fungi. The fungus sporophores are usually of a softer and more ephemeral type and their loca- tions permit a certain amount of bacterial and fungal decompo- sition, so that the resulting substrate is more easily utilized by the beetles. Beniek (1952) has reported Cis alni from such diverse fungi as Exidia glandulosa (Tremellaceae), Stereum rugosuni (Thelephoraceae), and Auricularia auricula-judae (Auriculariaeeae), while Lucas (1849) found the same species (as C. puncUdatus Lucas, non Gyllenhal) on Schizophyllum commune (Agaricaecae). Blatchley (1910, 1923) noted that 0. punctatus was taken by sifting debris from an oak log and that 0. pulchcr was found by beating dead branches of oak. Kraus (1908) bred Ennearthron fransvcrsatum from decaying rattan vines. Perris (1877) described the larva of Cis coluber from dead branches of chestnut and oak in which the fungus LAWRENCE: CLASSIFICATION OF CIIDAE 287 Thclcphora was oTowing; he also noted that the beetle could not be found on branches still remaining on the tree. In the same paper, he noted that the habits of C. alni and C. reflexicollis Abeille de Perrin were similar, and that C. ohlongus Mellie and C. pruinosulus were taken on fungus-infested elm branches. Zim- merman (1938, 1942) and Swezey (1954) have recorded a num- ber of Pacific island species from dead vines and branches, and those which I have examined seem to fall within this same group. One fact that makes the study of the particular group more difficult is that most of the shared characters are both simple and apparently primitive. Most of the special modifications which are characteristic of the family as a whole, such as the development of protibial structures for boring (expanded apices, teeth, combs, spines), various prosternal modifications (shorten- ing, carination, reduction of intercoxal process and increased coxal size), presence of sexual ornaments on the head and pro- notum of the male, and the development of more complex puncta- tion and vestiture, are all absent in the species of OrtJiocis. In addition, the generalized fungus-feeding habit may be considered primitive. Members of closely related families, such as the Lathridiidae and Corylophidae, share this habit with the species of Orthocis, and Crowson (1955) thinks that this may have been characteristic of the ancestral Cucujoidea. If these characters are primitive or plesiomorphic, they are, according to Hennig (1965), less reliable than derivative or apomorphic ones in deter- mining relationships. If, on the other hand, Orthocis represents a collection of two or more convergent groups, which have secondarily developed this type of feeding habit and thus have lost, through disuse, the structural modifications associated with boring, this convergence would probably be difficult to detect because of the simplicity of the resulting characters. I favor the hypothesis that these characters are primitive, rather than derivative, and that the feeding habits probably represent those of the common ancestor of the group. In certain other species, such as Hadraiile elongaiula and Maphoca hlaisdelli, which have similar biologies, this simplified condition also occurs, but other characters indicate that both of these are derivative forms, which may have evolved from a species in the Cis comptus group. I think that most of the species discussed above should be placed in a distinct genus, to which the name Orilwcis applies. The exact limits of the genus remain uncertain, especially with reference to the inclusion of Ch festivns and its relatives, and 288 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY it is hoped that Dr. Lohse will shed some light on the relation- ships of these European species. I would definitely exclude Cis angustus Hatch, which does have a truncate protibial apex, but which occurs on Fomes pinicola, has clypeal tubercles in the male, and lacks a margin on the elytral suture. The species of Dolichocis, although they share certain char- acters with Orthocis, should remain in a separate genus, differing by the absence of the margin on the elytral suture, the anteriorly constricted prothorax, cylindrical form, and vestiture of short, stout setae. The genus should include the species laricinus (Mel- lie), yuasai (Chujo), mayiitoha Dury and indistinctus Hatch. All four species occur on the larger, woody fruiting bodies of fungi, such as Fomcs pinicola and F. officinalis. Synonymies and redefinitions of Orthocis and DolicJiocis are given below. Orthocis Casey Orthocis Casey, 1898, Jour. New York Eiit. Soc, 6(2) :84; Kraus, 1908:77; Dalla Torre, 1911:20; Dury, 1917:13; Leng, 1920:247; Brethes, 1922:302. Type, by present designation, Orthocis aterrima Casey, 1898:84. Cis (Mellieicis) Lohse, 1964, Ent. Blatter, 60(2) :122. Type, by original designation, Cis alni Gyllenhal, 1813:386. NEW SYNONYMY. Enncarthron, Kraus, 1908:78. Cis (in part), auctt. Form elongate and somewhat depressed to narrowly elongate and cylindrical; vestiture of very short fine hairs or stouter bristles. Head slightly to moderately declined, only slightly cov- ered by pronotum; frontoclypeal ridge without distinct sexual modifications in male ; antennal fossa shallow. Antenna 9- or 10- segmented, with 3- segmented club, segments III and IV usually elongate ; maxillary stipes elongate, laeinia lateral, palp rela- tively stout ; prementum elongate. Pronotum subquadrate, sides narrowly margined to broadly margined and explanate, anterior angles truncate to slightly produced and rounded ; anterior edge simple in both sexes. Elytra usually parallel sided; punctation single and relatively uniformly distributed; suture margined posteriorly, the margin curved laterad just before apex. Pro- sternum flat to slightly tumid, longer than intercoxal process which is fairly broad ; procoxae subtransverse, narrowly open behind. Protibia only slightly expanded at apex, outer apical angle truncate or rounded. Metasternum slightly convex, the suture moderately long. Sternite III of male with a median pubescent fovea or patch. LAWRENCE: CLASSIFICATION OF CIIDAE 289 A large, widespread g:emis, but mainly tropicopolitan. Closely related to Cis and Dolichocis, from which it differs by the rounded protibial apices, margined elytral suture, and lack of distinct sexual ornaments in the male. The species appear to be general fungus feeders, occurring often on dead vines and branches penetrated by fungus mycelia. Dolichocis Dury Dolichocis Dury, 1919, Canad. Ent., 51:158; Hatch, 1962:234. Type, by monotypy, Dolichocis manitoha Dury, 1919:158. Cis (in part), Mellie, 1848:236; Lacordaire, 1857:551; Jaequelin Du Val, 1861:237; Seidlitz, 1872:44; Kiesenwetter, 1877:178. Ennearthron (in part), Abeille de Perrin, 1874:80; Eeitter, 1878:30; Seid- litz, 1891:285; Schilsky, 1900 :37B; Eeitter, 1902:59; Dalla Torre, 1911:23; Chujo, 1941:85. Form elongate, cylindrical; vestiture of short, stout, suberect bristles or squamae. Head moderately declined, partly covered by pronotum ; frontoclypeal ridge of male bituberculate, vertex simple or foveate ; antennal fossa shallow. Antenna 9-segmented, with 3-segmented club ; maxillary stipes subquadrate, lacinia subterminal, palp relatively stout. Pronotum almost as long as wide, somewhat constricted anteriorly, sides narrowly margined, anterior angles not or barely produced ; anterior edge simple in both sexes. Elytra elongate and subparallel ; punctation single and fairly uniformly distributed. Prosternum slightly tumid, slightly longer than intercoxal process which is fairly broad and blunt at apex ; procoxae subtransverse, narrowly open behind. Protibia only slightly expanded at apex, outer apical angle truncate or rounded. Metasternum slightly convex, suture mod- erately long. Sternite III of male with a median pubescent fovea. A small genus restricted to the Holarctic region. Closely re- lated to Cis and Orthocis, from which it differs in the rounded protibial apices, stout vestiture, constricted prothorax, and lack of a margin along the elytral suture. The 4 species all occur on Fomes pinicola and related fungi. REFERENCES Abeille de Perrin, Elzear 1874. Essai monographique sur les Cisides europeens et circamediter- raneens. Camoin, Marseille. 100 pp. 290 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Benick, L. 1952. Pilzkafer imd Kaferpilze. Acta Zool. Fennica, 70: l-l!50. Blatchley, William S. 1910. All illustrated descriptive catalogue of the Coleoptera or beetles (exclusive of the Ehynchophora) known to occur in Indiana. Bull. Indiana Dept. Geol. and Nat. Ees., 1: 1-1385. 1923. Notes on the Coleoptera of southern Florida with descriptions of new species. Canad. Ent., 55: 13-20, 30-36. Brethes, Juan 1922. Descripcion de varios coleopteros de Buenos Aires. Ann. Soc. Cient. Argent., 94: 263-305. Casey, Thomas L. 1898. Studies in the Ptinidae, Cioidae, and Sphindidae of America. Jour. NeAV York Ent. Soc, 6(2): 61-93. Chujo, Michio 1940. Descriptions of a new and a rare species of the Japanese ciid- beetles. Ins. Matsumurana, 14(4): 132-133. 1941. Description of a new ciid-beetle from Japan. Kontyu, 15(2) : 85-86. Ckowson, Eoy a. 1955. The natural classification of the families of Coleoptera. Lloyd, London, 187 pp. Dahlbom, Anders G. 1843. Hymenoptera Europae praecipue borealis . . . Tomus [1] : Sphex in sensu Linnaeano. Luiidlierg, Lund, xliv + 528 pp. Dalla Torre, K. W. von 1911. Cioidae. Pars 30. In W. Junk & S. Schenkling, edit., Coleop- terorum Catalogus. Junk, Berlin, 32 pp. Demarest, E. 1860. Tome 17. Coleopteres. Partie 3. In Jean C. Chenu, Encyclo- pedie d'histoire naturelle ou traite complet de cette science. Didot, Paris, 360 pp. Dury, Charles 1917. Synopsis of the coleopterous family Cisidae (Cioidae) of America north of Mexico. Jour. Cincinnati Soc. Nat. Hist., 22(2): 1-27. 1919. A new ciside genus with new species from Manitoba. Canad. Ent., 51 : 158. Fabricius, Johann 1792. Entomologia systematica emendata et aucta. Tomus I [Pars 1]. Prost, Copenhagen, xx + 330 pp. Gaubil, J. 1849. Catalogue synonymique des Coleopteres d 'Europe et d 'Algerie. Maison, Paris, 296 pp. Gyllenhal, Leonardo 1813. Insecta Svecica. Classis I. Coleoptera sive Eleuterata. Tomus I, Pars III. Leverentz, Scaris, 730 pp. LAWRENCE: CLASSIFICATION OP ClIDAE 291 1827. Tiiseeta Sveeiea, Classis I. Coleoptera sive Eleuterata. Tomus I. Pars IV. Fleischer, Leipzig, 761 pp. Hatch, Melville H. 1962. The beetles of the Pacific Northwest. Part III: Pselaphidae and Diversicornia I. Univ. Washington Press, Seattle, ix + 503 pp. Hennig, Willi 1965. Phylogenetic systematics. Ann. Eev. Ent., 10: 97-116. Jacobson, G. G. 1915. Zhuki Eossii i zapadnoi Evropi. Euskovodstvo k opredeleniyu zhukovi. Vip. XI. Devrient, Leningrad, pp. 865-1024. Jacqublin Du Val, Pierre N. C. 1861. Manuel entomologique. Genera des Coleopteres d 'Europe. Tome 3 [part]. DeyroUe, Paris, pp. 236-240, cat. p. 168. KlESENWETTER, H. VON 1877. Anobiidae, Cioidae, Aspidophoridae. In W. F. Erichson, Na- turgeschichte der Insecten Deutschlands. Abtheilung 1. Coleop- tera. Band 5, Hiilfte 1. Moeser, Berlin, pp. 169-200. Kraus, E. J. 1908. New bieolored Cioidae (Coleoptera). Proe. Ent. Soc. Washing- ton, 10: 74-81. Lacordaire, Jean T. 1857. Histoire naturelle des Inseetes. Genera des Coleopteres. Tome 4. Libr. Encycl. Eoret, Paris, 579 pp. Leng, Charles W. 1920. Catalogue of the Coleoptera of America, north of Mexico. Sherman, Mt. Vernon, N. Y., x + 470 pp. Leng, Charles W. and A. J. Mutchlek 1927. Supplement 1919 to 1924 (inclusive) to Catalogue of the Coleoptera of America, north of Mexico. Sherman, Mt. Vernon, N. Y., 78 pp. Lohse, Gustav A. 1964. Die in Mitteleuropa vertretenen Gattungen der Cisidae (1. Beitrag zur Kenntniss der mitteleuropaischen Cisidae). Ent. Blatter, 60(2): 116-122. Lucas, H. 1849. Exploration scientifique de I'Algerie, Zoologie. Histoire naturelle des animaux articules. 2 Partie. Inseetes. Imprim. Nat., Paris, 590 pp. Mayr, Ernst, E. G. Linsley and E. L. Usinger 1953. Methods and principles of systematic zoology. McGraw-Hill, New York, vii + 328 pp. Mellie, J. 1847. Melanges et nouvelles. Eev. Zool. Cuvierienne, 10: 108-110. 1848. Monographie de 1 'ancien genre Cis des auteurs. Ann. Soc. Ent. France, Ser. 2, 6: 205-274, 313-396, pis. 9-12. 292 BITLLETIN : MUSEUM OF COMPARATIVE ZOOLOGY MiYATAKE, MUTSUO 1954. studies on the Japanese Ciidae, I. (Coleoptera). Sei. Eep. Matsuyama Agr. Coll., 14: 40-67, 11 pis. Nakane, Takehiko and Akira Nobuchi 1955. On a new genus and six new species of ciid-beetles from Japan (Ciidae, Coleoptera). Sci. Rep. Saikyo Univ., Ser. A, 2(1): 47-52. Nobuchi, Akira 1955. Studies on the ciid-beetles from Japan. (I) With descriptions of a new genus and some new species (Ciidae, Coleoptera). Ent. Rev. Japan, 6(7) : 53-58, pi. 12. 1960a. Ciidae from Hokkaido (Coleoptera). Ent. Rev. Japan, 11(2): 37-41, 1 pi. 1960b. On a new species of Cis Latreille from Japan (Coleoptera, Ciidae). Akitu, 9: 65-66. Nyholm, Tord 1953. Bemerkungen iiber einige schwedische Cis-Artcn (Col., Ciidae). Ent. Tidskr., 74: 187-196. Paviour-Smith, Kitty 1960. The fruiting-bodies of macrofungi as habitats for beetles of the family Ciidae (Coleoptera). Oikos, 11(1): 43-71, 1 pi. Perris, Edouard 1877. Larves de Coleopteres [part]. Ann. Soc. Linn. Lyon, N.S., 23: 1-43, 14 pis. Peyerimhoff, p. de 1915. Notes sur la biologie de quelques Coleopteres phytophages du Nord Africain (Deuxieme serie). Ann. Soc. Ent. France, 84: 19-61. 1918. Nouveaux Coleopteres du Nord-Africain (Vingt-huitieme note: Faune du Pin d'Alep). Bull. Soc. Ent. France, 1918: 141-143. 1933. Les Coleopteres attaches aux coniferes dans le nord de 1 'Af rique. Ann. Soc. Ent. France, 102: 359-408. Redtenbacher, Ludwig 1847. Fauna Austriaca. Die Kafer [Part 3]. C. Ceroid, Vienna, pp. 321-480. Reitter, Edmund 1878. Nachtrage und Erganzungen zur Bearbeitung der Cioiden von H. V. Kiesenwetter. Deutsch. Ent. Zeitschr., 22(1): 21-30. 1902. Analytische Uebersicht dcr palearctischen Gattungen und Artcn der Coleopteren-Familien: Byrrhidae (Anobiidae) und Cioidae. Verhandl. Naturforsch. Ver. Briinn, 40: 1-61. Rosenhauer, Wilhelm G. 1847. Beitriige zur Insekten-Fauna Europas. Erstes Bandchen. Blaes- ing, Erlangen, x + 159 pp. RouBAL, Jan 1912. Koleopterologicke vysledky me cesty na Kavkaz v cervenci r. 1910. Casopis, 9: 25-28,124-26. LAWRENCE: CLASSIFICATION OF CIIDAE 293 1936. Besehreibung einer neuen Untergattung, einer Varietat und einer Form der Coleopteren. Festschr. 60 Geburtst. Embrik Strand, 1 : 53-58. 1937. Koleopterologische Notizen. Ent. Nachrichtenblatt, 11: 38-40. Saalas, Uunio 1923. Die Fichteiikafer Filmlands. II. Spezieller Teil 2 und Larven- bestimmungstab. Suom. Tiedeakat. Toimituskia, Sarja A, 22 (1): 1-746, 28 pis. Sahlberg, John 1926. Enumeratio Coleopteroruni Fenniae. Lamellicornes, Platysoma, Xylophagi, Fungicola. Ann. Soe. Zool.-Bot. Fennicae, 4(1): 1-169. SCHILSKY, J. 1900. Heft 37 in H. C. Kuster, G. Kraatz, and J. Schilsky, Die Kiifer Europas. Bauer & Raspe, Niirnberg, pp. i-iv, 1-100, A-W. Seidlitz, Georg 1872. Fauna Baltica. Die Kafer (Coleoptera) der Ostsee-Provinzen Russlands. Laakmann, Dorpat, xlii + 142 pp. 1891. Fauna Transylvanica. Die Kafer (Coleoptera) Siebenbiirgens. Hartung, Kaliningrad, Ivi + 192 + 914 pp. SwEZEY, Otto 1954. Forest entomology in Hawaii. Spec. Publ. Bernice P. Bishop Mus., No. 44, ix + 266 pp. Thomson, C. G. 1863. Skandinaviens Coleoptera, synoptiskt bearbetade. Tome 5. Lundberg, Lund, 340 pp. Wankowicz, Jean 1869. Notices entomologiques. Ann. Soe. Ent. France, Ser. 4, 9: 411- 422. Zimmerman, Elwood C. 1938. Ciidae of southeastern Polynesia (Coleoptera). Occ. Pap. Bernice P. Bishop Mus., 14: 199-217. 1942. Ciidae of Guam. In Insects of Guam — I. Bull. Bernice P. Bishop Mus., 172: 47-52. (Received 5 March 1965.) Bulletin of the Museum of Comparative Zoology HAEVARD UNIVERSITY Vol. 133, No. G THE FOSSIL ELEPHANT SHREWS (FAMILY MACROSCELIDIDAE) By Bryan Patterson With One Plate CAMBRIDGE, MASS., U.S.A. PRINTED FOR THE MUSEUM November 26, 1965 Publications Issued by or in Connection WITH THE MUSEUM OF COMPARATIVE ZOOLOGY HARVARD UNIVERSITY Bulletin (octavo) 1863 — The current volume is Vol. 133. Breviora (octavo) 1952 — No. 23-1: is current. Memoirs (quarto) 18G4-1D38 — Publication terminated with Vol. 55. Johnsonia (quarto) 1941 — A publication of the Department of JMollusks. Vol. 4, no. 43 is current. Occasional Papers of the Department of Mollusks (octavo) 1945 — Vol. 2, no. 31 is current. Proceedings of the Nevv^ England Zoological Club (octavo) 1899-1948 — Published in connection with the Museum. Publication terminated with Vol. 24. The continuing publications are issued at irregular intervals in num- bers which may be purchased separately. Prices and lists may be obtained from the Publications Office of the Museum of Comparative Zoology, Cambridge 38, Massachusetts. Peters' Check-List of Birds of the "World. Volume 1 is out of j)rint; volumes 2 to 7, 9, 10 and 15 are currently available. Phylogeny and Evolution of Crustacea. Proceedings of a confer- ence held at Cambridge, Mass., March 6-8, 1962. H. B. AVhittington and W. D. I. Rolfe, editors. Cambridge, Mass., 192 pp., 80 figs., 1963. Cloth bound, $6.75. Fishes of the Gulf of Maine, by Henry B. Bigelow and AVilliam C. Schroeder. AVashington, viii - 577 pp., 1953. Photo-offset reprint, $6.50. Publications of the Boston Society of Natural History The remaining stock of the scientific periodicals of the Bostoii Society of Natural History has been transferred to the Museum of Comparative Zoology for distribution. Prices for individual numbers may be had upon request. Bulletin of the Museum of Comparative Zoology HARVARD UNIVERSITY \'0L. 133, Xo. 6 THE FOSSIL ELEPHANT SHREWS (FAMILY MACROSCELIDIDAE) By Bryan Patterson With One Plate CAMBRIDGE, MASS., U.S.A. PRINTED FOR THE MUSEUM November, 1005 Bull. Mus. Comp. Zool., Harvard Univ., 133(6): 295-335, November, 1965. No. 6 — The Fossil Elephant Shrews (Family Macroscelididae) By Bryan Patterson INTRODUCTION The history of this peculiar group of mammals was osteusibly a blank until 1937, when Broom described an extinct species of Elephanfulus from the Pleistocene of South Africa. Four extinct genera, however, had been described earlier, from 1910 on, but, having been placed incorrectly in other groups of mammals, had gone unrecognized. One of these, Palacolhenfoides Stromer, has been detected as a macroscelidid by Butler and Hopwood (1957); the other three, Metoldobofes Schlosser, Myohyrax Andrews and Protypotheroides Stromer are here placed in the family for the first time. Palaeothentoides was originally described as a marsupial, Metoldobofes as a mixodectid insectivore, and Myohyrax and Protypotheroides as hyracoids. So extraordinary a situation is, I believe, without parallel in mammalian paleontology. This paper came into being in a rather roundabout way. Crea- tures with names such as Palaeothentoides and Protypotheroides have a certain fascination for anyone with a taste for South American fossil mammals. Being in Europe during the spring of 1957, I took the opportunity of visiting Rlunich and examining material of both forms. The collection included an important undescribed specimen of Palaeothentoides africanus, and this I was \Try kindly permitted to borrow for further study. Shortly after completing the description of it, and with the details of macro- scelidid dental structure fresh in mind, I had occasion to consult Schlosser 's memoir (1911) on Fayum mammals during a discussion of the dental formula of Parapithecus. There, staring up from the plate, was Metoldobotes, an ob\-ious macroscelidid. (Discussions of Parapithecus are perennial ; it is gratifying to be able to report this useful by-product of one of them.) On examining Protypotheroides beetzi in Munich I had noted a decided resemblance to Palaeo- thentoides, but at the time had considered it to be an interesting example of convergence, one about on a par with the resemblance to the interatherid typotheres. It was only after consulting Whit- worth's study (1954) of the Miocene hyracoids of Kenya in quite another connection that I began to suspect the Myohyracinae really were macroscelidids, a suspicion that hardened to conviction upon examination of Stromer's figures (1926) of postcranial remains from the Miocene of Southwest Africa. Here was a group 298 bulletin: museum of comparative zoology of elephant shrews that had masqueraded as ungulates for half a century. Broom's (1948) Mylomygale spier si from the Pleistocene, a form with hypsodont, rodent-like cheek teeth, revealed the former existence of yet another phylum of the family. It became clear that the surviving forms gi\'e little hint of a rather remark- able radiation that went on within the African continent through- out much of the Cenozoic. There is nothing remotely primate-like about the extinct phyla so far known. The fossil record in fact very definitely supports those who ha\'e been unable to accept the hypothesis of a close relationship between the elephant shrews and the trecshrews. As LcGros Clark has well put it (1959, pp. 318- 319n): "In retrospect it is difficult to understand this taxonomic association . . .the differences . . . are so marked as to make it clear that they are really quite divergent types." This study has been aided by National Science Foundation Grants G-3120 and GP 1188, which made it possible for me to examine specimens in Europe and in Kenya. For access to material I am indebted to the authorities of the British Museum (Natural History) and of The American Museum of Natural History, to Dr. Richai'd Dehm and to Dr. L. S. B. Leakey. Miss Margo Hayes has assembled the final manuscript and assisted in checking localities and references. The photographs are by Mr. Frank White and the drawings by Mrs. Dorothy Marsh. Figures of fossil macroscelidids are widely scattered in the literature and a number of them are in rather rare publications. Those pertinent to the work are accordingly redrawn here. Publication has been aided by National Science Foundation Grant GB-500. TAXONOMY AND MORPHOLOGY MACROSCELIDIDAE IMACROSCELIDINAE MetOLDOBOTES Schlosser Metoldobotes Schlosser, 1910, p. 507; Matthew, 1910, p. 702. Metolbodotes Schlosser, 1911, p. 70; Matthew, 1915, p. 467. Type species: M. stromeri Schlosser, 1910. Distribution: Early Oligocene, north Africa. Emended diagnosis: I3 (?), C, P4, M2. I3 not bifid, with pro- nounced vertical groove on lingual side; C bluntly pointed, three sided; Pi_3 short relative to P4 — M2; Pi single rooted, roots of P2 very closely appressed, P3 with small anterior and posterior cusps, without posterior accessory cusp; P4 and lower M with crista PATTERSON : FOSSIL ELEPHANT SHREWS 299 obliqua running to posterior face of trigonid, talonid of M3 short relative to trigonid. Horizontal ramus of mandible relatively short, deep; ascending ramus steep; symphysis long, extending to P3. ]\Ietoldobotes stromeri Schlosser (Fig. 1 c, d) Metoldobotes slronicri Schlosser, 1010, p. 507; Matthew, 1910, p. 702. Metolbodotes stromeri Schlosser, 1911, pp. 70-72, 147, 157, 163, 164; pi. 9, fig. 5. Type: An incomplete right horizontal ramus in the Stuttgart collections, with I3, C, P3 — M2, alveoli for I2, Pi, roots of P2. Hyyodigm: Type only. Horizon and locality: Fluviomarine series, early Oligocene; Fayum, Egypt. Diagnosis: As for the genus. The fragmentary type specimen appears to represent an animal somewhat larger than Rhynchocyon peter si and Protypotheroides beetzi, and hence the largest known member of the family. Discussion: Metoldobotes, with the exception of brief comments by Matthew and passing mentions in various editions of Zittel, has remained essentially unnoticed in the literature since its description. Schlosser tentatively assigned it to the JN'Iixodectidae, a reference which, as Matthew (1915, p. 467) stressed, had nothing to recommend it in the way of positive resemblances between the Fayum form and any mixodectid. The type ramus, incomplete anteriorly, preserves, in series, an anterior alveolus, an incisor, a partially erupted, conical tooth, the ah-eolus of a single-rooted tooth, the very closely appressed roots of a double-rooted tooth and the last four cheek teeth. Schlosser interpreted this array as Ii,2,3, C, P3,4, Mi,2,3, but he made no comparisons with any macroscelidid. Inspection of his figures at once reveals an impressive number of resemblances to the various members of this family, and suggests that the dental formula is in reaUty 1(0,2,3, C, Pi,2,3,4, Afi,2, as in Rhynchocyon, Petrodromus, Elephantulus, Macroscelides, and Mylomygale (an alveolus for Ii in this interpretation is lacking, but the specimen is incomplete anteriorly). In the ensuing remarks the teeth will be so designated. The crown of 1 3 is described by Schlosser as being about half the height of the root; this is true of the incisors of macroscelidines. A grooN'e is present on the lingual faces of the incisors of members of this subfamily, although in no case is it as pronounced as shown in Schlosser 's figure. The canine in Rhynchocyon is sometim3 5 (e.g. MCZ 38782), although by no means invariably, not fully 300 bulletin: museum of comparative zoology erupted until after the posterior cheek teeth have come into wear. Pi is single-rooted in Macroscelides, Mylomygale, and in some species of Elephant iilus^. The roots of P2 are closely appressed in Macroscelides. P3 of Mefoldohotes resembles that of all members of a Figure^ 1. r, d, Meioldobotes slronicri, ciowii view of (ifiititioii ;inil lateral view of iiKUitliWle, c X3, d x|; redrawn from Sclilosser. «, I'etrodroiiius nigriseia, MCZ 22434, 6, Rhi/nchoei/on petersi, MCZ 22573, ert)wn views of P3 - Ml, not to scale. ' Among the Elephantulus material in the Museum of Comparative Zoology collections I have found Pi to be single rooted in E. o-ularis (29 specimens), ruftsceris (•'>), and pulchtr (ti); variable in rupestris (3 single and 3 double); and double rooted in intufi (5). and ;i/sopes (2). PATTERSON: FOSSIL ELEPHANT SHREWS 301 the family in its proportions relative to P4 — M2 and in its posses- sion of low anterior and posterior cusps, and that of the macrosceH- dines in the absence of an accessory cusp on the posterior slope of the protoconid. The large P4 has a wide molariform talonid and a narrower, elongate trigonid. The tooth is typically macroscelidid in these features, and also as regards size relative to the molars and the structure and arrangement of the cusps and crests. The molars of Metoldobofes resemble those of the living macroscelidids in the nearly equal heights of the trigonids and talonids, the metaconids and entoconids higher than the protoconids and hypoconid, the absence of labial and lingual cingula and the struc- ture and positions of cusps and crests generally. The paraconid is median, as in the macrosceUdines. The crista obhqua runs to the center of the posterior face of the trigonid, as in the Miocene Rhynchocyon clarki (Butler and Hopwood, 1957, p. 10), rather than to the metaconid, as is the case in living forms. The relative lengths of Mi and M2 are approximately as in Rhynchocyon, and the talonid of AI2 appears to be about as small and short relative to the trigonid as in R. cirnei (e.g. MCZ 43735). In agreement with Rhynchocyon and the macroscelidines, the mental foramina are beneath Pi_2 and P4; the ascending ramus, to judge from Schlosser's figure, arises abruptly well behind the last molar. The masseteric fossa is shallow and, as in macroscelidines, extends down to the level of the tooth row. The horizontal ramus is deeper than in either the Rhynchocyoninae or the Macroscelidinae, shallower than in the hypsodont Myohyracinae. That part of the tooth row anterior to P4 is somewhat shorter relative to the length of the series as a whole than in living members of either of the first two subfamilies, but is approximately comparable to Myohyrax and Mylomygale in this respect. The symphysis, fide Schlosser, extends to P3 (his P4) and is hence longer than in all other known members of the family, in which it terminates beneath C or Pi. On the evidence available it is difficult to assign Metoldobotes to subfamily with any confidence. The Fayum form does not, even incipiently, display any of the specializations of myohyracines or mylomygalines. It does resemble both the Rhynchocyoninae and the Macroscelidinae, agreeing with one or the other now in this character, now in that, and differing from both of them in the long symphysis. Resemblances to Rhynchocyon — and differences from the macroscelidines — are the small size of the talonid of J\l2 rela- tive to the trigonid and the small size of M2 as a whole relative to Ml. M. stromeri resembles the Miocene R. clarki in that the crista obliqua runs only to the posterior face of the trigonid and not to the 302 bulletin: museum of comparative zoology apex of the metaconid, but this is interpretable simply as a primi- tive character possessed in common. Resemblances to the ]\Iacro- sceUdinae, or at least to some of them, are the lack of a posterior accessory cusp on P3, the small single-rooted Pi, the closely appressed roots of P2, the median position of the paraconid in the molars, the lingual groove of the incisor, and the steeply rising ascending ramus. In sum, the characters suggest relationship with the macroscelidines rather than with the rhynchocyonines, and I very tentatively place Metoldobotes in the Macroscelidinae. PalAEOTHENTOIDES Stromer Palaeothentoides Stromer, 1932, p. 185. Type species: P. africanus Stromer, 1932. Distribution: Early Pleistocene?, southwest Africa. Emended diagnosis: Lower postcanine formula P4, M3; Pi two- rooted, not incisiform; P2_3 with anterior cusps little separated from protoconids, P3 without metaconid and entoconid rudiments; P4 narrow, metaconid decidedly posterointernal to protoconid, reentrant valley between metaconid and entoconid nearly filled by swelling on crest running anteroexternally from entoconid, para- conid crest high, anterointernal sAvelling partially obhterating cleft between anterior crest and metaconid; Mi_2 with very slight, shallow clefts between paraconids and metaconids, sides of deep reentrants between metaconids and entoconids parallel, not ventrally converging; horizontal ramus of nearly even depth beneath cheek teeth, shghtly downcurving anteriorly. Palaeothentoides africanus Stromer (Fig. 2; PL 1) Palaeothentoides africanus Stromer, 1932, pp. 178-185, figs. la-2b; Butler and Hopwood, 1957, p. 11. Type: Munchen No. 1931. VII. la, left ramus with P3 - M2, alveoli for Pi_2, M3. Hypodigm: Type, and Munchen Nos. 1931. VII. lb, fragment of left ramus with M2-3 (now lost), and 1932. I. 501, left ramus with Pi - M3. Horizon: The "intermediate terrace" of Wagner and Merensky (1929, p. 29, fig. 5); age uncertain, possibly early Pleistocene. Stromer (1931, p. 41; 1932, p. 185) considered this to be ''wohl Mittelpliocan," which would now, with the transfer of the Villa- franchian to the Pleistocene, be regarded as late Phocene. The two other forms definitely identified by Stromer from this level, PATTERSON : FOSSIL ELEPHANT SHREWS 303 Enhydriodon and Hijaena (especially the latter), are not incon- sistent with a Pleistocene age. Locality: Klein Zee (or Kleinsee), near the mouth of the Buffels River on the coast of Little Namaqualand, some 25 miles SSE of Port Nolloth, Union of South Africa. Diagnosis: As for the genus. Comparable in size to the smaller hving macroscelidines. Description: The first premolar is a long, narrow, double-rooted tooth with a simple crown consisting of a procumbent protoconid, which extends forward beyond the anterior root, connected by a crest to a smaller posterior cusp. The lingual face of the tooth is very slightly convex, the labial vertically grooved between the cusps. P2 bears a small cusp on the anterior slope of the erect protoconid; labial and lingual grooves are present anterior to the protoconid, and the labial groove between protoconid and posterior cusp is much larger and deeper than in Pi. All these features are accentuated in P3 : the anterior cusp is larger, the grooves deeper, and the posterior cusp larger and wider; there is no metaconid or entoconid rudiment. These three teeth progressively increase in length and height. P1-3 of Palaeothentoides are very similar to the corresponding teeth of Nasilio, differing from those of the other living genera in various particulars. Thus, in Macroscelides, Pi is single-rooted and similar in structure to I2 — C, while P2-3 are higher crowned relative to length ; in Elephantulus, Pi has a higher protoconid, P2-3 have the anterior cusps well separated from the protoconid, P3 has metaconid and entoconid rudiments, and a rudiment of the metaconid is occasionally seen on P2; in Petro- dromus and Rhynchocyon, all three teeth are higher, more piercing (especially the caniniform Pi of the latter), anterior cusps are either lacking entirely (Rhynchocyon) or rudimentary {Petro- dromus), while the posterior cusps are very small in both and, in Rhynchocyon, confined to P3. P4, the longest of the cheek teeth, is submolariform, the talonid completely as in Mi_2, the narrow trigonid not. The entoconid is fully as large as in the molars and, as in them, higher than the hypoconid. In the trigonid, the anterior cusp is set off labially from the protoconid by a vertical groove wider and shallower than the corresponding ones on the anterior premolars. The groove be- tween these two cusps on the lingual side is shallow and partially filled by a buttress on the side of the ridge connecting them. The metaconid is well developed, nearly as high as the protoconid and decidedly posterointernal to it in position ; the short crest between the cusps bears a small, shallow groove on its lingual side. The 304 bulletin: museum of comparative zoology crista obliqua runs to the apex of the meta- conid. The labial reentrant between trigonid and talonid is fully as large and deep as in Mi_2, but the lingual is to a great extent filled by a vertical swelhng on the side of the ridge running from the entoconid to the crista obliqua. The sides of this lingual reentrant are nearly parallel for most of their heights and converge to form a U only near the base of the enamel. The tooth continues the progressive increase in crown height seen in Pi_3. Although thoroughly macroscelidid in structure, P4 is the most distinctive tooth of the series. None of the hving forms has the labial reentrant between trigonid and talonid nearly filled by a swelhng, and in none is the metaconid so far posterointernal to the protoconid. In all except Macroscelides the anterior crest and the meta- conid are widely separated by a deep groove and the sides of the lingual reentrant converge toward the base, forming a V. Mi_2 consist essentially of two triangular pris- matic columns connected by the narrow isthmus formed by the crista obhqua. The protoconids, hypoconids and paraconids are angulate, the metaconids and entoconids more rounded. The paraconids and metaconids are separated by very shallow vertical grooves. The lingual and labial reentrants are large, deep, and extend nearly to the base of the enamel; their sides are parallel and U-shaped below. The trigonids are wider and larger than the talonids, particularly on Mi. The metaconids and entoconids are higher than the protoconids and hypoconids, and there is a shght indication of a hypoconuhd. This cusp may have been larger on the unworn crown, if we may judge from Macroscelides in which it is very prominent on unerupted molars but rapidly be- comes worn away. On the hngual sides of the crests running forward from the entoconids are faint vertical swelhngs corresponding to the Figure 2. Palaeothentoides africanus, dorsal view of mandible, Munchen No. 1932. I. 501. X8. PATTERSON : FOSSIL ELEPHANT SHREWS 305 prominent one in P4. These molars are the highest crowned of the cheek tooth series. M3 is vestigial and much lower crowned than M2. It is composed of the trigonid only, on which the small metaconid is the highest element and the protoconid is sub- ordinated in the paraconid crest; metaconid and paraconid are separated lingually by a shallow depression, and there is a slight vertical ridge on the posterior face, the last vestige of the talonid. The enamel is continuous on all cheek teeth ; it is thick on P4 — M2 although thinning at the paraconids in the molars. M3 is indis- tinguishable from that of Nasilio. Mi_2, on the contrary, are quite different from those of this form and very close indeed to those of Macroscelides, differing only in their slightly lower crowns and, at a corresponding stage of wear, in the presence of enamel around the paraconids. In the other living forms, Mi_2 are somewhat lower crowned, have wide lingual grooves between the paraconids and metaconids, and V-shaped sides to the lingualr eentrants separating trigonids and talonids. Seen from above, the ramus curves very slightly inward from P2 forward and gently outward from M3 backward. There is no trace of the symphysis on the part preserved, indicating that, as in all living macroscelidids except Petrodromus, this did not extend posteriorly beyond the level of the canines. The inner face, as noted by Stromer, is nearly flat, the outer swells out gently oppo- site the molars. As in other forms, there is a posterior mental foramen beneath P4 and an anterior beneath Pi. The height of the horizontal ramus remains rather constant beneath the cheek teeth, decreasing less anteriorly than in the living forms. The ventral border is gently convex beneath the molars and P4, and shows a more marked tendency to turn down beneath Pi_2 than in any other form. As is usual in the group, the ascending ramus begins to rise well behind the last molar and the masseteric fossa is shallow and poorly defined. Discussion: Palaeothentoides is unquestionably a valid genus. It resembles Nasilio and Macroscelides, combining characters of both, and can be referred with assurance to the Alacroscelidinae. There is no need to belabor the fact that this form is no marsupial. A resemblance does exist, particularly in the trigonid, between P4 of Palaeothentoides and ^Ii of Palaeothentes, but this is far from exact. The labial and lingual reentrants separating trigonid and talonid that are large and deep in the African form are, for example, shallow in the South American one, and in any event the teeth Stromer compared are not homologous. P4 Ml Mo Ma 2.8 2.4 2.1 0.8 1.4 1.7 1.5 0.8 306 bulletin: museum of comparative zoology Stromer himself realized almost at once that Palaeothentoides was not a marsupial. A separate of his paper in my possession bears a tj^pewritten shp reading: "Berichfigung: Durch einen neuen Fund ist erwiesen, dass Palaeothentoides 4 Pin and 3 M hat, also sicher kein Didelphier ist." So far as I am aware, however, he never published this retraction. The "new find" referred to is of course No. 1932. I. 501, here described and figured for the first time. Measxirements, in mm, of Mi'mchen No. 1932. I. 501. Pi P2 P3 Length 1.4 2.0 2.3 Width 0.6 0.7 0.9 Length Pi - M3, 14.0; Pi_4, 8.6; Mi_3, 5.6. Height of ramus beneath Mi, external, 2.9. ElephantuLUS Thomas and Schwann Elephantulus Thomas and Schwann, 1906, p. 577. Elephayitomys Broom, 1937, p. 758. In the course of describing the Pleistocene E. langi, Broom observed that certain species of the genus, langi among them, have a molariform P-. Believing that E. rupestris, the type species, lacked this character, he proposed Elephantomys, with E. langi as type, for the reception of those species possessing it. Shortly thereafter, specimens from what he supposed to be the type locality of E. rupestris having come to hand, he concluded that this species did after all have a molariform P-, a fact which in his opinion effectively suppressed Elephantomys. He did not go on to erect a new genus for those forms with a non-molariform P'. Elephantomys was subsequently revi^'ed by Ellerman, Morrison- Scott and Hayman (1953, p. 8). Stating that P- of the type of E. rupestris was non-molariform, they recognized Elephantomys as a subgenus, distinguishing it on the basis of the molariform P- and the possession of less flattened bullae, in which the lateral (i.e. tympanic) portion is higher relative to the median (i.e. ento- tympanic) than in Elephantulus^ . These distinctions are not valid. As regards P-, what is involved is a forward extension of the molarisation field. In Rhynchocyon and Petrodromus this extends hardly or not at all beyond P^; in Nasilio and Macroscelides P^-s have been incorporated to the degree that two lingual cusps are 1 They also reduced Nasilio to the rank of a subgenus of Elephantulus. This does not seem justified. In addition to possessing Ms, Nasilio has higher-crowned posterior cheeli teeth and different hind limb proportions (Evans, 1942). PATTERSON: FOSSIL ELEPHANT SHREWS 307 present on them. Elephantulus is in a state of flux. Among the material available to me, P^ is molariform in intufi (5 specimens) and fuscipes (2), non-molariform in rupestris (5) and pulcher (7). Although predominantly non-molariform in ocularis (29) and rufcsccns (6), it is nevertheless \'ariable in these species, even within what are surely local populations. Thus, in rufescens from Mt. IMbololo, Kenya, of two specimens collected on the same day, one (MCZ 31800) has two lingual cusps on P^, another (MCZ 31802) one. Within ocularis, a small series from Unyanganyi, Tanganyika, Tanzania, includes two specimens (MCZ 25660 and 25683) with two lingual cusps on this tooth and three with one; a specimen from Dodoma, Tanganyika (MCZ 22841), has two hngual cusps on the left side and one on the right. Within the "non-molariform" species, P''' as well as P- is variable in respect of lingual cusp development. A distinction based on bulla structure cuts across one based on premolar structure. Thus the "non- molariform" pulcher has a "flattened" bulla, and the predomi- nantly "non-molariform" rufescens and ocularis have "less flattened" ones. Elephantomys does not merit recognition. A far reaching proposal for a division of Elephantulus, and indeed of the whole subfamily, has been advocated by Van der Horst, who, with co-workers, devoted many years to study of the embryology of the genus, with particular reference to E. myurus jamesoni (I employ Van der Horst's names in this paragraph). In the course of his work there emerged the remarkable facts that in this form approximately sixty eggs are liberated and approximately sixty corpora lutea develop in each ovary, only one of which be- comes implanted, the Graafian follicle is remarkably small, and fat globules are lacking in the ova. Macroscelides prohoscideus was found to agree in all these particulars, and E. capensis in all save for the presence of a few fat globules. E. intufi and E. rupestris stand in striking contrast. In these species only two eggs per ovary are liberated, the Graafian follicle is of normal type, and fat globules are present. Petrodromus tetradactylus is in agreement with them except for the apparent absence of fat globules. On the basis of all this. Van der Horst has suggested (e.g. 1944) that there are only two genera of macroscelidines and that the division passes through the genus Elephantulus of current usage. He has not spelled out what would result were his suggestion to be adopted, but this can be simply put. We would have two genera: Macro- scelides, with prohoscideus, myurus and capensis, and Petrodromus, with tetradactylus, rupestris and intufi; all other species would have to remain in limbo until comparable investigations had been 308 bulletin: museum of comparative zoology carried out on them. If these characters were indeed the touch- stone of macroscehdine systematics such a situation would be acceptable, but there is no real evidence that they are. The genera recognized by mammalogists over the years are clear-cut taxa, distinguishable by different combinations of characters. The sporadic occurrence of the curious ovarian characters suggests that these were independently acquired, or, alternatively, perhaps lost, at various times within the group. This possibility is reenforced if the classification, other than subgeneric, of the South African species and subspecies of Ele- phantuhis proposed by Ellerman, Morrison-Scott and Hayman is correct. Van der Horst's myurus jamesoni is their rupestris jamesoni, his capensis is their rupestris capensis, his intufi and rupestris may be their rupestris and intufi. His division of the subfamily would thus run between subspecies of rupestris in their arrangement. The genetic basis of the ovarian peculiarities may be of a rather simple sort. Elephantulus langi (Broom) Elephaniomys langi Broom, 1937, pp. 758-760, fig. 5. Elephantulus langi Broom, 1938, p. 251; 1948, p. 5. Horizon: Pleistocene. Locality: Cave deposit at Schurveberg, 15 miles west of Pretoria, Transvaal, Union of South Africa. E. langi is evidently represented by rather rich material from the Schurveberg cave deposit, and Broom's description is of the most preliminary sort. The relationship between langi and living forms remains to be determined. Elephantulus antiquus Broom Elephantulus antiquus Broom, 1948, pp. 5-6, fig. 3. Horizon: Earlier Pleistocene. Locality: Bolt's worldngs, Sterkfontein, Transvaal, Union of South Africa. E. antiquus is evidently distinct from E. langi — it has, e.g., a non-molariform P^ — but little more can be said. As in the case of E. langi, there is fairly abundant material, Broom's description is preliminary and incomplete, and the relationship to hving species is unknown. In 1946, Broom {in Broom and Schepers, p. 78) stated that: "The elephant shrew Elephantulus langi, or one very closely allied, occurs in the Plesianthropus cave. The type is from Schurveberg, Pretoria. It is common at Bolt's workings." Presumably the species there referred to is E. antiquus. PATTERSON: FOSSIL ELEPHANT SHREWS 309 Elephantulus rozeti (Duvernoy) From an archaeological site at Redeyef, Tunisia, Gobert (1912) recorded the presence of various genera of mammals, most of which he believed to be referable to living species. Macrocelides (sic) is among those listed. Three cultural levels occur at the site, the two lower Paleolithic and the upper ranging from transitional to Neolithic. Gobert gave no description of the mammalian remains, which it would appear from the text were found in the upper level. Romer (1928, pp. 100, 153, 161) lists this find as Macroscelides rozeti. A little uncertainty attaches to the determi- nation. Thomas (1901, 1913) split North African Elephantulus into two species, E. rozeti and E. deserti, the former with three and the latter with two subspecies. E. deserti is the more eastern of the two and, if valid, the Redeyef material might therefore be referable to it. More likely than not, however, subspecific distinc- tion, at most, is involved. Although Thomas had stated, in 1901, that deserti did not differ in size from rozeti, he claimed, in 1913, that it was smaller; the very few published measurements do not support the assertion. The differences appear to be confined to pelage color. Elephantulus rozeti, which dates from 1838, was long known as Macroscelides rozeti, and numerous specimens so labeled found their way into collections. Many of the labels were not changed when Thomas and Schwann transferred rozeti to their new genus Ele- phantulus. These labels have trapped trusting anatomists and paleontologists. A number of accounts and illustrations in the literature that purport to be of Macroscelides are actually of Elephantulus, based on E. rozeti (e.g. Evans 1942, Fiedler 1953, Grasse 1955, Saban 1956-1957, Van der Klaauw 1929, in part). RHYNCHOCYONINAE RhyncHOCYON Peters Rhynchocyon clarki Butler and Hopwood R. clarki Butler and Hopwood, 1957, pp. 4-11, figs. 2-3. Horizon: Early Miocene. Localities: Type from Songhor local fauna, Kenya; referred material from Rusinga Island, Kenya, found in the upper and lower Hiwegi beds and either in the Kiahera or in the lower part of the Kathwanga beds. This species, so well described by its authors, reveals, as they point out, that the two surviving subfamilies had diverged prior to 310 bulletin: museum of comparative zoology the Miocene. R. clarH ". . . as an early member of the Rhynchocyon Hneage . . . is . . . nearer to the common ancestor of the two groups, and possesses a number of primitive characters which have been lost in Recent representatives of both subfamiUes." The species is notably smaller than the living members of the genus, which suggests a relatively recent increase in size within the Rhynchocyon lineage. MYLOMYGALINAE subfam. nov. Diagnosis: I3, C, P4, M2. I1-3 small, subequal; C — P2 small, single rooted; P4 — M2 large, hypsodont, crowns complex, flat, of grinding type. Molars compressed anteroposteriorly, as wide as long; protoconid and hypoconid angulate, directed antero- externally; hypoconid and hypoconulid forming posterior lophid, entoconid set off from posterior lophid by deep reentrant; re- entrant between paraconid and metaconid situated on anterior face of tooth; M2 large relative to Mi. Talonid of P4 fully molari- form, trigonid larger, more elongate than in molars. Ventral border of horizontal ramus strongly convex, alveolar border con- cave beneath posterior cheek teeth. Mylomygale Broom Mylomxjgale Broom, 1948, p. 6. Type species: M. spiersi Broom. Distribution: Earlier Pleistocene, south Africa. Diagnosis: Sole known genus of the subfamily diagnosed above. Mylomygale spiersi Broom (Fig. 3) Mylomygale spiersi Broom, 1948, pp. 6-8, fig. 4, (1946, in Broom and Schepers, p. 28, fig. 1 N-Qi). Locality: "... a small cave about half a mile to the north of the cave which yielded the Taungs man-ape skull." Approximately 80 miles N. of Kimberly, Bechuanaland, Union of South Africa. Horizon: EarUer Pleistocene. ("This bone breccia is probably of approximately the same age as [those in] the other caves.") Diagnosis: Sole known species of the genus. Discussion: This remarkable little macrosceUdid enjo3'^s the distinction of being the only extinct genus correctly placed in the ' In this paper M. spiersi was figured and listed with the statement that: "It represents a new family of the Menotyphla." No diagnosis or description was given. PATTERSOX: FOSSIL ELEPHANT SHREWS 311 family by its describer. Discovery of Mylomygale revealed the existence of an otherwise unknown division of the family, one that evolved posterior cheek teeth that are as strikingly rodent-like as those of the myohyracines are ungulate-like. Broom's remark to the effect that had the molars been found isolated they would have been regarded as belonging to some peculiar hystricomorph rodent is no exaggeration. The type specimen preserves five small alveoli followed by four grinding teeth. Although he decided that the dental formula was probably I3, C, P4, M2, Broom was in some doubt as to whether the last two alveoli housed the roots of two teeth or of one. "As in all the living jMacroscelids the anterior premolars are double- rooted it might seem more probable that the two sockets held a single premolar, but on the other hand if the anterior premolar were double-rooted then there can only be three premolars, while all living Alacroscelids have four. ... I think it more likely that there were two small single-rooted premolars." This tentative conclusion was, I believe, the correct one. Contrary to Broom's statement, and as pointed out above, the roots of Pi are fused in Macroscelides, in some species of Elephantulus, and in the extinct Metoldobotes, while the roots of P2 are closely appressed in Macro- scelides. The anterior portion of the horizontal ramus is short in Mylomygale, and it is hence not surprising that the roots of P2 had fused. Figure 3. Mylomygale spiersi, lateral view of mandible and crown view of dentition, X4; redrawn from Broom. 312 bulletin: museum of comparative zoology P3, the most anterior tooth preserved, bears the same size rela- tion to P4, and P4 to the molars, as in other members of the family. P4 and the two molars are remarkable for their deep, narrow, and persistent reentrant folds. Comparison with an unworn molar of Macroscelides (Fig. 5h) permits an understanding of the cusp pattern. The anteroexternal angle is composed of the protoconid and the centrally situated paraconid. Hypoconid and hypoconulid are joined to form the posteroexternal and posterointernal angles and the posterior lophid. The metaconid makes up the antero- internal angle and the entoconid the central internal. This degree of independence of the entoconid is a departure from the usual macroscehdid condition, in which entoconid and hypoconulid tend to be connected (cf. Fig. la, b). The large size of M2 relative to Mi is another character peculiar to Mylomygale within the family; in all other known genera, even in the earhest, Metoldobotes, it is decidedly smaller than its predecessor in the series. The size of this tooth provides an example of the reversal of an evolutionary trend, the reversal in this case being associated \Yith. the later trend toward the acquisition of rodent-hke posterior cheek teeth. MYOHYRACINAE (= Myohyracidae Andrews 1914, Myohyracoidea Stromer 1926) (Figs. 4b; 5a, b, c, f, i, j ; 6a, c, e, g, i) Emended diagnosis: Macroscehdidae with complete dental formula; Ij"^ large, wdthout enamel on Ungual faces; posterior cheek teeth hypsodont; M3 greatly reduced; F^f submolariform, Pt essentially molariform; P^ - M^ with moderately undulant ectoloph, paracones and metacones with comparatively shallow labial grooves between them, parastyles and metastyles prominent, parastyles anteroexternal in P"' - :\I2; P^ - M^ with persisting fossettes, those of molars arranged in anterior and posterior pairs; P3 — M2 with two fossettids, one each in trigonid and talonid; horizontal ramus deep beneath posterior cheek teeth, mental foramen beneath P3. Distribution: Early Miocene, east and southwest Africa. Genera included: Myohyrax Andrews 1914, Protijpotheroides Stromer 1922. Discussion: Andrews (1914, pp. 169-171) described Myohyrax oswaldi on a fragment of a ramus with P3 — ^U and some isolated teeth, including an upper molar; this material gave no hint of the vestigial nature of M3. With such evidence in hand it would hardly occur to anyone to make a comparison with the macro- scehdids, and Andrews did not do so. He referred the genus to a PATTERSON: FOSSIL ELEPHANT SHREWS 313 new family of the Hyracoidea. All subsequent students have looked at myohyracines in this light and some of them have com- mented on how aberrant they are within that order. From the work of Stromer (1926) and of Whit worth (1954), it is possible to note that many of the characters in which they differ widely from hyracoids are actually points of resemblance to macroscelidids. a Figure 4. Lateral views of skull and mandible of, a, Nasilio brachyrhynchus, MCZ 43755, and, b, Myohyrax oswaldi, slightly modified from Whitworth; b X2, a not to scale. 314 bulletin: museum of comparative zoology The skull of Myohijrax, figured in outline by Whitworth, is not dissimilar to those of other members of the family (Fig. 4). The facial region in all is long, low and rather narrow; the cranium is short and high ; the zygoma arises over the rear of M^ ; and the glenoid cavity is situated high on the side of the skull. Whitworth shows a slight notch between nasal and premaxilla ; I was unable to detect this in the specimen. The palate is unfortunately not visible. The myohyracine mandible, if allowance be made for the increased depth beneath the hypsodont cheek teeth, is decidedly macroscelidid in appearance. The symphysis is short and shallow; the ascending ramus high and steeply rising, and the coronoid process small; the condyle is high and not expanded transversely; and the angle is hook-like and extended posterodorsally. p2 _ ]Vp of Macroscelides and Nasilio resemble the correspond- ing teeth of Myohyrax in a number of respects. The posterior cheek teeth of myohyracines are somewhat bowed outwardly (Whit- worth, 1954, pi. 6, fig. 2), P- — M^ are incHned backward and P3 — M2 are inclined forward; the bowing is incipient and the pitching definitely present in Macroscelides. M- of myohyracines is unre- duced, in correlation with the retention of M3 ; in Nasilio, in which M3 is retained, the posterior portion of A'P is less reduced than in the other living forms. The great reduction of M3 is, of course, a decided resemblance to the macroscelidids and a striking contrast to the hyracoids. The crown pattern of the upper molars of the myohyracines (Fig. 5) is basically macroscelidid and not hyracoid in such characters as the large, external paracone and metacone, the absence of a mesostyle (in this I agree with Andrews and with Hopwood, believing Whitworth 's mesostyle to be the paracone), and the position and relations of the lophs. In macroscelidines, especially Macroscelides and Nasilio, the lophs are relatively high, the protoloph going to the parastyle, the robust metaloph prima- rily to the paracone; the protocone is connected posteroexternally to the enlarged anteroexternal portion of the metaloph and the posteroloph is transverse, connecting metastyle and hypocone, all very much as in myohyracines. Between paracone, protoloph, metaloph and protocone anteriorly, and between metacone, metaloph and posteroloph posteriorly, two fossettes are isolated. These are aligned anteroexternally to posterointernally, and are the homologues of the anterior and posterior pairs of fossettes in the upper molars of myohyracines. In unerupted or little worn molars of Macroscelides tendencies toward division of each of these fossettes into two may be seen. Patterson: fossil elephant shrews 315 Figure 5. A comparison of myohyracine and macroscelidine cheek teeth. Myohyrax oswaldi: a, P^ - M'; c, M^;/, dm^;;, dm4. Protypotheroides beeizi: b, P2 - Ms; i, Ml. Macroscelides proboscideus: d, M' (unworn); e, M'; gr.dm^; h, Ml (unworn); k, diUi. a, b, c, f, i redrawn from Stromer, j from Whitworth; d, g, h, k MCZ 37022, e MCZ 37023. b X2; a, c, f, i X4; j X6.6; the rest not to scale. The crown structure of P4 and of the lower molars is close to that of Palaeothentoides and Macroscelides (Fig. 5). The lingual re- entrant between trigonid and talonid is less open than in these forms, and this narrowing was probably brought about by a swelling on the entoconid crest similar to but larger than that present in Palaeothentoides. Macroscelides has fossettids in the trigonids and talonids of unworn molars and these are closely com- parable, although much shallower than those occurring in Protypo- theroides. 316 bulletin: museum of comparative zoology The fourth upper milk molar of Myohyrax (Fig. 5f), like the permanent molars, is very similar to the corresponding tooth in macroscelidines. The differences that exist between it and that of, e.g., Macroscelides — parastyle less set off by grooves, more prominent posteroloph, nearly straight lingual wall, protocone not set off by an anterior groove — do not disguise the basic re- semblance. Dm4 of Myohyrax (Fig. 5j), although considerably worn, is again unmistakably macroscelidid in structure and quite unlike that of hyracoids. As in all members of the family, it is very long and low-crowned, with trigonid and talonid approximately equal in length. The paraconid area is set off by external and internal grooves from the large protoconid and the even larger metaconid, which was almost certainly twinned as it is in Macro- scelides. The paraconid was clearly anterocentral in position with a short labial crest and a longer lingual one terminating in a para- stylid. Lingual and labial reentrants between trigonid and talonid are essentially as in Macroscelides. The hypoconid is very large; the entoconid and hypoconulid have become united by wear, while the groove between hypoconid and hypoconulid still persists. With wear this would occur in Macroscelides. An entostylid is present anterior to the entoconid, set off by grooves from it and from the metaconid; a precisely similar structure occurs in dm4 of Petro- dromus. I differ from Whitworth as regards cusp homologies in this tooth. My parastylid is his paraconid, my entoconid and hypoconuhd are regarded by him as a stylar development and my entostylid is his entoconid. In hyracoids, dm'* is somewhat narrower relative to length than is M^, which it otherwise resembles very closely, and dm4 and Mi are nearly identical. Stromer (1926) described and figured various postcranial frag- ments, which he referred to Myohyrax. Except for an atlas, which may not be correctly identified, all of these are decidedly macro- scelidid in appearance. Figure 6 shows a selection of these frag- ments, redrawn from Stromer, compared with corresponding parts of a living member of the family. The resemblances are obvious and do not need to be elaborated. One point may be stressed. The astragalus is about as different as possible from that of hyracoids. This element in the latter is as distinctive in its way as are the corresponding bones of artiodactyls and perissodactyls. In the hjTacoid astragalus the articular area of the trochlea con- tinues distally over the medial side of the short neck into a curious, step-like surface for the reception of the long, stout internal malleolus of the tibia. This specialization had already been attained by early Oligocene, Fayum forms, the earliest known PATTERSON: FOSSIL ELEPHANT SHREWS 317 Figure 6. A comparison of myohyracine and rhynchocyonine postcranial elements. Myohyrax oswaldi: a, c, e, g, i; Rhynchocyon cirnei: b, d, f, h, j. a, b, proximal ends of humeri; c, d, proximal ends of ulnae; e, f, proximal ends of femora; g, h, astragali; i, j, calcanea. a, c, e X2; g, i, X4; R. cirnei not to scale. M. oswaldi redrawn from Stromer, R. cirnei MCZ 43735. members of the order (Schlosser, 1911, p. 126, pi. 13, fig. 2). The comparatively long-necked myohyracine astragalus shows no trace of such a structure, and agrees in all essentials with those of other macroscelidids (Fig. 6 g, h). It is regrettable that none of these pieces is complete enough to give any idea of the degree of fusion of the lower leg bones, not to mention the relative lengths of the limb segments or of the fore and hind Hmbs. Whether or not 318 bulletin: museum of comparative zoology the myohyracines were saltatorial, as are living forms^ remains unknown. Four species of early Miocene myohyracines have been de- scribed: Myohyrax oswaldi Andrews 1914, Pr-otypotheroides beetzi Stromer 1922, Myohyrax doederleini Stromer 1926, and Myohyrax osborni Hopwood 1929. Of these, M. oswaldi and M. doederleini are small and very similar, P. beetzi and M. osborni are much larger and very similar. Whitworth recognizes but two species, oswaldi and beetzi, and in this he is undoubtedly correct. He goes on to synonymize Protypotheroides with Myohyrax, but here I am unable to follow him. It appears to me that the differences between the two valid species, summarized in the diagnoses below, are of generic significance, as genera are defined in this family. INIyohyrax Andrews Myohyrax Andrews, 1914, p. 171. Type species: M. oswaldi Andrews, 1914. Distribution: Early Miocene, east and southwest Africa. Emended diagnosis: Myohyracines with cement in fossettes of cheek teeth; M^ single rooted; fossettids on P3 — M2 ephemeral. Myohyrax oswaldi Andrews Myohyrax oswaldi Andrews, 1914, pp. 169-171, pi. 28, figs. 4-6; Stromer, 1926, pp. 123-124, pi. 41, figs. 26-28; Whitworth, 1954, pp. 26-40, text-figs. 9-15, pi. 5, figs. 3-4, pi. 6. Myohyrax doederleini Stromer, 1926, pp. 120-123, text-fig. 19, pi. 41, figs. 1-23; Hopwood, 1929, p. 6, text-fig. 4. Horizon: Early Miocene. Localities: Kenya: Koru (type locaUty), Karungu, and^Rusinga Island (definitely recorded from the lower Hiwegi beds) ; South- West Africa: Elisabethf elder, a borehole some 37 km SSE of Liideritzbucht (Stromer), and "south of Llideritz Bay" (Hop- wood) . Diagnosis: As for the genus. Intermediate in size between the species of Petrodromus and those of the smaller macroscelidine genera. The combination of relatively persistent fossettes in the upper molars and ephemeral fossettids in the lower molars is almost precisely matched in Macroscelides. The species is common at 1 It is sometimes stated (e.g. Evans. 1942, p. 85) on the basis of observations by field workers that, despite their evident saltatorial adaptations, macroscelidids do not hop but run on all four feet. Run they obviously do but hopping has also been observed. PATTERSON: FOSSIL ELEPHANT SHREWS 319 certain localities. In South-West Africa, Stromer records over 100 from Elisabethfelder, and Whitworth (1958, p. 47) lists 120 from Karungu in Kenya. Protypotheroides Stromer Protypotheroides Stromer 1922, p. 333. Type species: P. beetzi Stromer, 1922. Distribution: Early Miocene, southwest Africa. Emended diagnosis: Myohyracines without cement in fossettes of cheek teeth; M^ two rooted; fossettids on P3 — M2 deep. Protypotheroides beetzi Stromer Protypotheroides beetzi Stromer, 1922, p. 333; 1926, pp. 124-125, pi. 41, figs. 29-31. Myohyrax osborni Hopwood, 1929, pp. 6-8, text-figs. 5-6. Myohyrax beetzi Whitworth, 1954, p. 26. Horizon: Early Miocene. Localities: Langental, some 10 km NNE of Bogenfels (Stromer), and "south of Llideritz Bay" (Hopwood), South-West Africa. Diagnosis: As for the genus. A large species, for this family; comparable in size to Rhynchocyon petersi the largest living form. Neither Stromer nor Hopwood mention cement in P. beetzi, and I was unable to detect any. The anterior wall of the alveolus of M^ is preserved in the type of "ilf . osborni'^ and shows the presence of two roots. The fossettids extend nearly to the bases of the posterior cheek teeth. P. beetzi is rare in comparison with AI. oswaldi, only eight specimens having been recorded. DISCUSSION Intrafamilial Relationships Among the living forms, Rhynchocyon stands apart in a number of characters — e.g., largely or wholly edentulous premaxillae, upper canines large and Pi caniniform, long and broad facial region, large cranial table, no palatal fenestrae, backwardly sloping ascending ramus, presence of chevron bones, radius and ulna not fused, digit I lacking and digit V reduced in the manus. The majority of authors agree in placing Rhynchocyon in a distinct subfamily and this seems justifiable. The extinct R. clarki shows that the subfamily was in existence by Miocene time, and in so doing provides us with the nearest approach to a phyletic hneage that we have. This species, as is not surprising, possesses some 320 bulletin: museum of comparative zoology macroscelidine characters; divergence of the two subfamiUes may date from earUer Ohgocene time. The Macroscelidinae, with four living and two extinct genera, form the core of the family, so far as present knowledge goes. The various forms differ unevenly within rather narrow limits. The heights of the cheek teeth range from brachyodont in Metoldohotes and Petrodromus to subhypsodont in Palaeothentoides, Nasilio and Macroscelides, yet two of the higher crowned forms, Palaeothen- toides and Nasilio, are primitive in their retention of a vestigial M3 {Metoldohotes, the earUest known macroscelidid, had already lost this tooth). Macroscelides stands alone in its possession of highly inflated bullae and epitympanic sinuses. Nasilio, advanced as regards molar height, has the tibia shorter relative to the femur than in either Elephantidus or Macroscelides and resembles in this respect the rather generahzed Petrodromus. The latter, in turn, is specialized as regards the loss of the hallux, and so on. Palaeo- thentoides and the living macroscelidines give the impression of being terminal twigs of a once more numerous and varied group. It is unsafe at present to assert that two or more members of the subfamily are closer to each other than to the rest phylogenetically, since characters in common could well have been achieved inde- pendently. The two extinct groups, Myohyracinae and Mylomygahnae, best regarded for the present as subfamilies, are highly specialized as regards their cheek teeth, the former in an unguIate-Uke, the latter in a rodent-Uke direction. Unfortunately, we know them only at moments in time — the two myohyracines in the earlier Miocene, Mylomygale in the Pleistocene — and hence have no direct evidence bearing on their phylogenies. The myohyracines, which alone in the family retain M^ as well as M3, may have branched off at an early date, possibly Eocene, the mylomygalines perhaps somewhat later. The ancestry of both groups may have lain in the IMacroscehdinae, but this is uncertain. The only extra- African form that has been referred to the family is Pseudorhynchocyon cayluxi Filhol (1892) from the Quercy Phosphorites. This very unsatisfactorily known animal was based on the posterior part of a left mandible, in which the alveoli of the last molar provide the only trace of the dentition. The ascending ramus is much inclined posteriorly, and Filhol saw in this a resemblance to Rhynchocyon. In fact, however, Pseudorhynchocyon in this respect goes far beyond conditions in the living form (Fig. 7). As Butler and Hop wood have pointed out, the ascending ramus of the Miocene R. clarki is considerably less inclined than in the PATTERSON: FOSSIL ELEPHANT SHREWS 321 -'--? .-^ a Figure 7. Posterior portions of mandibles of, a, Rhynchocyon clarki, b, Rhynchocyo7i petersi, c, Pseudorhynchocyon cayluxi. X2. a modified from Butler and Hopwood, b MCZ 22573, c redrawn from Filhol. living species, which suggests that such inchnation is of relatively- recent acquisition within the subfamily. Besides this, there is nothing at all macroscelidid-like about the structure of the ascend- ing ramus of Pseudorhynchocyon. The coronoid process is much stouter than in any member of the family and extends well above the condyle. The latter faces posteriorly and is level with the cheek teeth, not far above them. The angle is low, and a prominent masseteric crest leads downward and then upward from it, pro- jecting down below the level of the ventral border of the horizontal ramus. Butler and Hopwood conclude that "... the reference of Pseudorhynchocyon to the Macroscelididae is most improbable." They are quite right. Whatever the creature may be — and I can offer no constructive suggestion on this score — it is not a member of this wholly African family. 322 bulletin: museum of comparative zoology The Food of Macroscelidids The posterior cheek teeth of myohyracines are comparable as regards degree of hypsodont}^ and crown complexity to those of certain hypsodont notoungulates, or, in miniature, to those of late Miocene or early Pliocene Equinae. As both Hopwood and Whit- worth emphasize, such specialization can only be regarded as an adaptation to a diet consisting in large part of harsh vegetation. Mylomygale has posterior cheek teeth that are similar in height and complexity to those of various hj^psodont rodents. Within the Macroscelididae there have arisen two groups primarily adapted to an abrasive vegetable diet. It thus becomes important to ascer- tain if living members of the familj^ are to some extent herbi- vorous. Structurally, as has long been recognized, the macroscelidid jaw is basically that of a herbi\'ore. The jaw muscles, especially AI. temporalis, show resemblances to those of artiodactyls, although, as Fiedler (1953, p. 161) has pointed out, the disposition of the tendons (Sehnenskelet) is "insectivoran." The structure of the cheek teeth, particularly of the higher-crowned living forms, is con- sistent with a diet at least partially herbi^'orous. Is there e^•idence that these animals do in fact eat plants? Regrettably, no thorough study of the diet of any macroscelidid is available, and the anecdo- tal literature is unsatisfactory. Unsubstantiated assertions to the effect that macroscelidids are exclusively insectivorous are common. Reports by collectors that insects were found in stomachs are sometimes quoted, but insect remains are relatively easy to detect and would likely be looked for (macroscelidids being "insecti- vores"), whereas vegetable remains, if scanty, are more difficult to recognize and might even be passed over ^^'ithout comment on the supposition that they had been accidental!}^ ingested together with the "prey. Nevertheless, a few statements do suggest that these animals are omnivorous. Thus, Sclater (1901, p. 155) states, on the authority of Francis, a collector, that Petrodromus sultan is ". . . very partial to the droppings of the Livingstone buck (Nesotragus livingstonianus) .'" Shortridge writes, of Elephantuhis (1934, p. 21), ". . . although mainly insectivorous [they] are to some degree omnivorous, and may be caught in traps baited with meahes, quaker oats, etc."; and, of E. intufi, "examined stomach contents: insects and a small amount of vegetable matter" (p. 23). "The smaller species . . . feed mainly on ants . . . supplemented by tender shoots, roots and berries" (Walker et al. 1964, p. 134). The PATTERSON: FOSSIL ELEPHANT SHREWS 323 most convincing statement is by Broom (1898, p. 68): "Macro- scelides^, the Elephant-shrew, has a jaw which judging by analogy would certainly be related to a herbivorous form and the molar teeth would seem to be quite in harmony with this determination, and yet though Macroscelides is largely a vegetable feeder [italics mine] I have found in the stomach abundant remains of ants and even of fairly large beetles." Although quite inadequate to reveal how large a role vegetable food may play in the macroscelidid diet, this small budget of information does reveal that plants in one form or another are eaten. If early members of the family were similarly omnivorous, the evolution of predominantly herbi- vorous phyla is readily understandable. The Systematic Position of the Macroscelididae Real knowledge of the macroscelidids dates from 1829-, when Macroscelides was described by Smith, and of the tupaiids from 1821 -, when Raffles proposed Twpaia. Neither group formed part of Bowdich's Insectivora of 1821, based on Cuvier's "les insecti- vores" of 1817, which included representatives of a majority of the living families. Given the knowledge available in the earlier part of the 19th century, however, it was inevitable that both families would be placed in this order. As knowledge improved it was equally inevitable that differences between them and the rest of the Insectivora would become increasingly apparent. Peters (1864), recognizing this, divided the order Insectivora into two major, unnamed groups, one with, one without a caecum, and included the colugos with the tupaiids and macroscelidids in the first. Haeckel (1866, p. cix), excluding the colugos, gave to these groups the formal names Menotyphla and Lipotyphla, the former based equally on the Cladobatida ( = Tupaiidae) and the MacrosceUdea ( = Macroscehdidae) ^, the latter including the original Insectivora. ^ With this, a stage was set. Elephant shrews and tree shrews became firmly associated in the minds of many investigators, and discussions of affinities, particularly of the former group, fell into a 1 The old inclusive genus Macroscelides had not been subdivided at this date, and the question therefore arises as to whether Broom was dealins; with the genus as now restricted. It is known, however, that in 1897-98 he resided in Little Naniaqualand and hence was within the range of M. proboscideus. 2 Macroscelides had been known since 1800, but disguised as Sorex proboscideus, Tupaia since 1820, but disguised as Sorex glis. 3 Simpson (1931, p. 16n; 1945, pp. 176, 183) has stated that Menotyphla was based on the macroscelidids; this does not appear to be the case. 4 There is some tendency nowadays to employ Lipotyphla as an ordinal name for the In- sectivora minus the "Menotyphla." It needs to be emphasized, as McKenna (1963b, p. 4n) has done, that "Lipotyphla" is to all intents and purposes strictly synonymous with Bowdich's Insectivora. 324 bulletin: museum of comparative zoology rut. The characters the two families had in common were hailed as proof of close relationship, almost as though the possession of such things as caeca and normal mammahan zygomatic arches and pubic symphyses were pecuHar to them. Even after attention had become focused on the evident resemblances to primates shown by the tree shrews and strong doubts had been cast on the reahty of Haeckel's Menotyphla, some students continued to associate the two families closely, and even to waft the macroscelidids to the primate heights as a kind of ill-fitting tail to the tupaiid kite. Inclusion of the Tupaiidae in the Primates is a view becoming more and more wadely accepted. There is no occasion here to trace the development of this concept (Carlsson, 1909, 1922; Gregory, 1910; Le Gros Clark, 1934; Simpson, 1935, 1945; and others) or to review the imposing body of data that favors it. Some items of evidence that have not yet passed into the general literature may be mentioned, however. Henckel (1928) and Roux (1947) con- cluded that the chondrocranium of Tupaia excluded the family from the order Primates. This opinion was largely based on the absence of a septum interorbitale, a supposed hall mark of pri- mates. Grasse (1955, p. 1649), who recognized Menotyphla in the Haeckelian sense, utilized this to offset Saban's (1956-7) conclu- sion, based on a thoroughgoing study of the adult skull, that tupaiids were members of the order. Recently, Starck (1960, 1962), working on a wide variety of primates, has found the septum interorbitale to be a highly plastic structure without taxonomic significance, its presence or absence largely depending on the developmental stage under investigation. As he puts it, "Damit verlieren die Hypothesen (Henckel), die Tupaia aus der Prima- tenreihe ausschhessen woUen, ihre Hauptstiitze."^ The placenta- tion of tupaiids was very poorly known until quite recently. Meister and Davis (1956) have helped to fill this gap with their description of three stages in Tupaia minor. They conclude that "morphologically the placenta and other fetal membranes of Tupaia are almost an ideal starting point from which to derive the corresponding structures of the primates." Of particular interest for the present study is their further conclusion, based on compari- son with Van der Horst's drawings (1950) of Elephantulus, that ". . . the placenta and fetal membranes in these two forms differ in almost every respect except placental type"; the data ". . . sup- port the view that the tree-shrews and elephant shrews are not 1 Since this was sent to press, W. Spatz' study of the ontogeny of the cranium of Tupaia glis (Morphol. Jahrb., 106: 321-416, 1964) has come to hand. He concludes that on this evidence tupaiids are primates. PATTERSON: FOSSIL ELEPHANT SHREWS 325 closely related." The little evidence available on ectoparasites supports primate affinities for the former but not for the latter (Patterson, 1957, pp. 23, 26). A few uncertainties remain concern- ing the propriety of including tupaiids in the Primates. Some see the structure of the hand as a bar to inclusion while others do not. Buettner-Janusch and Buettner-Janusch (1964, p. 87) find that the electrophoretic behavior of the hemoglobin differs from that of other prosimians.^ Jane, Campbell and Yashon (1965) have shown that in Tupaia the pyramidal tract occurs in the dorsal funiculus of the spinal cord, whereas in insectivores it occurs in the ventral funiculus and in primates in the lateral. ^ However, only one other prosimian, Nycticehus, has thus far been studied in this connection. A wider investigation may reveal that the distinction is not a clear cut one. It should be recalled, to introduce a note of caution, that until 1952 Tupaia could be said to differ from all primates in its possession of an outer bar of Jacobson's cartilage. In that year Eloff demonstrated the presence of the bar in Galago senegalensis. Certain facts do, of course, await further assessment, but the weight of the evidence now decidedly favors the ordinal reference ; the burden of proof has shifted. The Macroscehdidae, to anticipate a little, are without much doubt a very ancient family that probably arose early in the Cenozoic, possibly even toward the end of the Mesozoic. In some respects specialized, they are nevertheless basically rather primi- tive. It is not surprising therefore that they should to varying degrees resemble other groups of mammals of more or less com- parable antiquity. These resemblances, particularly those to the Insectivora and to the tupaiid primates, have been interpreted as indicative of close affinity to one or the other of these groups, wrongly interpreted I now believe. As regards the soft anatomy, Le Gros Clark (1933, p. 1004) has stated of the brain that "... it would be difficult to conceive two small mammalian brains which are more fundamentally different and divergent in their structure than those of Macroscelides and Tupaia.'' Stephan and Spatz (1962) and Stephan and Andy (1964) also emphasize that the macroscelidid brain differs from those of Insectivora in various ways, notably in the much larger mesen- cephalon and hippocampus. The organ of Jacobson (Broom, 1902, 1 On the basis of serum protein comparisons Goodman (1963, p. 137) concludes: "Although evidence for a definitive taxonomic assignment of the tree shrews has not been gathered, the serological data demonstrate that the tree shrews have affinities with the Primates. (Elephant shrews, placed by some taxonomists with tree shrews, do not show any primate or tupaiid affinities.)" 2 In this, as in certain other respects, Tupaia may be simply primitive. The tract is dorsal in monotremes, marsupials, edentates and rodents. 326 bulletin: museum of comparative zoology 1915) is of marsupial type, resembling that of Twpaia (presumably- primitive in both groups) and very different from that of insecti- vores. Carlsson, although concluding that macroscehdids were close to erinaceids, did record certain characters in the musculature in which they resembled the tupaiids (1909, p. 396). The testes remain abdominal in elephant shrews, whereas the tree shrews have a well developed scrotum. Both, of course, have a caecum, but this is much larger in the elephant shrews. In sum, the evi- dence of the soft parts would appear to oppose close relationship to either the Tupaiidae or to the Insectivora. The distinction between macroscehdids and tupaiids in placenta- tion has been mentioned above (p. 324), as has the remarkable number of eggs liberated from the ovaries of certain of the Macro- scelidinae (p. 307). The two famihes also differ as regards the young. In the elephant shrews these are decidedly precocial, being born fully haired, with the eyes open, and capable of active locomo- tion within a very short time (Hoesch and von Lehmann, 1956, p. 17; ^Yalker et al. 1964, p. 135)^ In tree shrews — and also in the Insectivora generally (Herter, 1957, p. 31) — this is not the case. Uterine bleeding has been described by Van der Horst (1954, and references there cited) in Elephantulus "myurus,^' with the suggestion that this foreshadows the menstrual cycle of the higher primates. The bleeding is of an unusual type, however. During diestrum a polyp-like growth forms in one part of the uterus and disintegrates at the end of the stage; coiled arteries, which are ''enormously developed," are confined to this part. In all likeli- hood this is simply another macroscelidid peculiarity. The dentition of macroscehdids, particularly the posterior cheek teeth, is unlike that of any other group of mammals. Carlsson saw resemblances to the teeth of Erinaceidae, but these are not close. The Macroscelididae have at times been placed in the Insectivora "Dilambdodonta" although there is nothing whatsoever dilambdo- dont about their molars. Frechkop (1931) has stated that the cheek teeth resemble those of ungulates more than those of any other major group. This is correct — the myohyracines evolved molars so ungulate-like as to mislead some very competent students — but there is no detailed similarity to any particular group of 1 A very curious observation concerning the young of Elephantulus rupeslris has been reported. Fitzsimons (1920, pp. 12-13) quotes an observer, van Musschenbroek, who noticed that the two young ". . . were hanging on to something on top of the shoulder blades. On examination I found they were two teats, one on either side. [The mother] carried them by these teats [and] took good care to see that they were hanging on before she hopped away like a miniature kangaroo." As regards the position of the teats confirmation is supplied by Burton (1955), who states that Hayman found them to be situated high on the flanks just behind and above the scapulae. PATTERSON : FOSSIL ELEPHANT SHREWS 327 hoofed mammals. The resemblance is an interesting example of convergence, as Friant (1935) concluded. On the basis of his observations, Frechkop asserted that the macroscelidids were on the ungulate road, just as the tupaiids were on the primate one. In this, of course, he went too far. All, or practically all, ungulate orders have surely emerged from the Condylarthra (to which the Arctocyonidae properly belong), and there is nothing suggestively condylarthran in the elephant shrews. Nevertheless, the myohyra- cines do raise the possibility that placental mammals could achieve "ungulate" grade independently of that order.^ The dentition of the extinct macroscelidids gives no hint as to relations with any other group. The earliest known form, Metoldobotes, had the typical pattern and had lost the last molar. The most recent study of the skeleton of the Alacroscelididae is that of Evans (1942), who compared representatives of all genera except Macroscelides (see p. 309) with those of Tupaia and Echino- sorex. He concluded that the macroscehdids resembled Tupaia in 30 osteological features and Echinosorex in 13. Further, he claimed that out of 40 lemuroid features mentioned by Gregory and by Carlsson as occurring in Tupaia, the macroscelidids shared in 32. This "simple morphological balance," as Simpson (1945, p. 176) called it, has been seized on as evidence for the reality of Menotyphla sensu Haeckel (e.g. Grasse, 1955, p. 1649; Heim de Balzac and Bourhere, 1955, p. 1691). It does not provide such evidence. Evans' study is in fact a classic example of the "rut discussions" mentioned above. He has shown beyond question that tupaiids and macroscelidids differ in a number of features, some of them more or less similar, from Echinosorex, which, as a true insectivore, has various characters not found in other major groups of mammals. This is a far cry from proving a close rela- tionship between elephant shrews and tree shrews, however. Looking beyond the restricted prosimian-tupaiid-macroscelidid- insectivore circle, it becomes apparent at once that few of Evans' characters in common between tupaiids and macroscelidids are confined to these families. Furthermore, the resemblance between the two in many of the features cited is far from close. Some of the common features, e.g. the relatively large braincase, were in all probability independently acquired (the braincases at least house very dissimilar brains); others, e.g. the structure of the orbito- temporal region and of the zygomatic arch, are simply primitive eutherian, or therian, characters inherited by both; others again. 1 The only other possible candidates for such a distinction that I can think of are the notoungulates. I am not suggesting a macroscelidid-notoungulate relationship! 328 bulletin: museum of comparative zoology e.g. the presence of a free centrale and of a third trochanter on the femur, are characters so widely possessed as to be meaningless in this context; yet others, e.g. large auditory bullae and slender coronoid process of the mandible, are not sufficiently similar structurally to quahfy as significant resemblances; and so on. The same apphes to the features cited as occurring in common between macroscehdids and lemuroids. Butler (1956, p. 476) has listed certain cranial characters in which the Macroscelididae, Dermop- tera, Tupaiidae and Lemuroidea resemble each other. Some of these features are of the same sort as those cited by Evans. As regards the JNIacroscehdidae, at least, they are equally open to question, as Butler recognized. The evidence from the hard parts seems to me to point in the same direction as that from the soft : namely, that the macroscehdids are sharply distinct from both Tupaiidae and Insectivora. Resemblances to ungulates do exist — Rhynchocijon and Orycteropus are remarkably similar in the disposi- tion of the bones in the orbitotemporal region (cf. figs. 124 and 177 in Gregory, 1920), the rostral and caudal entotympanics of elephant shrews compare rather closely with those of notoungulates, fusion of distal elements in the hmbs occurs in macroscehdids and in hoofed mammals, the astragalus has a fairly long neck in some small and primitive ungulates, etc. — but these appear to be either convergent or simply primitive in both. Evans concluded, on the basis of superficial resemblances, that Anagale from the early Ohgocene of Mongolia "... is, in many osteological features, intermediate between the Alacroscelidae and the Tupaiidae and is either the common ancestor of the two families or quite close to it." New evidence, derived from a hitherto undescribed specimen of Anagale, and from Anagalopsis, reveals that the cheek teeth are quite different from those of either family, and that the tympanic forms the lateral portion of the bulla instead of being a ring enclosed by the entotympanic, as Simpson (1931) had supposed. Basing his conclusion on this evidence and on the very peculiar structure of the unguals, AIcKenna (1963a) has re- moved the Anagahdae from the Tupaioidea, where Simpson had placed them, and hsted them as Eutheria incertae sedis. Whatever the anagalids may prove to be, they are not related to the macro- scehdids. Metoldobotes, it may be recalled, was contemporary with Anagale. As will by now be evident, I believe the elephant shrews to be a group of mammals distinct from both the Insectivora and the tupaiid primates. Their known distribution is exclusively African, and they make their first appearance in the record in the earliest PATTERSON: FOSSIL ELEPHANT SHREWS 329 adequately known mammalian fauna of that continent. They are accompanied there by an array of mammalian groups unknown elsewhere in deposits of earlier or similar date — hyracoids, arsinoitheres, moeritheres, barytheres, proboscideans and catar- rhine primates. So notable a degree of endemism argues a long isolation of Africa^ (Darlington, 1957, pp. 365, 590; Patterson, 1957, p. 45), one lasting throughout much of the Eogene at least. Macroscelidids may well have been members of this "old African" fauna, survivors from the later Cretaceous beginnings of the Eutheria. As a group they are more diversified and contain more genera than almost one-third of the currently recognized orders of eutherian mammals. They are, I believe, worthy of ordinal rank. Butler (1956) has proposed for them the ordinal name Macro- scelidea^ without definition. Such action seems preferable to restriction of Menotyphla to the elephant shrews. ]\ienotyphla has long had a proto-primate flavor, and since the tupaiid half of the artificial assemblage almost surely is primate and the macro- scelidid half assuredly is not, it hardly seems desirable to attempt perpetuation of so ambiguous a name for the latter alone. The order may be definied as follows: MACROSCELIDEA Dentition I°3^, C\ , Ft, M2I3 ; Ft large, molarif orm ; upper cheek teeth without mesostyles; M3, when present, greatly reduced; posterior cheek teeth brachyodont to hypsodont. Skull with complete zygomatic arch; orbits large, open posteriorly; maxilla not extending into orbital wall, palatine with orbital wing; auditory bulla compound, ectotympanic, rostral and caudal entotympanics, ahsphenoid, squamosal, periotic participating ; mandible with high ascending ramus, condyle well above level of cheek teeth, coronoid process small. Pelvis with pubic symphysis; humerus with entepicondylar foramen; distal segments of legs longer than proximal; radius and ulna fused or closely appressed, tibia and 1 When this isolation began and ended is of course uncertain. Darlington, on the basis of the rather few northern forms that occur in the Fayum deposits, believes that a connection had by then become established. This does not seem certain ; that only two or three of a great many northern groups would have made their way over a land bridge had this been fully in existence is rather unlikely. The Fayum rodents belong to a family not known in Eurasia; they may well have descended from waif ancestors transported during the period of isolation. 2 Somewhat vaguely, however, since in the body of his paper (p. 479) he expressed doubt as to the propriety of including the elephant shrews in the Insectivora ("Lipotyphla" in his terminology), suggesting that they should either ". . . be included in the Primates as an out- lying suborder, or a new order, Macroscelidea, should be created for them." In the summary (p. 480) he simply remarked that they are ". . . placed in a new order, Macroscelidea." Haeckel's prior — and invalid (by modern standards) — use of the same name for the family does not constitute preoccupation. I was previously dubious about recognition of the order (1957, p. 23), but with the increase in knowledge of the fossil record my doubts have disappeared. 330 bulletin: museum of comparative zoology fibula fused; pollex and hallux reduced or absent; astragalar neck moderately long. Proboscis long, flexible; organ of Jacobson of marsupial type; brain with relatively large mesencephalon and speciahzed hippocampus ; caecum relatively large ; testes abdominal. One family, Macroscelididae, with four subfamilies: Macro- scelidinae, Rhynchocyoninae, Mylomygahnae, jNIyohyracinae. Known range: Early Oligocene to Recent, Africa. SUMMARY The Macroscelididae, a wholly African group so far as known, includes four subfamihes, two of which are extinct. The Macro- scelidinae date from the early Oligocene of the Fayum, where they are represented by Metoldobofes, a form originally referred to the insectivore family JNIixodectidae. Palaeothentoides of the early Pleistocene (?), first described as a marsupial, is a valid member of the subfamily. Extinct species of Elephantulus are known from the Pleistocene. The early Miocene Rhynchocyon clarki provides the only fossil record of the Rhynchocyoninae. The subfamily Mylomygalinae is proposed for the Pleistocene Mylomygale, a remarkable form with hypsodont posterior cheek teeth convergent toward those of various rodents. The Myohyracinae, hitherto placed in the Hyracoidea as Myohyracidae or ]\Iyohyracoidea, are represented by the early Miocene Alyohyrax and Protypotheroides. Their posterior cheek teeth are decidedly ungulate-like and com- parable in complexity and degree of hypsodonty to those of Equinae and hypsodont Notoungulata. Pseudorhynchocyon cayluxi from the Quercy Phosphorites is not a member of the family. The extinct subfamilies were beyond doubt predominantly herbivorous. Some evidence indicates that the surviving forms are to a degree omnivorous. The affinities of the family are reviewed and the conclusion reached that macroscehdids are not closely related either to the tupaiid primates or to the insectivores. Resemblances to ungulates are either convergent or primitive. Butler's order Macroscelidea is recognized for the reception of the group and a definition is offered. REFERENCES CITED Andrews, C. W. 1914. On the lower Miocene vertebrates from British East Africa col- lected by Dr. Felix Oswald. Quart. Jour. Geol. Soc. London, 70: 163-186. PATTERSON: FOSSIL ELEPHANT SHREWS 331 Broom, R. 1898. On the affinities and habits of Thylacoleo. Proc. Linn. Soc. New South Wales, 23: 57-74. 1902. On the organ of Jacobson in the elephant-shrew (Macroscelides proboscideus). Proc. Zool. Soc. London, 1902: 224-228. 1915. On the organ of Jacobson and its relations in the "Insectivora" — Part L Tupaiia and Gymnura. Ibid., 1915: 157-162. 1937. On some new Pleistocene mammals from limestone caves of the Transvaal. South Afr. Jour. Sci., 33: 750-768. 1938. Note on the premolars of the elephant shrews. Ann. Transvaal Mus., 19: 251-252. 1948. Some South African Pliocene and Pleistocene mammals. Ibid., 21: 1-38. Broom, R. and G. W. H. Schepers 1946. The South African fossil ape-men, the Australopithecinae. Mem. Transvaal Mus., No. 2: 1-272. Buettner-Janusch, J. and V. Buettner-Janusch 1964. Hemoglobins of Primates. In Evolutionary and genetic biology of Primates, ed. J. Buettner-Janusch. Vol. 2, pp. 75-90. New York and London: Academic Press. Burton, M. 1955. Elephant shrews. lUus. London News, 226: 412. Butler, P. M. 1956. The skull of Idops and the classification of the Insectivora. Proc. Zool. Soc. London, 126: 453-481. Butler, P. M. and A. T. Hopwood 1957. Insectivora and Chiroptera from the Miocene rocks of Kenya Colony. Brit. Mus. (Nat. Hist.). Fossil Mammals of Africa, No. 13: 1-35. Carlsson, a. 1909. Die Macroscelidae und ihre Beziehungen zu den iibrigen Insecti- voren. Zool. Jahrb., Abt. Syst. Okol. Geogr. Tiere, 28: 349-400. 1922. tlber die Tupaiidae und ihre Beziehungen zu den Insectivora und den Prosimiae. Acta Zool., 3: 227-270. Darlington, P. J., Jr. 1957. Zoogeography, the geographical distribution of animals. New York: John Wiley and Sons, Inc., pp. i-xiii, 1-675. Ellerman, J. R., T. C. S. Morrison-Scott and R. W. Hayman 1953. Southern African mammals 1758 to 1951: a reclassification. London: British Museum (Natural History), pp. 1-363. Eloff, F. C. 1952. On the organ of Jacobson and the nasal floor cartilages in the chondrocranium of Galago senegalensis. Proc. Zool. Soc. London, 121: 651-655. Evans, F. G. 1942. Osteology and relationships of the elephant shrews {Macroscel- ididae). Bull. Amer. Mus. Nat. Hist., 80: 85-125. 332 bulletin: museum of comparative zoology Fiedler, W. 1953. Die Kaumuskulatur der Insectivora. Acta Anat., 18: 101-175. FiLHOL, H. 1892. Note sur un insectivore nouveau. Bull. Soc. Philom., (8) 4: 134. FiTZSIMONS, F. W. 1920. The natural history of South Africa. Mammals. Vol. IV. London: Longmans, Green and Co., pp. i-xix, 1-271. Frechkop, S. 1931. Note prcliminaire sur la dentition et la position systematique des MacrosceUdae. Bull. Mus. Roy. Hist. Nat. Belg., 7 (6): 1-11. Friant, M. 1935. La morphologic des dents jugales chez les macroscelidides. Proc. Zool. Soc. London, 1935: 145-153. GOBERT, E. 1912. L'abri do Redeyef. L'Anthropologie, 23: 151-168. Goodman, M. 1963. Serum protein comparisons by two-dimensional starch-gel electro- phoresis and immunodiffusion plate reactions applied to the systematics of primates and other mammals. Proc. XVI Internat. Congr. Zool., 4: 133-142. Grass:^, P.-P. 1955. Ordre des insectivores. Anatomic et reproduction. Affinites zoologiques des diverses families entres elles et avec les autres ordres de mammiferes. In Traite de Zoologie, P.-P. Grasse ed. Paris: Masson et Cie. Vol. 17, Mammiferes, fasc. 2: 1574-1653. Gregory, W. K. 1910. The orders of mammals. Bull. Amer. Mus. Nat. Hist., 27: 1-524. 1920. Studies in comparative myology and osteology: No. IV. A review of the evolution of the lacrymal bone of vertebrates with special reference to that of mammals. Ibid., 42: 95-263. Haeckel, E. 1866. Generelle Morphologic der Organismen. Berlin: Georg Reimer, Vol 2. Pp. i-cix, 1-462. Heim de Balsac, H. and F. Bourliere 1955. Ordre des insectivores. Systematique. In Traite de Zoologie, P.-P. Grasse, ed. Paris: Masson ct Cie. Vol. 17, Mammiferes, fasc. 2: 1653-1697. Henckel, K. O. 1928. Das Primordialcranium von Tupaiia und der Ursprung der Primaten. Zeitschr. Anat. Entwickl., 86: 204-227. Herter, K. 1957. Das Verhalten der Insectivorcn. In Handbuch der Zoologie, J.-G. Helmcke, H. von Lengerken and D. Starck, eds. Berlin: Walter de Gruyter u. Co. Bd. 8, lief. 9, 10(10): 1-50. HoEscH, W. and E. von Lehmann 1956. Zur Saugetier-Fauna Sudwestafrikas. Bonn. Zool. Beitr., 7: 8-57. PATTERSON: FOSSIL ELEPHANT SHREWS 333 HopwooD, A. T. 1929. New and little-kuowu maninials from the Miocene of Africa. Amer. Mus. Novit., No. 344: 1-9. Jane, J. A., C. B. G. Campbell and D. Yashon 1965. Pyramidal tract: a comparison of two prosimian primates. Science, 147: 153-155. Le Gros Clark, W. E. 1933. The brain of the Insectivora. Proc. Zool. Soc. London, 1932: 975-1013. 1934. Early forerunners of man. A morphological study of the evo- lutionary origin of the primates. London and Baltimore: Balliere, Tindall and Co.x, William Wood and Co. Pp. i-xvi, 1-296. 1959. The antecedents of man. An introduction to the evolution of the Primates. Edinburgh: The University Press. Pp. i-ix, 1-374. McKenna, M. C. 1963a. New evidence against tupaioid affinities of the mammalian family Anagalidae. Amer. Mus. Novit., No. 215S: 1-16. 1963b. Primitive Paleocene and Eocene Apatemyidae (Mammalia, Insectivora) and the primate-insectivore boundary. Ibid., No. 2160: 1-39. Matthew, W. D. 1910. Schlosser on Fayum mammals. A preliminary notice of Dr. Schlosser's studies upon the collection made in the Oligocene of Egypt for the Stuttgart Museum by Herr Markgraf. Amer. Nat., 49: 700-703. 1915. A revision of the Wasatch and Wind River faunas. Part IV — • Entelonychia, Primates, Insectivora (part). Bull. Amer. Mus. Nat. Hist., 34: 429-483. Meister, W. and D. D. Davis 1956. Placentation of the pygmy treeshrew. Fieldiana: Zool., 35: 71-84. Patterson, B. 1957. Mammalian phylogeny. In Premiere symposium sur la specificite parasitaire des parasites des vertebres. Neuchatel: Paul Aldinger. Pp. 15-48. Peters, W. 1864. Ueber die Saugethier-Gattung Sulenodon. Abh. K. Akail Wiss. Berlin, 1863: 1-22. ROMER, A. S. 1928. Pleistocene mammals of Algeria. Fauna of the Paleolithic station of Mechta-el-Arbi. Bull. Logan Mus., 1: 80-163. Roux, G. H. 1947. The cranial development of certain Ethiopian insectivores and its bearing on the mutual affinities of the group. Acta Zool., 28: 165-307. Saban, R. 1956- Les affinites du genre Tiipaiia Raffles 1821, d'apres les caracteres 1957. morphologiques de la tete osseuse. Ann. Paleont., 42: 169-224 (1956); 43: 1-44 (1957). 334 bulletin: museum of comparative zoology SCHLOSSER, M. 1910. tjber einige fossile Siiugetiere aus dem Oligociin von Agj^pten. Zool. Anz., 35: 500-508. 1911. Beitrage zur Kenntnis der oligozanen Landsiiugetiere aus dem Faj'um, Agypten. Beitr. Pal. Geol. Osterr.-Ungar. Orients, 24: 51-167. SCLATER, \V. L. 1901. The mammals of South Africa. Vol. II. Rotlentia, Chiroptera, Insectivora, Cetacea and Edentata. London: R. H. Porter. Pp. i-xii, 1-241. Shortridge, G. C. 1934. The mammals of South West Africa. A biological account of the forms occurring in that region. Vol.1. Pp. i-xxv, 1-4157. London: William Heinemann Ltd. Simpson, G. G. 1931. A new insectivore from the Oligocene, Ulan Gochu horizon, of Mongolia. Amer. Mus. Novit., No. 505: 1-22. 1935. The Tiffany fauna, upper Paleocene. II. Structuie and relation- ships of Plesiadapis. Ibid., No. 816: 1-30. 1945. The principles of classification and a classification of mammals. Bull. Amer. Mus. Nat. Hist., 85: i-xvi, 1-350. Satrck, D. ' 1960. Das Cranium eines Schimpansenfetus {Pan troglodytes [Blumen- bach 1799]) von 71 mm SchStlg., nebst Bemerkuiigen liber tlie Korperform von Schimpansenfeten. (Beitrag zur Kenntnis des Primatencraniums II). Morphol. Jahrb., 100: 559-647. 1932. Das Cranium von Propithecus spec. (Prosimiae, Lemuriformes, Indi-iidae). (Beitrage zur Kenntnis des Primaten-Craniums 111). Bibl. primat., 1: 163-196. Stephan, H. and O. J. Andy 1964. Quantitative comparisons of brain structures from insectivores to primates. Amer. Zool., 4: 59-74. Stephan, H. and H. Spatz 1962. Vergleichend-anatomische Untersuchungen an Insectivorengehii-- nen. IV, Gehirne afrikanischer Insectivoren. Versuch cinei' Zuordnung von Hirnbau und Lebensvveise. Moiphol. Jahrb., 103: 108-174. Stromer, E. 1922. Erste Mitteilung iiber tertiiire Wirbeltier-Reste aus Deutsch- Siidwestafrika. Sitzungsber. math.-phys. Kl. Bayer. Akad. Wiss., 1921: :« 1-340. 1926. Reste Land- und Siisswasser-bewohnender \Mrl)eltiere aus den Diamantfeldern Deutsch-Siidwestafrikas. //; Kaiser, E., Die Diamantenwiiste Siidwestafrikas, vol. 2. Pp. 107-153. Berlin: Dietrich Riemer. 1931. Reste Siisswasser und Land bewohnender Wirbeltiere aus den Diamantfeldern Klein-Namaqualandes (Siidwestafrika). Sitz- ungsber. math.-nat. Abt. Bayer. Akad. Wi.ss., 1931: 17-47. PATTERSON: FOSSIL ELEPHANT SHREWS 335 1932. Palacothcntoides africanus nov. gen., nov. sp(>c., oiti erste.s Iknitcltier aus Afrika. Ibid., 1931: 177-190. Thomas, O. 1901. List of small mammals obtained by Mr. A. E. Pease, M.P., during his recent expedition to Abyssinia, with descriptions of three new forms of Macroscelides. Ann. Mag. Nat. Hist., (7) 8: 154-156. 1913. List of mammals obtained by the Hon. Walter Rothschild, Ernst Hartert and Carl Hilgert in western Algeria during 1913. Novit. Zool., 20: 5S6-591. Thomas, O. and H. Schwann 190B. The Rudd exploration of South Africa. V. List of mammals obtained by Mr. Grant in N. E. Transvaal. Proc. Zool. Soc. London, 1906: 575-591. V'an der Horst, C. J. 1944. Remarks on the systematics of Elephantulus. Jour. Mammal., 25: 77-82. 1950. The placentation of Elephantulus. Trans. Roy. Soc. S. Afr., 32: 435-629. 1954. Elephantulus going into anoestrus: menstruation and abortion. Phil. Trans. Roy. Soc. London, 238 B: 27-61. Van der Klaauw, C. J. 1929. On the development of the tympanic region of the skull in the Macroscelididae. Proc. Zool. Soc. London, 1929: 491-560. Wagner, P. A. and H. Merensky 1929. The diamond deposits on the coast of Little Namaqualand. With an appendix on the palaeontology of the Namaqualand coastal deposits by S. H. Haughton. Trans. Geol. Soc. S. Afr., 31: 1-41. Walker, E. P., F. Warnick, K. I. Lange, H. E. Uible, S. E. Hamlet, M. a. Davis and P. F. Wright 1964. Mammals of the world. Vol.1. Pp. i-1, 1-644. Baltimore: The Johns Hopkins Press. Whitwurth, T. 1954. The Miocene hyracoids of Ea.st Africa. Brit. Mus. (Nat. Hist.). Fossil Mammals of Africa, No. 7: 1-58. 1958. Miocene ruminants of East Africa. Ibid., No. 15: 1-50. (Received 19 February 1965.) Bulletin of the Museum of Comparative Zoology HAEVARD UNIVERSITY Vol. 133, No. 7 PANAMANIAN SPIDERS OF THE GENUS TMARUS (ARANEAE THOAIISIDAE) By Arthur M. Chickering CAMBRIDGE, MASS., U.S.A. PRINTED FOR THE MUSEUM November 26, 1965 Publications Issued by or in Connection WITH THE MUSEUM OF COMPARATIVE ZOOLOGY HARVARD UNIVERSITY Bulletin (octavo) 1863 — The current volume is Vol. 133. Breviora (octavo) 1952 — No. 235 is current. Memoirs (quarto) 1864-1938 — Publication terminated with Vol. 55. Johnsonia (quarto) 1941 — A publication of the Department of Mollusks. Vol. 4, no. 43 is current. Occasional Papers of the Department op Mollusks (octavo) 1945 — Vol. 2, no. 31 is current. Proceedings op the New England Zoological Club (octavo) 1899-1948 — Published in connection with the Museum. Publication terminated with Vol. 24. Tlie continuing publications are issued at irregular intervals in num- bers which may be purchased separately. Prices and lists may be obtained from the Publications Office of the Museum of Comparative Zoology, Cambridge 38, Massachusetts. Peters' Check-List of Birds of the "World. Volume 1 is out of print; volumes 2 to 7, 9, 10 and 15 are currently available. Phylogeny and Evolution of Crustacea. Proceedings of a confer- ence held at Cambridge, Mass., March 6-8, 1962. H. B. AVhittington and W. D. I. Rolfe, editors. Cambridge, Mass., 192 pp., 80 figs., 1963. Cloth bound, $6.75. Fishes of the Gulf of Maine, by Henry B. Bigelow and William C. Schroeder. Washington, viii - 577 pp., 1953. Photo-offset reprint, $6.50. Publications of the Boston Society of Natural History The remaining stock of the scientific periodicals of the Boston Society of Natural History has been transferred to the Museum of Co'upaTO'ive Zoo'ogy for distribution. Prices for individual numbers may be liad upon request. Bulletin of the Museum of Comparative Zoology HAEVARD UNIVERSITY Vol. 133, No. 7 PANAMANIAN SPIDERS OF THE GENUS TMARUS (ARANEAE THOMISIDAE) By Arthur M. Chickering CAMBRIDGE, MASS., U.S.A. PRINTED FOR THE MUSEUM November, 1965 Bull. Mus. Comp. Zool., Harvard Univ., 133 (7): 337-368, November, 1965 No. 7. — Panamanian Spiders of the Genus Tmarus {Araneae, Thomisidae) By Arthur M. Chickering The Pickard-Cambridges (1889-1905) recognized seven species of Tmarus from Panama. Petrunkevitch (1925) reported only a single immature specimen of the genus from Panama. Banks (1929) reported two species now regarded as T. ineptns O.P.- Cambridge and T. studiosus O.P.-Cambridge. As a result of my earlier study of the genus (1950) I was able to recognize eighteen species. Four of these were known only from females, seven known only from males, with the remaining seven probably known from both sexes. Roewer (1954) listed twenty species known from Panama. This list includes T. separatus Banks and T. interritus Keyserhng, but this inclusion seems to have been an error. T. separatus Banks is apparently known only from Mexico and T. interritus Keyserling only from Brazil. Since the publication of my previous paper on this genus (1950), I have collected spiders in Panama during four additional periods as a result of which I have accumulated a considerable number of specimens in this generic group. These have been completely worked over with the result that I am now compelled to recognize twelve additional species, thus making a total of thirty species known from this small area. It continues to be very difficult to match up males and females correctly and I am obliged to state that I have some doubts con- cerning the accuracy with which this has been done in the past. As I have frequently stated in my published papers, I am deeply grateful for the many privileges extended to me by the staff of the Museum of Comparative Zoology, Harvard University, over a period of many years. My studies could never have been con- tinued without this aid and encouragement. Special acknowledge- ments should be extended to Dr. Ernst Mayr, Director; Dr. P. J. Darlington, Jr., Alexander Agassiz Professor of Zoology; and Dr. Herbert W. Levi, Associate Curator of Arachnology. Grant No. GB-1801 from the National Science Foundation made it possible for me to spend seven months making collections of spiders in the West Indies and Panama during the latter part of 1963 and the first five months of 1964. This grant is also making it possible for me to continue my studies at the Museum of Com- parative Zoology for a considerable period. A Guggenheim Fellowship made it possible for me to collect in Jamaica, W. I., and in Panama, in 1957 and 1958, and also to spend four months in the 340 bulletin: museum of comparative zoology British ^Museum (Natural History) studying important collections. Dr. G. Owen Evans and Mr. D. J. Clark, Department of Zoology (Arachnida), British Museum (Natural History), have recently loaned me important specimens of the genus Tmarus for study in the preparation of this paper. All types named in this publication together with all other specimens of the genus Tmarus in my personal collection will be deposited in the JNIuseum of Comparative Zoology, Harvard University. Genus Tmarus Simon, 1875 The type species of the genus is Tmarus piger (Walckenaer), 1802; widely distributed in Europe and Asia. The genus is cosmo- pohtan in distribution and is most abundant in the neotropical region. In addition to the general family characteristics, generic features common to the species from Panama may be stated as follows: Chelicerae without teeth; only slightly porrect; essentially parallel. Lip much longer than wide at base. Eyes: eight in two rows; viewed from above, posterior row definitely recurved, anterior row slightly so; lateral eyes larger than medians, on well defined tubercles; posterior tubercles the larger; central quadrangle usually wider behind than long and wider behind than in front. Clypeus porrect, often conspicuously so; height equal to several diameters of AME. Carapace usually abruptly declined to pos- terior border; without a median thoracic groove or fovea. Legs: spinose; usually 1243 in order of length but occasionally 2143 or other minor deviation; one and two nearly equal in length and longer than three and four which are also nearly equal in length. Tarsal claws toothed; palpal claw in females also toothed. Ab- domen often with a posterior tubercle or other modification; with many short, stiff spines and spinules. Male palpi usually well supplied with apophyses, and female epigyna usually well defined and often complicated. To conserve space certain features pos- sessed in common by all species under consideration will not be specifically mentioned in the technical descriptions of new species in this paper. For the same reason a more concentrated description than usual in my published papers will be adopted in this and succeeding publications. When eyes are referred to by the abbre- viations Ar^IE, ALE, PME and PLE, the anterior medians, an- terior laterals, posterior medians and posterior laterals are respec- tively designated. chickering: tmarus from Panama 341 A complete list of the species of the genus Tmarus from Panama as they are now recognized may be given as follows: Tmarus aculeatus Chickering; T. hucculentus Chickering; T. cognatus Chickering; T. conforfus Chickering; T. corruptus 0. P.-Cambridge; T. crefatus sp. nov. ; T. curvus Chickering; T. decens 0. P.-Cam- bridge; T. decorus sp. nov.; T. hinnphreyi sp. nov.; T. impedus sp. nov.; T. ineptus O. P.-Cambridge; T. innotus sp. nov.; T. innumus sp. nov.; T. intentus 0. P.-Cambridge; T. levii sp. nov.; T. longus sp. nov.; T. morosus Chickering; T. mundulus O. P.-Cambridge; T. obsecus sp. nov.; T. parki Chickering; T. pauper 0. P.-Cam- bridge; T. peregrinus Chickering; T. probus Chickering; T. pro- dudus Chickering; T. protobius sp. nov.; T. rubinus sp. nov.; T. sigillatus Chickering; T. stiidiosus 0. P.-Cambridge; T. vitusus sp. nov. Ten of the species listed above are known only from males ; ten species are known only from females ; the remaining ten are probably known from both sexes. Key to the males of known species of Tmarus from Panama 1. Species with embolus definitely and conspicuously curled either at anterior end of bulb on ventral side or on retrolateral side {contortus, curvus, irwrosus, produdus) 2 la. Species with embolus either extended more or less around the margin of bulb and not definitely curled or, apparently, restricted to anterior end of bulb {aculeatus, cretatus, decorus, humphreyi, ineptus, innotus, intentus, mundulus, obsecus, parki, pauper, peregrinus, probus, sigillatus, studiosus, vitusus) 5 2. Palpal tibial apophyses short, only ventral one strongly chitinized; cymbium deeply excavate at retrolateral basal corner; embolus deeply grooved, arises near middle of anterior border of bulb, makes a complete retrolateral circle and then loops across middle of bulb again (fig. 4, 1950) T. contortus 2a. Palpal features not as given above {curvus, morosus, productus) 3 3. Palp: tibia deeply excavate retrolaterally and distally; with a strongly cliitinized apophysis of moderate length ventral to excavation; embolus makes a complete circle distal to anterior margin of bulb, then passes retrolaterally to terminate in a finely dentate tip (fig. 7, 1950) . . T. curvus 3a. \\ ithout palpal features as given above {tnorosus, productus) 4 4. Palp: ventral, retrolateral, tibial apophysis strongly chitinized and dis- tall}" knobbed; dorsal apophysis a sharply pointed spine; near anterior margin of Ijulb the embolus turns toward base, then passes to retro- lateral side and extends nearly to tip of cymbium as a long, slender filament (figs. 13-14, 1950) T. morosus 4a. Palp: with a pair of long, slender, retrolateral, tibial apophyses almost meeting distally, with a hook attached near base of the shorter apo- physis; deeply grooved embolus makes a loop near distal margin of bulb, 342 bulletin: museum of comparative zoology then a second loop near base of bulb, continues as a fine filament to distal end of cymbium (fig. 25, 1950) T. produdus 5. Palp : species with at least the dorsal retrolateral, tibial apophysis elongated and well developed (aculeatus, cretatus, decorus, hurnphreyi, parki, pere- grinus, probus, studiosus, vitusus) 6 5a. Palp: species with retrolateral tibial apophyses less well developed; either with but one apophysis or with two shorter and poorly developed apophyses {ineptus, intentus, innotus, mundulus, obsecus, pauper, sigil- latus) 14 6. Palp : with a retrolateral, dorsal, tibial apophysis extending nearly to tip of bulb {aculeatus, decorus) 7 6a. Palp: with a retrolateral, dorsal, tibial apophysis extending at most little beyond middle of bulb {cretatus, hurnphreyi, parki, peregrinus, probus, studiosus, vitusus) 8 7. Palp: with the dorsal, retrolateral, tibial apophysis distally bifurcated (fig. 1, 1950) T. aculeatus 7a. Palp: with the dorsal, retrolateral, tibial apophysis simple at tip, not bifurcated (Figs. 7-8) T. decorus 8. Palp: with the dorsal, retrolateral, tibial apophysis subdivided near its base into a short, pointed process and a long, pointed extension (figs. 29-30, 1950) T. studiosus 8a. Palp: without any basal division of the dorsal, retrolateral, tibial apophy- sis {cretatus, hurnphreyi, parki, peregrinus, probus, vitusus) 9 9. Palp: with the dorsal, retrolateral, tibial apophysis definitely serrated along its external edge near middle (fig. 23, 1950) T. probus 9a. Palp: without any serration along external edge of dorsal, retrolateral, tibial apophysis {cretatus, hutnphrei/i, parki, peregrinus, vitusus) 10 10. Palp: with the ventral, retrolateral, tibial apophysis a relatively massive structure, deeply notched at distal end (Fig. 10) T. hurnphreyi 10a. Palp: with the ventral, retrolateral, tibial apophysis relatively smaller and unnotohed at its distal end {cretatus, parki, peregrinus, vitusus). .11 11. Palp: dorsal, retrolateral, tibial apophj-sis with two nearly right angles in its total length (fig. 20, 1950) T. parki 11a. Palp: with donsal, retrolateral, tibial apophysis not provided with right- angled bends {cretatus, peregrinus, vitusus) 12 12. Palp: dorsal, retrolateral, tibial apophysis flask -like in shape (fig. 22, 1950) T. peregrinus 12a. Palp: dorsal, retrolateral, tibial apophysis not flask-like in shape {cretatus, vitusus) 13 13. Palp: dorsal, retrolateral, tibial apophysis elongated; ventral, tibial apophysis definitely hammer-headed (Figs. 1-2) T. cretatus 13a. Palp: dorsal, retrolateral, tibial apophysi.3 robust, with a very slender terminal spine; ventral, retrolateral apophysis short, broad, slightly in- dented terminally T. vitusus 14. Palp: with a definitely recurved hook at base of tarsal bulb (fig. 28, 1950) T. sigillatus 14a. Palp: tarsal bulb without any definite recurved hook at its base {ineptus, intentus, innotus, mundulus, obsecus, pauper) 15 chickering: tmarus from Panama 343 15. Palp: with a pair of short, pointed processes emerging from a rounded depression at anterior end of bulb (fig. 10, 1950) T. ineptus 15a. Palp: bulb without such structures as given above {intentus, innotus, mundulus, obsecus, pauper) 16 16. Palp: tarsal bulb with a single, bifurcated process emerging from a rounded depression at anterior end (F. P. -Cambridge's fig. 26, table 10) T. intentus 16a. Palp: tarsal bulb without such a structure as given above (innotus, mundulus, obsecus, pauper) 17 17. Palp: ventral, retrolateral, tibial apophysis a short, blunt process; con- siderably longer than very short, dorsal, retrolateral apophysis (figs. 16-17, 1950) T. mundulus 17a. Palp: tibial apophyses not as given above (innotus, obsecus, pauper) . . 18 18. Palp: both tibial apophyses short, bluntly pointed processes (O. P.- Cambridge's fig. 8, table 12) T. pauper 18a. Palp: retrolateral, tibial apophyses unlike those given above (innotus, obsecus) 19 19. Posterior fourth of abdomen sharply constricted but with no discernible tubercle (Fig. 27) T. obsecus 1 9a. Abdomen with no marked constriction; with a small but distinct dorsal tubercle in last quarter T. innotus No satisfactory key has yet been devised to aid in the identi- fication of females of known species from Panama. Tmarus aculeatus Chickering Tmarus aculeatus Chickering, 1950. Bull. Mus. Comp. Zool., 103(4): 217, fig. 1. The male holotype from Barro Colorado Island, C. Z., together with several male paratypes from localities in the Panama Canal Zone and nearby parts of Panama are all in the Museum of Comparative Zoology. Roewer, 1954; Bonnet, 1959. Since 1950 the species has been taken only on Barro Colorado Island, C. Z., Summit, C. Z., and Arraijan, R. P. The female is still unknown. Tmarus bucculentus Chickering Tmarus bucculentus Chickering, 1950, Bull. Mus. Comp. Zool., 103(4): 220, fig. 2. The female holotype from Ft. Randolph, Panama Canal Zone, August, 1936, together with a female paratype, are in the Museum of Com- parative Zoology. Roewer, 1954: Bonnet, 1959. This species, known only from females, has been collected in three different localities in the Canal Zone since my publication in 1950. 344 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Tmarus cognatus Chickering Tmarus cognatus Chickering, 1950, Bull. Mus. Comp. Zool. 103(4): 222, fig. 3. The female holotype and two female paratypes from the Panama Canal Zone Forest Preserve, August, 1939, are in the Museum of Compara- tive Zoology. Roewer, 1954; Bonnet, 1959. This species, also known only from females, has not been taken since the original specimens were collected. Tmarus contortus Chickering Tmarus contortus Chickering, 1950, Bull. Mus. Comp. Zool., 103(4): 224, figs. 4-5. The male holotype, female paratype, one male paratype and three immature specimens, all taken in the Madden Dam region, Panama Canal Zone, August, 1939, are in the Museum of Comparative Zoology. Roewer, 1954; Bonnet, 1959. A very small number of both sexes have been taken since the establishment of the species and all have come from the Canal Zone Forest Preserve. Tmarus corruptus O. P.-Cambridge Tmarus corruptus O. P.-Cambridge, 1892, Arachnida-Araneida, 1: 95, pi. 12 fig. 10. The holotype female from BugaV)a, Panama, is in the British Museum (Natural History). F. P.-Cambridge, 1900; Chickering, 1950; Roewer, 1954; Bonnet, 1959. This species remains poorly known and only from females. The elder P.-Cambridge had the holotype from Bugaba, Panama; F. P.-Cambridge beUeved that he had the species from Mexico. It has not yet appeared in my collections. A female on loan from the British jMuseum (Natural History) does not show the two pairs of minute canals drawn so clearly by F. P.-Cambridge. O. P.-Cambridge's figure 10c, plate 12, is nearer to what I see in the epigynum than what is shown in F. P.-Cambridge's figure 32, plate 10. In view of the uncertainties here and the scarcity of material I am reserving further treatment of the species pending the acquisition of more specimens for a comparative study. Tmarus cretatus sp. nov. Figures 1-6 The specific name is a Latin adjective suggested by the chalky appearance of the body. Male holotype. Total length, including slightly porrect cheli- cerae, 3.84 mm. Carapace 1.49 mm long, 1.56 mm wide opposite chickering: tmarus from Panama 345 interval between second and third coxae where it is widest; 0.47 mm tall and, therefore, less than }/^ as tall as wide. Eyes: viewed from above, posterior row strongly recurved, anterior row gently so; central ocular quadrangle wider behind than in front in ratio of 13 : 10, longer than wide behind in ratio of 15 : 13; posterior row occupies about 9/13 of width of carapace at that level; ratio of eyes AME : ALE : PME : PLE = 5.5 : 12 : 4.5 : 13; AjNIE separated from one another by nearly twice their diameter, from ALE by a little more than twice their diameter; PME sepa- rated from one another by a Uttle less than four times their Tmarus cretatus sp. nov. Fig. 1. Left palp of male, ventral view. Fig. 2. Left palpal tibia, retrolateral view. Fig. 3. Dorsal view of abdomen of male. Fig. 4. Dorsal view of abdomen of female. Fig. 5. Epigynum, from below. Fig. 6. Spermathecae of female paratype, cleared and turned outward to show internal arrangement. 346 bulletin: museum of comparative zoology diameter, from PLE by six times their diameter ; laterals separated by 23^ times the diameter of ALE. Height of clypeus equal to 43^ times the diameter of AAIE. Legs 1234. Palp: essential features shown in Figures 1-2; tibial apophyses quite distinctive. Ab- domen: posterodorsal tubercle, prominent in the female, appears here as only a slight rise. Color in alcohol: general chalky appear- ance important; first two pairs of femora with numerous white flecks on prolateral surfaces; venter of abdomen with a median, longitudinal, light brownish stripe and a narrow, white stripe on each side (in some paratypes the median stripe is nearly black). Female paratype. Total length, including bases of chelicerae and slightly extended posterior spinnerets, 5.84 mm. Carapace 2.08 mm long, 2.21 mm wide opposite interval between second and third coxae where it is widest. Clypeus somewhat more porrect than in male; height equal to five times the diameter of AAIE. Legs 1243. Abdomen: with a prominent posterodorsal tubercle as indicated in Figure 4. Epigynum : essentially as shown in Figures 5-6. Color: essentially as in male; median, ventral, abdominal stripe is dark brown. Type locality. The male holotype is from Summit, Panama Canal Zone, July, 1950 ; the female paratype was taken in the same locality in August, 1950. One male paratype was taken at Summit, C. Z., July, 1950; several female paratypes are in the collection from Summit, C. Z., July and August, 1950, and Summit Gardens, C. Z., August, 1954. Tmarus curvus Chickering Tmarus curvus Chickering, 1950, Bull. Mus. Comp. Zool., 103(4): 228, figs. 7-8. The male holotype and female paratype were taken on Barro Colorado Island, Panama Canal Zone, August, 1936. Paratypes of both sexes were reported from several localities in the Canal Zone and in Panama proper; all specimens are in the Museum of Comparative Zoology. Roewer, 1954; Bonnet, 1959. The species has been collected during two recent collecting periods in the Canal Zone. Tmarus decens O. P.-Cambridge Tmarus decens O. P.-Cambridge, 1892, Arachnida-Araneida, 1: 98, pi. 12, fig. 9. The holotype female is in the British Museum (Natural History). F. P.-Cambridge, 1900; Chickering, 1950; Roewer, 1954; Bonnet, 1959. This species, known only from the female, has not yet appeared in my collections and is known only from Bugaba, Panama. chickering: tmarus from Panama 347 Tmarus decorus sp. nov. Figures 7-8 The specific name is a Latin adjective suggested by the decora- tive pattern on the cephalothorax and dorsal areas of the abdomen. Male holotype. Total length, including somewhat porrect cheli- cerae, to tip of anal tubercle 4.39 mm. Carapace 1.69 mm long; 1.5 mm wide opposite second coxae where it is widest; about 0.68 mm tall and, therefore, less than half as tall as wide. Eyes: viewed from above, posterior row strongly recurved, anterior row sUghtly so; viewed from in front, anterior row almost straight, measured by centers ; central ocular quadrangle wider behind than in front in ratio of about 16 : 11; about as wide behind as long. Ratio of eyes AME : ALE : PME : PLE = 5:11 : 6.5 : 10. AME separated from one another by nearly 2.5 times their diameter, from ALE by three times their diameter. PAIE separated from one another by slightly less than three times their diameter, from PLE by nearly 4.5 times their diameter. Laterals separated by nearly three times the diameter of PLE. Clypeus quite porrect; height equal to about seven times the diameter of AME (only chitinized area measured); apparently bearing only a single, slender spine at each ventrolateral angle. Sternum: sternal suture gently recurved; not extended between fourth coxae which are separated by about three-fourths of their width. Legs: 1243; with numerous long, slender spines. Palp: most essential features shown in Figures 7-8; patella and tibia short; the latter with two conspicuous apophyses. Abdomen: a slight swelhng may indicate a more or less prominent posterodorsal tubercle in the female when the latter is known ; with numerous long, slender spines ; only a little more than twice as long as wide. Color in alcohol: carapace yellowish with red and pink decorations; a narrow red marginal band nearly encircles this part of the body from opposite the interval between LE to posterior border where there is a central gap in the colored margin; within the dorsal area there is an irregular, bright red figure consisting of a series of short radii sur- rounding a yellowish area; the clypeus bears a pair of relatively large, irregular, reddish spots; the chelicerae are yellowish with a central pink spot on each; legs basically yellowish but with many small red or pink dots and larger reddish blotches; fourth leg and, to a lesser extent, the third leg with the reddish dots united into narrow stripes especially on the patellae and tibiae; abdomen with a series of reddish and white transverse bands extending laterally and posteriorly; the venter is yellowish with a dusty brown center. 348 bulletin: museum of comparative zoology Tyye locality. The holotype male is from Barro Colorado Island, Panama Canal Zone, February, 1958. There are no para- types and the female is unknown. Tmarus humphreyi sp. nov. Figures 9-13 For some time the males and females here regarded as repre- senting one species were considered as two. Recently, however, after studying the paratypes of both sexes it was decided that it would be safe to combine them. The species is named in honor of Richard L. Humphrey, M. D., a former student of mine and a delightful companion on an extended collecting trip in Panama. Male holotype. Total length, including somewhat porrect chelicerae, 3.74 mm. Carapace 1.39 mm long, 1.19 mm wide opposite anterior border of first pair of legs; 0.55 mm tall and, therefore, a little less than half as tall as wide. Eyes : viewed from above, posterior row rather strongly recurved, anterior row gently so ; viewed from in front, anterior row nearly straight, measured by centers; central ocular quadrangle wider behind than in front in ratio of 25 : 18, longer than wide behind in ratio of 27 : 25; posterior row occupies nearly four-fifths of width of carapace at that level. Ratio of eyes AME : ALE : PIME : PLE = 5 : 10 : 6.5 : 9. AME separated from one another by twice their diam- eter, from ALE by a slightly greater distance. PAIE separated from one another by slightly more than twice their diameter, from PLE by about 3.5 times their diameter. Laterals separated by about twice the diameter of PLE. Clypeus: quite porrect; with seven long, slender spines along ventral border; height, including membranous border, equal to nearly six times the diameter of AME. Chelicerae and maxillae essentially typical of the genus. Sternum: bluntly terminated posteriorly and only slightly ex- tended between fourth coxae which are separated by about two- thirds of their width; otherwise essentially typical of the genus. Legs: 1243. Palp: most essential features shown in Figures 9-11. Abdomen: almost squarely truncate at anterior dorsal border; a small posterodorsal tubercle is curved downward; dorsum with numerous slender spines each arising from a small, basal tubercle. Color in alcohol: carapace reddish brown with many irregular, dark, oblique stripes along lateral sides; legs yellowish with fine, black dotting; sternum brownish with darker spots extending from coxae toward the center; abdomen a dull black dorsally with a narrow, central, yellowish stripe extending through the anterior half; lateral sides of abdomen with alternating, irregularly narrow, chickering: tmarus from Panama 349 dark and light stripes; venter with a broad, nearly black, central stripe and a narrow, irregular light stripe on each side. Female paratype. Total length 5 mm, including bases of some- what porrect chelicerae; carapace 1.76 mm long, 1.45 mm wide opposite second coxae, 0.84 mm tall and, therefore, a Uttle more than half as tall as wdde, with the usual supply of long, slender spines. Eyes: central ocular quadrangle wider behind than in front in ratio of 15 : 11; about as long as wide behind; ratio of eyes Tmarus decorus sp. nov. Fig. 7. Tibia and tarsus of left male palp, ventral view. Fig. 8. Tibia and tarsus of left male palp, retrolateral view. Tmarus humphreyi sp. nov. Fig. 9. Tibia and tarsus of left male palp, ventral view. Fig. 10. Tibia and tarsus of left male palp, retrolateral view. Fig. 11. Tibia and tarsus of left male palp, dorsal view. Fig. 12. Posterior end of abdomen of female. Fig. 13. Epigynum from below. 350 bulletin: museum of comparative zoology AAIE : ALE : PME : PLE = 5 : 10 : 6 : 8.5; AME separated from one another by 2.6 times their diameter, from ALE by nearly three times their diameter; PAIE separated from one another by sUghtly less than three times their diameter, from PLE by nearly 14/3 of their diameter; laterals separated by 2.2 times the diameter of ALE. Clypeus: strongly porrect; height, including membranous ventral border, equal to slightly more than six times the diameter of AAIE. Legs: 1243. Abdomen: with a short posterodorsal tubercle (Fig. 12); with the usual supply of slender spines each arising from a minute tubercle. Epigynum: quite distinctive; with a short, median scape; essentials shown in Figure 13; dissection reveals a pair of conspicuously coiled spermathecae not seen in external appearance. Color in alcohol : generally much lighter than in male ; first and second legs yellowish with many reddish brown, irregular spots; first and second femora with prolateral surfaces conspicuously reddish brown with many fine, black dots each associated with a stiff bristle or spine; third and fourth legs much lighter; abdomen yellowish brown dorsally with lighter spots and narrow stripes, with alternating light and darker lateral stripes, ventrally with a broad, dark brown, median stripe. Type locality. Alale holotype and female paratype from Gam- boa, Panama Canal Zone, July, 1954. Paratypes of both sexes collected at the following localities in the Canal Zone: Summit, August, 1950; Summit Gardens, August and July, 1954; Gamboa, July and August, 1954 and January, 1958; Corozal, July, 1954 and December, 1957; Balboa, i\lay, 1964. Tmarus impedus sp. nov. Figures 14-15 The name of the species is an arbitrary combination of letters. Female holotype. Total length, including the somewhat porrect chelicerae, 6.11 mm. Carapace 2.41 mm long; 2.34 mm wide opposite posterior border of second coxae where it is widest; 0.72 mm tall and, therefore, only one-third as tall as wide; ventral margin regularly rounded from opposite PLE to posterior border; other features as usual in females of the genus from Panama. Eyes: viewed from above, posterior row moderately recurved, anterior row slightly so; viewed from in front, anterior row nearly straight, measured by centers; central ocular quadrangle wider behind than in front in ratio of 4 : 3, wider behind than long in ratio of about 10 : 9. Ratio of eyes AAIE : ALE : PAIE : PLE = 5 : 12 : 7 : 12. AME separated from one another by 3.6 times chickering: tmarus from Panama 351 their diameter, from ALE by nearly four times their diameter; PME separated from one another by nearly four times their diameter, from PLE by nearly five times their diameter; laterals separated by slightly more than 2.5 times their diameter. Clypeus quite porrect; height, including membranous ventral border, nearly equal to 8.5 times the diameter of AME; with the usual slender spines. Chehcerae, maxillae, and lip, all essentially typical of females of the genus from Panama. Sternum: scutiform; moderately convex; longer than wide in ratio of 5 : 4; widest be- tween second coxae; with many bristles; not extended between Tmarus impedus sp. nov. Fig. 14. Abdomen of female, lateral view, right side. Fig. 15. Epigynum, from below. Tmarus innotus sp. nov. Fig. 16. Tibia and tarsus of left male palp, ventral view. Fig. 17. Tibia and tarsus of left male palp, retrolateral view. 352 bulletin: museum of comparative zoology fourth coxae which are separated by slightly more than half their width. Legs: 1243; with numerous spines. Abdomen: very spiny; anterior, dorsal border slightly raised; slightly anterior to middle of dorsum a pair of small, conical tubercles arise, each topped by a fairly robust spine; a little behind the middle of the dorsum 's a low tubercle bearing numerous spines (Fig. 14). Epigynum: some- what distinctive (Fig. 15). Color in alcohol: the carapace has a triangular area which is yellow or whitish with small, irregular, Hght, reddish brown dots and spots extending from PLE to the top of the posterior decUvity; from the top of the dechvity a series of radiating, irregular, whitish lines extend downward over a yellowish brown background; legs yellowish with many reddish spots; prolateral surfaces of first two pairs of femora mottled with white spots; mouth parts yellowish; sternum hght yellowish; abdomen with dorsum and dorsolateral sides a conglomerate of reddish spots together with irregular, whitish spots, streaks, and numerous black dots ; the pair of small dorsal tubercles are largely black; venter with a median, light brownish stripe flanked on each side by an irregularly dotted white area. It seems probable that there would be much variation in the color pattern of a large population of the species. Type locality. The holotype female is from Summit, Panama Canal Zone, August, 1950. No paratypes have appeared in my collection and the male is unknown. Tmarus ineptus 0. P.-Cambridge Tmarus ineptus O. P.-Cambridge, 1892, Arachnida-Araneida, 1: 94, pi. 13, fig. 3. The holotype female is in the British Museum (Natural History). F. P.-Cambridge, 1900; Chickering, 1950; Roewer, 1954; Bonnet, 1959. The specimens which I have assigned to this species appear to represent the most abundant of all of the species now known from Panama. I have it from many localities in the Panama Canal Zone and in Panama proper. The P.-Cambridges had the species only from Bugaba, Panama. A comparison of the figures published by the two P.-Cambridges, together with an examination of the single female on loan from the British INIuseum (Natural History) and the specimens in my own collection, poses serious questions regard- ing my previous treatment of the species. The two P.-Cambridges represented the epigynum quite differently in their respective illustrations. It now seems quite possible that it will be necessary to recognize the species in my collection as a new species. Until I have an opportunity to study all of the specimens now in the chickering: tmarus from Panama 353 P.-Cambridge collection and, if possible, to collect in the vicinity of Bugaba I do not think it advisable to take a more positive stand in regard to the matter. Tmarus innotus sp. nov. Figures 16-17 The name of the species is an arbitrary combination of letters. Male holotype. Total length, including somewhat porrect cheli- cerae, 2.8 mm. Carapace 1.1 mm long; 1.08 mm wide; 0.54 mm tall. Eyes: viewed from above, posterior row quite strongly re- curved, anterior row moderately so; viewed from in front, anterior row slightly procurved, measured by centers; central ocular quadrangle wider behind than in front in ratio of 26 : 17, wider behind than long in ratio of 13 : 12; posterior row occupies about seven -tenths of width of carapace. Ratio of eyes AME : ALE : PME : PLE = 4:8:5:8. AME separated from one another by 2.5 times their diameter, from ALE by 2.75 times their diameter; PAIE separated from one another by 3 times their diameter, from PLE by slightly more than this; laterals separated by 2.25 times their diameter. Clypeus: wdth a long, slender spine on each side beneath interval separating AAIE from ALE; height equal to 4 times the diameter of AAIE (membranous ventral border not in- cluded). CheHcerae, maxillae, lip and sternum all essentially typical of males of the genus from Panama. Legs: 1243. Palp: exposed parts much simpler than usual in the genus from Panama ; tibial apophyses unlike others seen by me ; distinguishing features shown in Figures 16-17. Abdomen: a little more than tw^ce as long as wide; posterior end somewhat extended; with a small but distinct dorsal tubercle probably indicating that the female has a well developed posterodorsal tubercle extending above the spinnerets. Color in alcohol: lateral sides of carapace a mottled, medium reddish brown; clypeus and ocular region yellowish with numerous light brown spots; dorsal area of carapace with a some- what triangular, yellowish area with apex extending just over the top of the posterior declivity and including brownish spots of different sizes and form ; mouth parts and sternum yellowish with variations. First and second femora with irregular, whitish spots on prolateral surfaces; first and second metatarsi reddish. Abdomen with an irregular whitish stripe in middorsal region; remainder of dorsum and lateral regions yellowish with minute reddish dots at bases of numerous spines; venter yellowish white. 354 bulletin: museum of comparative zoology Type locality. Male holotype from Barro Colorado Island, Panama Canal Zone, July 30, 1954. There are no paratypes and the female is unknown. Tmarus innumus sp. nov. Figures 18-20 The name of the species is an arbitrary combination of letters. Female holotype. Total length, including sUghtly porrect cheli- cerae, 9.95 mm. Carapace 3.25 mm long, 2.89 mm wide, 1.37 mm tall. Eyes : viewed from above, posterior row moderately recurved, anterior row slightly so; viewed from in front, anterior row nearly straight, measured by centers; central ocular quadrangle wider behind than in front in ratio of about 4:3; wider behind than long in the same ratio. Ratio of eyes AME : ALE : PME : PLE = 8 : 16 : 9.5 : 14. AME separated from one another by four times Tmarus ituiumus sp. nov. Fig. 18. Abdomen of female, dorsal view. Fig. 19. Posterior end of abdomen of female, lateral view, right side. Fig. 20. Epigynum, seen from below. Tmarus levii sp. nov. Fig. 21. Epigynum, seen from below. chickering: tmarus from Panama 355 their diameter; from ALE by slightly less than this. PME sepa- rated from one another by about five times their diameter; from PLE by nearly six times their diameter. Laterals separated by twice the diameter of ALE. Clypeus: moderately porrect; height, including membranous ventral border, equal to nearly nine times the diameter of AME. Chelicerae, maxillae, and lip all essentially typical of females of the genus from Panama. Sternum : typical of females of the genus from Panama; not extended between fourth coxae which are separated by two-fifths of their width. Legs : 2143 ; spines on first and second metatarsi shorter, more robust, and more numerous than usual in females from Panama. Epigynum some- what distinctive as shown in Figure 20. Carapace brownish with three pairs of narrow stripes radiating over lateral sides ; posterior decli\'ity yellowish at top but dark brown elsewhere ; a small brown circular area surrounds each PAIE. Sternum light brown. Legs yellowish with many black or gray dots and spots ; first and second femora conspicuously dotted with black on prolateral and ventral surfaces; first patellae nearly black ventrally. Abdomen con- spicuously chalky white dorsally with irregular, blackish bars and spots in posterior half; lateral sides irregularly white with brown spots and irregular oblique rows of black dots and short bars; venter with a light brown, median stripe flanked on each side by a row of brownish dots. Type locality. Female holotype from Summit Gardens, Panama Canal Zone, July 19, 1954; one female paratype from Summit, C. Z., July 21, 1950. Epigynum of paratype slightly different from that of the holotype but not significantly so in my judgment. The male is unknown. Tmarus intentus 0. P.-Cambridge Tmarus intentus O. P.-Cambridge, 1892, Arachnida-Araneida, 1: 96, pi. 13, fig. 1. The holotype male from Guatemala and the female from Bugaba, Panama, are in the British Museum (Natural History). F. P.-Cambridge, 1900; Chickering, 1950; Roewer, 1954; Bonnet, 1959. This appears to be a rare species in regions where I have col- lected. It has not appeared in my collections since 1936 when I took two females on Barro Colorado Island in July. Some uncer- tainties appear to exist in respect to this species, however. Two females on loan from the British Museum (Natural History) differ quite markedly in general appearance although the epigyna are closely similar. I hope to be able to study all specimens regarded by the P.-Cambridges as belonging to this species and, if possible, to collect in regions from which the species has been reported. 356 bulletin: museum of comparative zoology Tmarus levii sp. nov. Figure 21 This species is named after Dr. Herbert W. Levi, Associate Curator of Arachnology, Aluseum of Comparative Zoology, Harvard University. Female holotype. Total length, including bases of chelicerae, 6.37 mm. Carapace 2.4 mm long; 2.05 mm wide; about 0.98 mm tall; without a distinct thoracic groove; with numerous minute tubercles each with a long, slender spine. Eyes : viewed from above, posterior row moderately recurved, anterior row only slightly so; viewed from in front, anterior row slightly procurved, measured by centers. Central ocular quadrangle wider behind than in front in ratio of 25 : 18; wider behind than long in nearly the same ratio. Ratio of eyes AAIE : ALE : PAIE : PLE = 7 : 10.5 : 6.5 : 10. AME separated from one another by about 3.5 times their di- ameter, from ALE by about three times their diameter. PAIE separated from one another by nearly six times their diameter, from PLE by slightly less. Laterals separated by about three times the diameter of PLE. Clypeus : with a row of six spines along ventral border and with numerous bristles turned toward the middle line; height equal to nearly eight times the diameter of AME, including ventral membranous border. Chelicerae, maxil- lae, and lip all essentially typical of females of the genus from Panama. Sternum not extended between fourth coxae which are separated by slightly less than one-third of their width. Legs: 1243; spines essentially as usual in females of the genus from Panama; trichobothria observed on tibiae, metatarsi, tarsi and also on palpal tibiae. Abdomen: anterior border with a row of slender spines; dorsal surface with many slender spines, all ap- parently erectile. Distinctive features of epigynum shown in Figure 21. Carapace yellowish with light brown stripes radiating irregularly from region just above beginning of posterior declivity together with numerous brownish streaks and spots; posterior declivity light brownish with nearly white, irregular border. Clypeus yellowish with many reddish brown dots. Sternum, lip and maxillae all yellowish. Legs yellowish with many reddish dots at bases of spines; first and second femora conspicuously dotted with red on a white background along prolateral surfaces. Ab- domen: dorsum w^hitish with many fine, dark dots and many larger, i-eddish dots at bases of spines; venter whitish along a broad, central stripe and darker laterally with conspicuous, narrow grooves and semitransparent dots. chickering: tmarus from Panama 357 Type locality. Female holotype from Barro Colorado Island, Panama Canal Zone, June, 1950. There are no paratypes and the male is unknown. Tmarus longus sp. nov. Figures 22-23 The name of the species is a Latin adjective meaning extended, suggested by the conspicuous posterior extension of the abdomen. Female holotype. Total length 5.33, including somewhat porrect chelicerae. Carapace 1.67 mm long; 1.36 mm wide; 0.64 mm tall. Eyes: viewed from above, posterior row moderately recurved, anterior row slightly so; viewed from in front, anterior row slightly recurved, measured by centers; central ocular quadrangle wider behind than in front in ratio of 5 : 3, wider behind than long in ratio of 15 : 14; posterior row occupies about two-thirds of width of carapace at that level. Ratio of eyes AME : ALE : PME : PLE = 4 : 11 : 5.5 : 10. AME separated from one another by slightly less than three times their diameter, from ALE by slightly less than four times their diameter; PME separated from one another by nearly 3.5 times their diameter, from PLE by nearly five times their diameter; laterals separated by nearly twice the diameter of ALE. Clypeus: moderately porrect; with a row of seven spines near ventral margin with the second from each end being very long ; height equal to about 7.5 times the diameter of AME. Chelicerae, maxillae, and lip essentially typical of females of the genus from Panama. Sternum quite convex; with many long, stiff bristles; not Tmarus longus sp. nov. Fig. 22. Posterior end of abdomen, lateral view, right side. Fig. 23. Epigynum, seen from below. 358 bulletin: museum of comparative zoology extended between fourth coxae which are separated by five- thirteenths of their width. Legs: 1243; first two pairs of legs unusually long ; spines are few in number and less conspicuous than usual in the genus from Panama. Abdomen long and slender; posterior end conspicuously extended (Fig. 22) ; with many spines. Epigynum complicated ; with some parts quite obscure ; unlike any other seen by the author (Fig. 23). Carapace with a somewhat triangular, yellowish area extending from PLE to top of posterior declivity which has a narrow white boundary and an incomplete, white, longitudinal line through the middle; dorsolateral sides light brownish with white lines and streaks extending ventrally in somewhat radiate fashion; a fairly broad, white, longitudinal band extends along lateral sides ; ocular area whitish with a pair of short, yellowish, longitudinal stripes between PAIE. Palps and chelicerae a mixture of yellowish and white; other mouth parts yellowish. First two pairs of legs whitish with brown dots and larger spots, the latter concentrated along prolateral surfaces of first two pairs of femora; third and fourth legs yellowish with few reddish brown dots and larger spots. Abdomen: with a fairly well defined folium bounded laterally by an irregular, narrow, brown stripe extending through about three-fourths of length of abdomen; lateral and ventrolateral sides yellowish white with several narrow, longi- tudinal folds; venter yellowish with a narrow, irregular, whitish, ventrolateral border on each side. Type locality. Holotype female from Panama Canal Zone Forest Preserve, Panama Canal Zone, July, 1954; one immature paratype taken in the same locality, January, 1958; the male is unknown. Tmarus morosus Chickering Tmarus jnorosus Chickering, 1950, Bull. Mus. Comp. Zool., 103(4): 235, figs. 13-15. The male holotype and the female paratype are in the Museum of Comparative Zoology. Roewer, 1954: Bonnet, 1959. The holotype is from Barro Colorado Island, Panama Canal Zone; the female paratype is from the Aladden Dam region, C. Z.; paratypes are in the collection from Barro Colorado Island, C. Z. Forest Preserve, Summit Gardens, Fort Sherman, and Chilibre, all in the Canal Zone. Tmarus mundulus 0. P.-Cambridge Tmarus mundulus O. P.-Cambridge, 1892, Arachnida-Araneida, 1: 95, pi. 12, fig. 11. The holotype female from Bugaba, Panama, is in the British Museum (Natural History). F. P.-Cambridge, 1900; Chickering, 1950; Roewer, 1954; Bonnet, 1959. chickering: tmarus from Panama 359 The P.-Cambridges had only females. What are believed to be both sexes have appeared in my collection in large numbers from many localities in the Canal Zone. Tmarus obsecus sp. nov. Figures 24-29 In spite of some lingering uncertainties it is regarded as fairly safe to match the sexes as they are presented below. The name of the species is an arbitrary combination of letters. Male holohjpe. Total length, including bases of chelicerae, 4.49 mm. Carapace: 1.54 mm long; 1.32 mm wide; 0.57 mm tall. Eyes : viewed from above, posterior row rather strongly recurved, anterior row moderately so, not including strongly convex lenses; viewed from in front, anterior row very slightly procurved, meas- ured by centers. Central ocular quadrangle wider behind than in front in ratio of 30 : 21 ; wider behind than long in ratio of 15 : 13. PLE extend somewhat beyond lateral margins of carapace at their level. Ratio of eyes AAIE : ALE : FME : PLE = 6:11 : 5 : 9. AA'IE separated from one another by 11/6 of their diameter, from ALE by 7/3 of their diameter. PAIE separated from one another by four times their diameter, from PLE by slightly more than five times their diameter. Height of clypeus, including membranous ventral border, equal to 3.5 times the diameter of AME. Cheli- cerae, maxillae and lip typical of males of the genus in Panama. Sternum only slightly convex; posterior end bluntly pointed but not extended between fourth coxae which are separated by nearly one half their width. Legs: 1243. Palp: distinctive features shown in Figures 24-26; both tibial apophyses short, blunt. Abdomen: elongated; lateral sides nearly parallel throughout two-thirds of length and then sharply narrowed; a sUght widening just before constriction is taken to indicate a tendency to inflate as in the female. Carapace with steep posterior declivity a fairly uniform brown; radiating from top of posterior declivity are several irregular, yellowish and brown stripes covering most of dorsal surface and extending laterally to make a complicated color pat- tern ; clypeus a dark brown ; anterior surface of chelicerae brownish with yellowish stripes. Sternum brown with elongated darker spots extending toward center from margin. Legs generally light brown with many irregular, reddish and yellowish spots; first femora with a dark brown stripe along prolateral surface; second femora with a similar stripe but much broken into separate spots; third and fourth legs much lighter. Abdomen: dorsum brownish 360 bulletin: museum of comparative zoology in general but this results from a conglomeration of brown streaks and spots on a light yellowish background ; a median lighter streak extends through the entire length of the dorsum ; lateral sides with narrow rows of elongated, brown spots and streaks alternating with yellowish streaks; venter with a broad, brownish, median, longi- tudinal stripe flanked on each side by a narrower yello^^dsh stripe. Tmarus ohseciis sp. nov. Fig. 24. Tibia and tarsus of left male palp, ventral view. Fig. 25. Tibia and tarsus of left male palp, retrolateral view. Fig. 26. Basal, palpal, tarsal apophysis, prolateral view. Fig. 27. Abdomen of male, dorsal view. Fig. 28. Abdomen of female, dorsal view. Fig. 29. Epigynum, seen from below. chickering: tmarus from Panama 361 Female paratype. Total length, including somewhat extended chelicerae, 8.13 mm. Carapace 2.47 mni long; 2.2 mm wide; about 0.98 mm tall. Eyes: viewed from above, posterior row moderately- recurved, anterior row shghtly so; viewed from in front, anterior row straight, measured by centers. Central ocular quadrangle wider behind than in front in ratio of about 3:2; wider behind than long in about the same ratio. Ratio of eyes AME : ALE : PAIE : PLE = 6 : 13 : 8 : 11. AME separated from one another by shghtly more than three times their diameter, from ALE by about 3.5 times their diameter. PAIE separated from one another by slightly more than four times their diameter, from PLE by about 4.75 times their diameter. Laterals separated by nearly twice the diameter of PLE. Clypeus quite porrect; ventral margin with numerous spines (some indicated only by scars) ; height, including membranous ventral border, equal to slightly more than six times the diameter of AME. Chehcerae, maxillae and lip typical of females of the genus from Panama. Sternum moderately convex; not extended between fourth coxae which are separated by a little more than one-sixth of their width. Legs: 1243. Abdomen: inflated laterally in posterior half as shown in Figure 28. Carapace reddish brown with several narrow, yellowish lines and reddish brown bands radiating from top of posterior declivity; lateral sides irregularly spotted with reddish brown on a yellowish back- ground ; steep posterior declivity nearly all reddish brown. Cheli- cerae brownish with a yellowish stripe along median boundary. Lip brown, lighter at distal end. Maxillae yellowish in general but brownish in median halves. Sternum brown. First and second legs brownish in general with irregular yellowish spots; femora of these legs with a broad, brown, prolateral stripe throughout; third and fourth legs much lighter with the femora also with a broad, pro- lateral stripe much hghter in color. Abdomen: dorsum dark colored with many small, red dots and larger reddish spots ; extend- ing through the middle of the anterior half of the dorsum is a narrow, irregular, dark stripe fringed with yellow ; this is continued to the anal tubercle by a broken, median yellow stripe; the inflated region has a broken, yellow bar extending across and down the lateral sides ; lateral sides with a series of narrow, alternating, black and reddish, longitudinal stripes; venter with a fairly broad, me- dian, longitudinal, light brown stripe flanked on each side by a broader, yellow stripe with dark, irregular dots. Type locality. Male holotype from Summit, Panama Canal Zone, August, 1950; female paratype from Barro Colorado Island, C. Z., January, 1958; one immature female paratype from Summit, C. Z., November, 1946 (N. L. H. Kraus). 362 bulletin: museum of comparative zollogy Tmarus parki Chickering Tmarus parki Chickering, 1950, Bull. Mus. Comp. Zool., 103(4): 242, figs. 19-20. The male holotype from Barro Colorado Island, Panama Canal Zone, is in the Museum of Comparative Zoology. Roewer, 1954; Bonnet, 1959. One additional male of this species was taken on Barro Colo- rado Island, C. Z., January, 1958. This specimen has a much brighter color pattern than was shown by the holotype; the cara- pace is brightly colored somewhat like that of T. decorus sp. nov. ; the dorsum of the abdomen has each spine arising from a red dot surrounded by a yellowish ring. The female is unknown. Tmarus pauper O. P.-Cambridge Tmarus pauper O. P.-Cambridge, 1892, Arachnida-Araneida, 1: 96, pi. 12, fig. 8. The male holotype is in the British Museum (Natural History). F. P.-Cambridge, 1900; Chickering, 1950; Roewer, 1954; Bonnet, 1959. The P.-Cambridges had only the male from Bugaba, Panama. The species has not yet appeared in my collections and, so far as I know, has not been reported since the original collections were made. Tmarus peregrinus Chickering Tmarus peregrinus Chickering, 1950, Bull. Mus. Comp. Zool., 103(4): 245, figs. 21-22. The male holotype is in the Museum of Comparative Zoology. Roewer, 1954; Bonnet, 1959. The male holotype was taken on Barro Colorado Island, Panama Canal Zone, July, 1934 and has not appeared in my collections since that time. The female is unknown. Tmarus probus Chickering Tmarus probus Chickering, 1950, Bull. Mus. Comp. Zool., 103(4): 247, figs. 23-24. The male holotype together with several paratype males from several different localities in the Panama Canal Zone and Panama proper are in the Museum of Comparative Zoology. Roewer, 1954; Bonnet, 1959. The species has been collected on four different dates since the publication of my paper in 1950, and all within the Canal Zone. Dr. James Zetek contributed a specimen taken in El Cermeno, Panama, January, 1940. CHICKERING : TMARUS FROM PANAMA 363 Tmarus productus Chickering Tmarus productus Chickering, 1950, Bull. Mus. Comp. ZooL, 103(4): 249, figs. 25-26. The male holotype from Porto Bello, Panama, is in the Museum of Comparative Zoology. Roewer, 1954; Bonnet, 1959. The species has appeared in my collections only twice in recent years: Summit, July, 1950, and Summit Gardens, July, 1954, Panama Canal Zone. It seems possible that the female T. impedus sp. nov. is the missing female for this species. Tmarus protobius sp. nov. Figure 30 The name for this species is an arbitrary combination of letters. Female holotype. Total length 5.46 mm, including somewhat porrect chehcerae. Carapace 1 .65 mm long ; 1 .36 mm wide ; 0.77 mm tall. Eyes: viewed from above, posterior row rather strongly re- curved, anterior row moderately so, exclusive of the very convex lenses of ALE; viewed from in front, anterior row gently pro- curved, measured by centers. Central ocular quadrangle wider behind than in front in ratio of 29 : 19; slightly wider than long. Ratio of eyes AME : ALE : PME : PLE = 4.5 : 11.5 : 7 : 10.5. AME separated from one another by slightly more than twice their diameter, from ALE by a little more than three times their diameter. PAIE separated from one another by about 2.5 times their diameter, from PLE by a little more than 3.5 times their diameter. Laterals separated by twice the diameter of PLE. Clypeus: moderately porrect; height, including membranous ventral border, nearly six times the diameter of AME. Chehcerae, maxillae, and lip essentially typical of females of the genus from Panama. Sternum: moderately convex; posterior end sharply pointed, not extended between fourth coxae which are separated by five-thirteenths of their width. Legs: 1243. Abdomen : with a well defined posterodorsal tubercle. Epigynum: somewhat distinctive (Fig. 30) ; quite unlike that seen in any other species known to me. Carapace in general hght yellowish but with irregular, brownish lines radiating from a central light colored region out over dorso- lateral sides ; posterior decUvity light brownish in center and light yellowish on lateral sides. Chehcerae: each with a narrow, light yellowish, lateral stripe and with anterior surface a mixture of irregular light lines and light brownish spots. Other mouth parts yellowish. Sternum: yellowish with an irregular row of red dots along each lateral region. Legs: in general yellowish with many 364 bulletin: museum of comparative zoology reddish dots, white spots and streaks ; first femora with an irregu- lar, prolateral, brown stripe with red dots and streaks along margins; second femora with a whitish, prolateral stripe and many red dots; ventral surfaces of segments from patellae to tarsi are similarly but less conspicuously colored. Abdomen : with a dorsal, light brownish folium bordered by narrow, white lines ; lateral sides whitish with fine dots arranged in irregular, longitudinal lines; ventrolaterally there is a series of light brownish, irregular, narrow stripes; venter hght yellowish with a pair of dotted Unes close to midventral region and a single dotted line on each side near the lateral border. Type locality. Holotype female from Barro Colorado Island, Panama Canal Zone, August, 1950. One paratype female from Summit Gardens, C. Z., July, 1954. The male is unknown. Tmarus rubinus sp. nov. Figures 31-32 The name of the species is an arbitrary combination of letters. Female holotype. Total length, including somewhat porrect chelicerae, 6.11 mm. Carapace 1.96 mm long; 1.69 mm wide; about 0.98 mm tall. Eyes: viewed from above, posterior row moderately recurved, anterior row gently so; viewed from in front, anterior row probably straight (raised cuticle makes it difficult to measure accurately). Central ocular quadrangle wider behind than in front in ratio of about 3:2; wider behind than long in ratio of about 4 : 3. Posterior row of eyes occupies about five-sixths of width of carapace. Ratio of eyes AAIE : ALE : PAIE : PLE = 5.5 : 12.5 : 7.5 : 10. AAIE separated from one another by slightly more than three times their diameter, from ALE by about the same distance. PAIE separated from one another by slightly less than four times their diameter, from PLE by about the same distance. Laterals separated by 1.8 times the diameter of PLE. Clypeus moderately porrect; height equal to about six times the diameter of AME. Chelicerae, maxillae, and lip essentially typical of females of the genus from Panama. Sternum: only slightly convex; not extended between fourth coxae which are separated only by about one-fifth of their width. Legs: 1 = 234. Abdomen: with posterodorsal tubercle quite conspicuously developed (Fig. 31); otherwise essentially typical of females of the genus from Panama. Epigynum: obscurely distinctive; essentials shown in Figure 32. Carapace light brownish in general with whitish spots and streaks ; dorsal radiations, so conspicuous in several species, are here obscure and poorly defined; posterior decUvity with a whitish spot chickering: tmarus from Panama 365 32 Tmarus protobius sp. nov. Fig. 30. Epigynum, seen from below. Tmarus rubinus sp. nov. Fig. 31. Posterior end of abdomen of female, lateral view, right side. Fig. 32. Epigynum, seen from below. at top with remainder a variable brownish. Sternum hght yellow- ish. Legs: yellowish in general with many small, reddish brown spots ; first two pairs of femora conspicuously mottled with reddish brown along the whole prolateral surface; dorsal and lateral sur- faces of abdomen light yellowish with many small, reddish and whitish dots, irregular spots and streaks; venter with a broad, unicolorous, yellowish, median stripe; lateral sides of the median stripe irregularly whitish. Type locality. Female holotype from Summit Gardens, Panama Canal Zone, July, 1954. There are no paratypes and the male is unknown. Tmarus sigillatus Chickering Tmarus sigillatus Chickering, 1950, Bull. Mus. Comp. Zool., 103(4): 252, figs. 27-28. The male holotype from the Panama Canal Zone Forest Pre- serve, C. Z., is in the Museum of Comparative Zoology. Roewer, 1954; Bonnet, 1959. Several males have been added to the collection from Summit, C. Z. Forest Preserve, and Summit Gardens, all in the Canal Zone and all taken during the past fourteen years. The female remains unknown. Tmarus studiosus 0. P.-Cambridge Tmarus sludiosus O. P.-Cambridge, 1892, Arachnida-Araneida, 1:97, pi. 12, fig. 6. Types of both sexes are in the British Museum (Natural History), from Bugaba, Panama. F. P.-Cambridge, 1900; Chickering, 1950; Roewer, 1954; Bonnet, 1959. 366 bulletin: museum of comparative zoology The collection now includes both sexes of this species from many locaUties in the Canal Zone where it appears to be abundant. Tmarus vitusus sp. nov. Figures 33-35 The name of the species is an arbitrary combination of letters. Male holotype. Total length 4.68 mm. Carapace 1.76 mm long; 1.69 mm wide; about 0.68 mm tall. Eyes: viewed from above, posterior row quite strongly recurved, anterior row gently so (exclusive of very convex lenses of ALE) ; viewed from in front, anterior row almost straight, measured by centers. Central ocular quadrangle wider behind than in front in ratio of 10 : 7; wider behind than long in ratio of 5 : 4. Posterior row of eyes occupies nearly full width of carapace. Ratio of eyes A]\IE : ALE : PME : PLE = 5 : 11 : 7 : 10. AME separated from one another by 3.6 times their diameter, from ALE by three times their diameter. PME separated from one another by 3.7 times their diameter, from PLE by shghtly more than 4.4 times their diameter. Laterals 34 Tmarus vitusus sp. nov. Fig. 33. Left palp of male, ventral view. Fig. 34. Left palpal tibia, retrolateral view. Fig. 35. Posterior end of abdomen, lateral view. chickering: tmarus from Panama 367 separated by 2.5 times the diameter of PLE. Clypeus: with a row of six long, slender spines near ventral border and a single similar, medial spine below the interval separating AME. Chelicerae, maxillae, and lip essentially typical of males of the genus from Panama. Sternum moderately convex; squarely terminated half- way between fourth coxae which are separated by slightly more than half their width. Legs: 1243. Palp: essential features shown in Figures 33-34. Abdomen: with a small but definite postero- dorsal tubercle (Fig. 35), probably indicative of a more or less prominent tubercle in the female of the species. Carapace with a nearly triangular area outlined in yellowish enclosing a brownish region just behind PME; just posterior to the apex of this region is a somewhat rounded, yellowish area reaching a short distance down the posterior decUvity; remainder of dorsal region, posterior declivity and lateral sides are a mottled yellowish and brownish color; interocular and clypeal region a dotted, yellowish coloration. Legs yellowish with many brown dots, especially marked on pro- lateral surfaces of femora, patellae, and tibiae of first and second legs. Sternum yellowish with fine, brown, irregular dots. Abdomen : dorsal and lateral sides in general grayish with a darker median streak extending through anterior half and four narrow, darker, cross bars poorly outlined; venter with a rounded, brownish spot in front of genital groove and a median, longitudinal, light brown stripe bordered by very light areas and extending nearly to the spinnerets. Type locality. Male holotype from Gamboa, Panama Canal Zone, July 24, 1954. There are no paratypes and the female is unknown. BIBLIOGRAPHY Banks, Nathan 1898. Arachnida from Baja California and other parts of Mexico. Proc. California Acad. Sci., Ser. 3, Zool., 1(7): 205-309, 5 pis. 1929. Spiders from Panama. Bull. Mus. Comp. Zool., 69: 53-96, 4 pis. Bonnet, Pierre 1959. Bibliographia Araneorum. Toulouse. Vol. 2 (5). Cambridge, O. P.-, and F. P. -Cambridge 1889-1905. Arachnida-Araneida. Vols. I-II. In: Biologia Centrali- Americana. Dulau & Co., London. Chickering, A. M. 1950. The spider genus Tmarus (Thomisidae) in Panama. Bull. Mus. Comp. Zool., 103 (4): 213-255, 4 pis. Keyserling, Graf E. von 1880. Die Spinnen Amerikas. 1. Laterigradae. Baur & Raspe, Nurnberg. 368 bulletin: museum of comparative zoology Petrunkevitch, Alexander 1911. A synonymic index-catalogue of spiders of North, Central, South America, etc. Bull. Amer. Mus. Nat. Hist., 29: 1-809. 1925. Arachnida from Panama. Trans. Conn. Acad. Arts Sci., 27: 51-248. ROEWER, C. Fr. 1954. Katalog der Araneae. Brussels, Vol. 2a: 1-923. (Received 29 June 1965.) INDEX TO SPECIES aculeatus 343 levii 356 bucculentus 343 longus 357 cognatus 344 morosus 358 contortus 344 mundulus 358 corruptus 344 obsecus 359 cretatus 344 parki 362 curvus 346 -pauper 362 decens 346 peregrinus 362 decorus 347 probus 362 humphreyi 348 produdus 363 impedus 350 protobius 363 ineptus 352 rubinus 364 innotus 353 sigillatus 365 innumus 354 studiosus 365 intentus 355 vitusus 366 Bulletin of the Museum of Comparative Zoology HARVARD UNIVERSITY Vol. 133, No. 8 THE RELATIONSHIPS OF FOUR SMALL HISPANIOLAN ELEUTHERODACTYLUS (LEPTODACTYLIDAE) By Albert Schwartz CAMBRIDGE, MASS., U.S.A. PRINTED FOR THE MUSEUM January 20, 1966 Publications Issued by or in Connection WITH THE MUSEUM OF COMPARATIVE ZOOLOGY HARVARD UNIVERSITY Bulletin (octavo) 1863 — The current volume is Vol. 133. Breviora (octavo) 1952 — No. 237 is current. Memoirs (quarto) 1864-1938 — Publication terminated with Vol. 55. Johnsonia (quarto) 1941 — A publication of the Department of Mollusks. Vol. 4, no. 43 is current. Occasional Papers of the Department op Mollusks (octavo) 1945 — Vol. 2, no. 31 is current. Proceedings of the New England Zoological, Club (octavo) 1899-1948 — Published in connection with the Museum. Publication terminated with Vol. 24. The continuing publications are issued at irregular intervals in num- bers which may be purchased separately. Prices and lists may be obtained from the Publications Office of the Museum of Comparative Zoology, Cambridge 38, Massachusetts. Peters' Check-List of Birds of the World. Volume 1 is out of print; volumes 2 to 7, 9, 10 and 15 are currently available. Phylogeny and Evolution of Crustacea. Proceedings of a confer- ence held at Cambridge, Mass., March 6-8, 1962. H. B. Whittington and W. D. I. Rolfe, editors. Cambridge, Mass., 192 pp., 80 figs., 1963. Cloth bound, $6.75. Fishes of the Gulf of Maine, by Henry B. Bigelow and AVilliam C. Schroeder. Washington, viii - 577 pp., 1953. Photo-offset reprint, $6.50. Publications of the Boston Society of Natural History The remaining stock of the scientific periodicals of the Boston Society of Natural History has been transferred to the Museum of Comparative Zoology for distribution. Prices for individual numbers may be had upon request. Bulletin of the Museum of Comparative Zoology HARVARD UNIVERSITY Vol. 133, No. 8 THE RELATIONSHIPS OF FOUR SMALL HISPANIOLAN ELEUTHERODACTYLUS (LEPTODACTYLIDAE) By Albert Schwartz CAMBRIDGE, MASS., U.S.A. PRINTED FOR THE MUSEUM January, 1966 Bull. Mus. Comp. Zool., Harvard Univ., 133(8): 369-399, January, 1966. No. 8 — The Relationships Of Four Small Hispaniolan Eleutherodactylus (Leptodactylidae) By Albert Schwartz 10,000 SW S4th Street Miami, Florida 33143 Perhaps the most confusing group of the genus Eleutherodactylus in the West Indies is that composed of the small to minute frogs of Hispaniola, none of which has a snout-vent length in excess of 25 mm. To this assemblage belong E. minutus Noble 1923, E. abbotti Cochran 1923, E. audanti Cochran 1934, and E. haitianus Barbour 1942 (= E. intermedins Cochran 1941, preoccupied). E. abbotti was described from Laguna, Samana Province, RepubUca Dominicana, E. audanti from Peak la Selle (= Mont la Selle), Dept. de I'Ouest, Haiti. The remaining two forms were described from the interior uplands of the Cordillera Central in La Vega Province (Repiiblica Dominicana), minutus from near Paso Bajito, Jarabacoa-Constanza Trail, and haitianus from Loma Rucilla, 8C00 to 10,0C0 feet. Although minutus and haitianus are still known only from various localities in the Cordillera, abbotti has been reported from many localities in the Repubhca Dominicana and Haiti. E. audanti is here reported for the first time from outside of the Massif de la Selle. Shreve and Wilhams (1963: 320-323) discussed at some length the situation of the species audanti and abbotti in the Port-au-Prince region. Conclusions drawn from my own experience in the field in both Haiti and the Republica Dominicana in 1962 and 1963 differ from theirs, and are drawn in part from a large body of fresh material from critical localities both outside and within the range of their particular study. Through the courtesy and cooperation of the following curators, I have been able to study specimens of this group of frogs : Ernest E. Williams, Museum of Comparative Zoology (MCZ) ; Doris M. Cochran, United States National Museum (USNM) ; and Charles M. Bogert and Margaret Bullitt, American Museum of Natural History (AMNH). Specimens in my own collection are designated Albert Schwartz Field Series (ASFS). In the field in Hispaniola I had the capable assistance of Patricia A. Heinlein and Ronald F. Klinikowski, Dennis R. Paulson, David C. Leber, and Richard Thomas. To all of them I express my sincere thanks for their interest in these small frogs. The illustrations for the present paper are the work of Klinikowski and Leber; they again merit my gratitude for their endeavors. 372 bulletin: museum of comparative zoology ELEUTHERODACTYLUS ABBOTTI AND ELEUTHERO- DACTYLUS AUDANTI ON THE SOUTH ISLAND ^ Although E. abbotti was known from the south island by only two doubtfully identified specimens (from Petionville and Fond- des-Negres, Haiti) at the time of Cochran's monograph on the herpetology of Hispaniola (1941), it is now known from many localities there. Shreve and Williams (1963) in a study of a large number of small frogs from the La Selle region presented the following con- clusions: 1) audanti is a subspecies of abbotti because "of the presence in the lowlands of the Port-au-Prince region and on the southwest peninsula and in Barahona of equivocal specimens which appear to be in various grades and degrees intermediate between audanti and abbotti"; 2) there are no absolute differences between these two species; 3) at intermediate levels north of the Massif de La Selle, such as Furcy, there exist populations of frogs, some of which may be identified as abbotti, others as audanti, and still others intermediate between these two species; and 4) mate- rial from various lowland and highland localities (Peninsula de Barahona, IMorne de Ca3'ette, Petionville, Fond-des-Negres, Thiotte) cannot be referred with certainty to either abbotti or audanti and are considered to be intergrades. An interesting sidelight on Shreve and Williams' study was the discovery of "the presence in the foothills of the La Hotte region ... of frogs much more like abbotti than any others in the area south of the Cul de Sac Plain." This analysis would suggest that abbotti and audanti are conspecific, and that audanti is restricted to the highest peaks of the La Selle, intergrades with abbotti at lower levels, and is replaced by abbotti in the lowlands. However, while collecting at Furcy in the summer of 1962, it quickly became obvious that we were dealing with two types of small frogs. Both are vocally very similar; the call of each is a series of highpitched "tuck" 's, followed by a sharply ascending "wheep" (although a series of "tuck-wheep, tuck-wheep" 's may be interspersed in the normal call series, usually at its end or beginning). The call of the smaller of these two frogs (audanti) was distinctly higher than that of the larger (abbotti), but other- wise the calls were identical in structure. However, these two forms differ markedly from one another in that abbotti has a 1 The terms "south island" and "north island" are used in this paper as, for example, by Wilhams, 1961. SCHWARTZ : SMALL HISPANIOLAN ELEUTHERODACTYLUS 373 grayish green to tan ground color, is without definite leg bars and postanal triangle (Fig. 1, left), and has an all yellow venter, whereas audanti has a much more brightly colored dorsum, most often some shade of reddish brown or buff, with a distinct and clear-cut postanal triangle and leg bars (Fig. 1, right), and a gray belly which is often spotted. Fig. 1. Left: Eleuiherodactylus abbotti, adult female, ASFS X1649, Furcy, 5600', Dept. de I'Ouest, Haiti; snout-vent length 23.0 mm. Right: Eleutherodactylus audanti audanti, adult female, ASFS X2362, 2.4 mi. S Kenscoff, Dept. de I'Ouest, Haiti; snout-vent length 23.3 mm. All the preserved specimens of these frogs collected by others in the Furcy area can be separated into these two categories without difficulty. Naturally, the yellow ventral coloration is no longer present; the distinct leg and postanal markings of audanti remain, however, and these are a ready means of differentiation between the two species. Occasional "intermediates" (i.e., audanti with slightly less clear-cut postanal triangle or abbotti with more definite leg bands) I consider to be within the natural range of variation of each of the two species involved; the Furcy popula- tion is in no way composed of a large number of intergrades with abbotti and audanti at the two extremes. It is, rather, composed of frogs which are readily assignable to either audanti or abbotti on the basis of pattern, with a very occasional specimen of each whose markings slightly resemble those of the other species in degree of clarity. Thus, at least at Furcy, audanti and abbotti appear not as two races of one form, but as two sympatric species. The following table (Table I) is based on a selection of ten 374 bulletin: museum of comparative zoology specimens of each sex of each species (using as the criterion for species the pattern described above) from the south island Haitian uplands. These series include the largest member of each sex in each case, and the type (a female) of audanti, as well as three other adult female paratypes. jMeasurements and ratios indicate the following: in males, abhotti reaches a larger size than audanti (in females, the reverse appears to be true, but this is apparently an artifact of the abhotti sample, since there are female abhotti from medium elevations in the Cordillera Central which have a snout- vent length equal to that of the largest audanti); measurements of femur, tibia, and fourth toe are diagnostic, as is the tibia/ snout-vent ratio. Table P (measurements in milHmeters) Tibia/snout- Snout-vent Tibia Femur vent ratio length aver- aver- aver- (maximum) range age range age range age abhotti d" 19.3 8.8-10.1 9.3 7.7- 8.5 8.0 48.6-55.4 52.2 audanti (d 18.4 7.2- 8.0 7.6 6.1- 7.0 6.6 40.8-46.8 43.8 abhotti^ 23.6 11.1-13.0 11.9 9.7-11.4 10.2 50.4-56.6 53.5 audanti 9 25.3 9.0-lO.S 10.1 8.0- 9.8 9.1 38.6-47.1 43.6 The following additional descriptive notes on the patterns of the two species should be helpful. E. audanti is well illustrated by Cochran (1941: 66). The heavily and distinctly crossbarred limbs and the dark postanal triangle are clearly showTi. The leg bars — one on the crus, one on the pes, and two incomplete bars on the thigh — are usually outlined in preserved material by pale bands, which set the bars off very distinctly from the ground color. There is a single bar on the antebrachium, and another on the wrist, again outUned by pale color. The dorsal pattern may have a middorsal stripe. The major feature of the dorsal pattern is a scapular X, the anterior limbs of which are usually fused to the dark interocular bar; often the area between the interocular bar and the anterior arms of the X is hkewise dark, thus giving a 1 Head measurements, not included in the table, are less clear-cut. In males all these measurements overlap greatly, although abhotti averages consistently higher. In females, abhotti averages larger in head length (8.2 vs. 8.1) and naris to eye (2.4 vs. 2.2), but slightly less in head width (8.4 vs. 8.5) and diameter of tj'mpanum (1.4 vs. 1.5); diameter of eye averages the same in both species (3.0), although the eye of abhotti reaches a larger maximum size. In no head measurements are the females of the two species separable. SCHWARTZ: SMALL HISPANIOLAN ELEUTHERODACTYLUS 375 rather complex occipital and scapular figure. The sides may be dark like the back, or may have a series of about five horizontal bars before the groin. Two fresh specimens have a tan dorsal band from snout to vent, with darker brown sides, the scapular X faint but still present. The ventral dark pigmentation is variable, but always present; there may be stippling, mottling, or even blotching on the throat and belly; the underside of the hindUmbs is always heavily stippled with brown. The pale snout of audanti is a characteristic of the species. E. abbotti, on the other hand, is generally paler than audanti when preserved, and although it has the scapular X which is joined to the interocular bar, the same number of Umb bands as audanti, and even, at times, has the vertical side stripes, none of the features is so bold and diagrammatic as in audanti. The entire dorsal surface is irregularly mottled with darker color, and the individual pattern elements are lost in the general obliterative effect of the dorsal pigmentation. There is no clear-cut postanal triangle, although the postanal area is somewhat darker than the rest of the concealed surfaces. The limb bars are not set off from the ground color by a distinct pale area; in fact, they have been so much invaded by pale color that they are no longer conspicuous. Usually the ventral surfaces are almost immaculate, although there may be some diffuse stippling on the throat, and an occa- sional individual has some belly stippling. Cochran (1941 : 67-68) has described in great detail the palhd dorsal coloration and asymmetrical spotting which occur in some specimens of audanti. Apparently the entire dorsum loses its basic pattern, and a blotchy, irregular, asymmetrical pattern is super- imposed upon the now unicolor back, so that the preserved specimen is pale, variously and irregularly mottled with dark brown. Of the large series from Furcy available, only two indi- viduals show this mottling, and even in them it is apparently in its earliest stages. The back of one still retains some semblance of pattern, but there are large licheniform patches on the hind- limbs. The other is uniform pale pinkish dorsally, but the hind- legs show expansion of the dark pigment from the crural bands and elsewhere on the legs. This irregularly mottled state is more common in specimens from the high La Selle, where large numbers in any given series may be mottled. The faded back and increased mottling is not a condition of age, since some tiny juveniles, as well as adult and subadult males and females all show the mottled state. Invariably, from any single locality, there are both "normal" and mottled individuals in the sample. 376 bulletin: museum of comparative zoology Shreve and Williams (1963: 322) stated that "nothing . . . comparable to the orange or asymmetrically pigmented specimens of audanti occurs in the abbotti populations north of the Cul de Sac." However, there is a single specimen of abbot li from the lowlands near Pimentel in the Repubhca Dominicana (and thus well outside the range of audanti) which clearly shows this con- dition. In addition, there are frogs (for example, one from Furcy, MCZ 33549, one from Savane Zombi, :\1CZ 31953, and two from 10.5 miles south of Cabral, ASFS V71 and V83) which are typical abbotti in size, proportion, and residual pattern, and show no audanti influence in these respects, but which, nevertheless, have the mottled condition of audanti. These could, perhaps, be con- sidered intergrades. However, there are typical abbotti available from the same localities, and, of the three localities, only at Furcy do the two species occur together. Since abbotti, when far outside audanti genetic influence, may manifest this sort of peculiar spotting, I regard these specimens as within the chromatic varia- tion of abbotti. Since audanti and abbotti are members of the auriculatus group, both have external vocal sacs, granular bellies, small patch-Uke vomerine teeth, and enlarged digital discs. I am unable to dis- tinguish the two species on any structural basis ; one possibiUty is that the vomerine teeth of abbotti are slightly more oblique than those of audanti, but this is at best a very subjective character. Like most members of the auriculatus group, both audanti and abbotti lack inguinal glands. E. audanti is known from near the Dominico-Haitian border (Foret des Pins) west to the vicinity of Pic Alacaya (foothills. Massif de la Hotte). It is restricted to the uplands — the lowest locality whose altitude is known and whence audanti has been collected is Peneau, 5000 feet. A possibly lower locality is 2.4 miles south of Kenscoff, but no altitude is available. The highest locality is Mont Cabaio (7000 feet), although specimens are recorded from Alont la Selle, without elevation given (the summit of jNIont la Selle is slightly over 8000 feet). The specimens from Pic Alacaya (MCZ 21551-53) represent the only Haitian audanti outside of the Massif de la Selle ; they do not differ in pattern from La Selle specimens. The adult female of the series has a tibia/ snout-vent ratio of 48.9, slightly greater than that reported above for La Selle frogs. In the La Selle region, abbotti occurs with audanti at elevations up to 5600 feet (Peneau, Furcy, 2.4 miles south of Kenscoff). There are specimens of abbotti from Savane Zombi (4200 feet) but SCHWARTZ : SMALL HISPANIOLAN ELEUTHERODACTYLUS 377 none from Foret des Pins at 5800 feet where audanti has been collected. E. abhotti has also been collected from the lower southern slopes of the La Selle (one specimen, Thiotte, about 3000 feet); from the northern lowlands of the eastern Tiburon Peninsula (one specimen, IMorne de Cayette) ; from the western extremity of the Tiburon, both north (one specimen, Marfranc, 120 feet) and south (eight specimens, Camp Perrin, 1000 feet) of the ^Massif de la Hotte; and from the intermediate southern slopes of this range (five specimens, Carrefour Canon, 500 feet). (See Fig. 5 for distribution of these two species.) A large amount of fresh material from the Peninsula de Bara- hona and the Sierra de Baoruco indicates that abhotti occurs there as well. In this region it has been taken from near sea level (La Cienaga) up to 3700 feet in the Sierra de Baoruco. In fact, abhotti is the dominant small frog in this entire region. The absence of audanti from the Sierra de Baoruco may be more apparent than real. Since audanti in the La Selle has not been taken lower than about 5000 feet, it may well not occur at lower elevations in the Sierra de Baoruco. Most collecting in these mountains has been in the Valle de Polo region, whose elevation is less than that for the lowest record of audanti to the west. In summary, E. abhotti is widespread throughout the south island, occurring from about sea level to elevations of 5600 feet. The species occurs not only in the lowlands, but in the La Hotte- La Selle-Baoruco massif up to moderate elevations. E. audanti, on the other hand, is known only from elevations above 5000 feet in the massifs de la Hotte and la Selle, and in the latter range is extremely abundant at these higher elevations. Its occurrence on the Sierra de Baoruco requires confirmation. ^ Both species occur together at elevations between 5000 and 5600 feet, at least in the INIassif de la Selle. 1 Since the above was written, David C. Leber and Richard Thomas, in the summer of 1964, succeeded in securing E. audanti in the Sierra de Baoruco. Near the Dominico-Haitian border, eleven specimens of audanti were secured between 4 and 11 kilometers northeast of Los Arroyos, Pedernales Province, at elevations between 5600 feet and 7200 feet. In this same general region, nine specimens of E. abbotti were also collected; these are from six localities ranging in elevation from 2200 feet to 5800 feet. The latter high elevation gives an increase of altitudinal overlap between audanti and abhotti of 200 feet in the southern massifs. At one locality (5 km NE Los Arroyos) both species were collected together. These new localities are not included on the map, nor are these specimens included in the computations. 378 bulletin: museum of comparative zoology ELEUTHERODACTYLUS ABBOTTI AND ELEUTHERO- DACTYLUS AUDANTI ON THE NORTH ISLAND E. abhotti has long been known to occur throughout much of the north island. Described from Laguna on the Peninsula de Samana, this frog was subsequently reported from many localities in the Republica Dominicana (Cochran, 1941; Mertens, 1939), from the Dominico-Haitian border east to the Peninsula de Samana and the south shore of the Bahia de Samana. Its dis- tribution in Haiti is poorly known ; it has been reported only from near Limbe (Lynn, 1958), the Citadelle (Cochran, 1941: 61), and Grande Riviere (Shreve and Williams, 1963: 322). The relatively small number of specimens of abhotti in collections prompted Shreve and Williams to suggest that abhotti w^as "nowhere very abundant." On the contrary, abhotti is an extremely abundant frog, somewhat more so in the uplands than in the lowlands. Not only can it be collected at night, when huge choruses make the forest resound, but can often be secured with ease during the day in piles of coconut trash and old, rotting, and very wet piles of cacao husks. In the Cordillera Central abhotti occurs up to elevations of 6000 feet (Loma Vieja; 9.1 miles north of Constanza; 9.3 miles north of Constanza). In the Sierra de Neiba it occurs at elevations as high as 5000 feet (14.5 miles south of Elias Pina). The Cordillera elevation is slightly in excess of the highest known records in the Massif de la Selle, but not strikingly so. The altitude of major abundance in the Cordillera is apparently about 3600-4000 feet, where abhotti forms the largest portion of nocturnal frog choruses in broadleaf gallery forest along rivers in pinewoods. The measurements of three series of abhotti (ten males and ten females each) from southern Haiti (La Selle), the interior Domini- can uplands (Cordillera), and northern Republica Dominicana are tabulated below (Table II) and reveal certain differences among them. (Each series included the largest members of both sexes; in each series most females were gravid.) In coloration and pattern there appear to be no differences among the various populations studied. The typical dorsal ground color varies from gray to some shade of tan or light brown ; there is a dark interocular bar, crossbars on the limbs, a yellowish to whitish-gray belly, and a yellow \'ocal sac. Despite its wide geographic and altitudinal distribution, abhotti has apparently not differentiated into races. SCHWARTZ: SMALL HISPANIOLAN ELEUTHERODACTYLUS 379 Table II. Means and extremes of nine measurements and one ratio in three populations of Ekuthcrodactylus abbotti consisting of the ten largest specimens of each sex from each region. (No races are recognized.) Republica northern Dominicana Republica Haitian uplands uplands Dominicana lOd" lOd" lOcf snout-vent length 17.9 ( 16.4-19.3) 18.5 (17.9-18.9) 17.8 (17.0-18.8) head length 6.5 ( 6.1- 7.0) 6.4 ( 6.0- 6.9) 6.4 ( 6.0- 6.8) head width 6.6 ( 6.0- 7.1) 6.5 ( 6.3- 7.3) 6.6 ( 6.0- 7.4) tympanum 1.3 ( 1.1- 1.4) 1.2 ( 1.0- 1.3) 1.2 ( 1.0- 1.5) eye 2.6 ( 2.5- 2.7) 2.4 ( 2.2- 2.6) 2.7 ( 2.3- 3.0) naris to eye 1.8 ( 1.6- 2.0) 1.9 ( 1.7- 2.0) 1.8 ( 1.7- 2.0) femur 8.0 ( 7.7- 8.5) 7.7 ( 7.1- 8.3) 7.8 ( 7.5- 8.2) tibia 9.3 ( 8.8-10.1) 8.8 ( 8.5- 9.3) 8.8 ( 8.5- 9.3) fourth toe 8.3 ( 7.5- 9.3) 7.6 ( 6.6- 8.1) 7.5 ( 7.0- 7.8) tibia /snout- vent ratio 43.8 ( 40.8-46.8) 47.9 (45.5-51.1) 50.0 (45.7-52.5) 109 109 109 snout-vent length 22.2 ( 20.9-23.6) 22.7 (20.4-25.4) 21.4 (20.9-22.8) head length 8.2 ( 7.4- 9.3) 8.1 ( 7.1- 8.8) 7.8 ( 7.5- 8.4) head width 8.4 ( , 7.9- 9.2) 8.1 ( 6.9- 9.6) 7.9 ( 7.5- 8.7) tympanum 1.4 ; 1.3- 1.6) 1.5 ( 1.2- 1.7) 1.4 ( 1.2- 1.6) eye 3.0 ; 2.8- 3.4) 2.9 ( 2.5- 3.4) 3.0 ( 2.8- 3.2) naris to eye 2.4 < : 2.0- 2.8) 2.3 ( 1.8- 2.6) 2.3 ( 2.1- 2.7) femur 10.2 ; 9.7-11.4) 9.9 ( 8.6-10.8) 9.1 ( 8.3- 9.8) tibia 11.9 ;il. 1-13.0) 11.1 ( 9.7-12.5) 10.5 ( 9.9-11.2) fourth toe 10.3 ; 9.0-11.3) 9.4 ( 7.9-10.6) 8.8 ( 8.2- 9.7) tibia /snout-vent ratio 53.5 (50.4-56.5) 48.9 (45.6-51.3) 49.2 (47.2-51.9) Two samples of small frogs, from the Sierra de Neiba and the Cordillera Central, merit special attention. The series from the Sierra de Neiba consists of nine frogs from three localities ranging in elevation from 4750 feet to 5950 feet; there are eight adult males and one juvenile. The Cordillera series is made up of nine adult male frogs taken while calling ; the elevation for part of this lot is 5000 feet (Valle de Culata). The specimens from 4 miles (7 km) north of Constanza have no recorded elevation, but are from above 5000 feet. These two small series are distinctly different from abbotti and are quite like audanti from the Massif de la Selle. The bold postanal triangle and the conspicuously banded limbs ally these small frogs with audanti; all are from high elevations. The voice of the Cordillera specimens resembled the high-pitched calls 380 bulletin: museum of comparative zoology of audanti more than the lower calls of abhotti. At Valle de Culata these audanti and abhotti were heard calling in the same abandoned field, which was grown up in Pteris and blackberries. The call of audayiti at this locality reminded me very distinctly of that of E. auricvlatvs in Cuba — a long series of telegraphic chcks (= "tucks") with an occasional, almost inaudible "wheep" at the end of the series. The abhotti call at this locality was that typical for the species throughout its range, and was quite obviously different from the call of audanti. It is remarkable that, despite the large number of frogs which have been collected from the Cordillera Central, audanti is pres- ently known only from the Valle de Culata region. This valley is not especially distinctive, being rather small and mostly cut over or burned for pasture. E. audanti is an upland species, with apparently isolated popu- lations; the Sierra de Neiba and Cordillera populations differ not only from one another but also from the La Selle form. For the Sierra de Neiba population I propose the name : Eleutherodactylus audanti notidodes ^ new subspecies Holotijpe. MCZ 43204, an adult male, from 20 km (11.7 miles) southwest Hondo Valle, elevation 5950 ft., Independencia Prov- ince, Repiiblica Dominicana, one of a series taken 11 August 1963 by David C. Leber and Richard Thomas. Original number ASFS V371. Parafypes. ASFS V372-74, AAINH 71990-92, same data as type; ASFS V385, 14.5 km (8.4 miles) SW Hondo Valle, 4750 ft., San Rafael Province, Repiiblica Dominicana, 11 August 1983, R. Thomas; MCZ 43205, 25 km (14.5 miles) S Elias Pina, 5000 ft., San Rafael Province, RepiibHca Dominicana, 17 August 1963, A. Schwartz. Diagnosis. A subspecies of E. audanti characterized (in males; females unknown) by larger size than a. audanti (male a. audanti to 18.4 mm, male a. notidodes to 21.9), longer hind legs, ratio of tibia/snout-vent length higher (46.5 in notidodes, 43.8 in audanti), little or no dark ventral pigmentation, and hind leg crossbars distinct but not prominently set off from ground color by pale outlining. Description of type. An adult male with the following measure- ments (in mm) and ratio : snout-vent length, 21 .9 ; head length, 7.5 ; ' From the Greek noiis, notidis, moisture, wet, and -odes, dweller. SCHWARTZ: SMALL HISPANIOLAN ELEUTHERODACTYLUS 381 head width, 7.4; diameter of tympanum, 1.2; diameter of eye, 2.7; naris to eye, 2.2; femur, 7.9; tibia, 10.1; fourth toe, 8.6; tibia/ snout-vent length, 46.1. Head width equal to head length; snout truncate with nares conspicuous at anterior end of canthus ros- tralis ; diameter of eye greater than distance from naris to anterior corner of eye; interorbital space 2.6, about equal to diameter of eye; diameter of tympanum much less than diameter of eye, distance from tympanum to eye equal to about three-quarters diameter of tympanum. Digital discs present, that of digit three the largest and equal to about two-thirds area of tympanum. Fingers moderate in length, unwebbed, 3-4-2-1 in order of de- creasing length; subarticular tubercles well developed, pale gray. Toes moderate in length, unwebbed, 4-3-5-2-1 in order of decreas- ing length ; subarticular tubercles dark gray and prominent. Heels touch when femora are held at right angles to body axis. Dorsum finely warty or shagreened with a raised median line from snout to above vent. Throat and belly granular; vocal sac present, large, extending posteriorly to between foreUmbs, heavily glandular anteriorly. Inguinal glands absent. Posterior surface of thighs with large juxtaposed rounded granules. Tongue small, oval, entire, free behind, its greatest width about one-half that of floor of mouth. Vomerine teeth in two sharply oblique patches, begin- ning within the median border of the choanae, and separated from the choanae by a distance equal to slightly less than the diameter of a choana, and from each other by a distance equal to the length of one tooth row. Coloration of type in life. Dorsal ground color tan with a darker brown interocular bar and a scapular X; snout, anterior to interocular bar, slightly paler than back ; area between interocular bar and two anterior arms of X suffused with darker brown ; back in general rather uniformly tan, but on the sides this breaks down into a series of about four lateral bars, separated from one another by creamy gray from the belly. Hindlimbs tan with two crossbars on the thigh, neither especially prominent and neither outlined with pale color; a more conspicuous crural cross-bar, vaguely outlined with pale color; and a single cross-bar on the pes. Fore- limbs with one antebrachial bar and a wrist bar. A prominent dark bar from the snout through the eye to the forelimb insertion and covering the upper half of the tympanum. A prominent and very dark brown postanal triangle, extending onto the underside of the thighs. Ventral ground color creamy gray with scattered dark brown stippling, most concentrated on the vocal sac and undersides of the limbs. Iris silvery above. 382 bulletin: museum of comparative zoology Variation. Variation in measurements and ratios is shown in Table III. Structurally all the paratypes resemble the type. The rather widely separated vomerine teeth and the angulated arrange- ment of the two series in relationship to one another are common features. Table III. Means and extremes of three populations of Eleuthero- daciylus audanti. (Females unknown from Sierra de Neiba and Cordillera Central. Sample from Haiti consists of the ten largest specimens of each sex.) E. a. audanti E. a. notidodes E. a. melatrigonum [Haitian uplands] [Sierra de Neiba] [Cordillera Central] lOd" 8cf 9d^ snout-vent length 17.4 (16.2-18.4) 19.9 (17.9-21.9) 18.2 (17.6-20.3) head length 6.1 ( 5.7- 6.4) 7.0 ( 6.5- 7.5) 6.5 ( 6.0- 6.9) head width 6.3 ( 5.8- 6.6) 7.3 ( 6.3- 7.9) 6.6 ( 6.1- 7.3) tympanum 1.1 ( 0.9- 1.3) 1.2 ( 1.0- 1.4) 1.2 ( 0.9- 1.4) eye 2.4 ( 2.2- 2.5) 2.8 ( 2.4- 3.1) 2.6 ( 2.0- 3.0) naris to eye 1.7 ( 1.5- 1.8) 2.2 ( 2.0- 2.4) 1.9 ( 1.6- 2.2) femur 6.6 ( 6.1- 7.0) 7.7 ( 6.9- 8.5) 7.2 ( 6.4- 8.0) tibia 7.6 ( 7.2- 8.0) 9.9 ( 8.4-10.1) 8.2 ( 7.8- 8.5) fourth toe 6.6 ( 5.9- 7.1) 8.0 ( 7.2- 8.6) 7.0 ( 6.6- 7.5) tibia/snout-vent ratio 43.8 (40.8-46.8) 109 46.5 (44.7-48.3) 45.0 (40.9-47.4) snout-vent length 23.1 (22.1-25.3) head length 8.1 ( 7.4- 8.6) head width 8.5 ( 7.3- 9.2) tympanum 1.5 ( 1.2- 1.6) eye 3.0 ( 2.7- 3.2) naris to eye 2.2 ( 2.0- 2.4) femur 9.1 ( 8.0- 9.8) tibia 10.1 ( 9.0-10.8) fourth toe 9.2 ( 8.4- 9.8) tibia/snout-vent ratio 43.6 (38.6-47.1) In coloration, the paratypes are quite variable. All but one were tan in life, but only a single specimen has the rather uniform dorsal pigmentation of the type. In all, the leg bars are more prominent than in the type, although never outlined boldly with paler color. Five specimens have a dark tan dorsum, somewhat irregularly pigmented, with a vague pair of pale reversed paren- theses, dark interocular bar, scapular X, and pale snouts. Only one of these normally colored specimens has any indication of ventral mottling, and this is sparse and confined to the sides of SCHWARTZ: SMALL HISPANIOLAN ELEUTHERODACTYLUS 383 the abdomen. All the other individuals have only faint dark stippling vcntrally. Two males have a pinkish orange dorsum with irregular dark brown mottling on the limbs, sides, sacrum, and head — the typical asymmetrical audanti pattern described in detail by Cochran. Their dorsal blotching carries over onto the venter, but is not heavy there. The smallest paratype (snout- vent 11.1 mm) is pale pink, with prominent limb bars, but with no irregular dark mottling. Comparisons. E. a. notidodes requires comparison both with the nominate race and with E. abhotti, which occurs with it. Com- pared with a. audanti, notidodes reaches a larger size, and averages greater in all measurements. There is virtually no overlap of extremes in measurements of head length, eye, naris to eye, femur, tibia, and fourth toe. E. a. notidodes has a longer tibia than does a. audanti, although the overlap of the ratios is great. Both subspecies are much alike in coloration and pattern; both show some individuals with asymmetrically blotched dorsal pigmenta- tion. Both have a dark postanal triangle and both show dark limb bars, although those of notidodes are not outlined in paler color. The ventral pigmentation is much heavier in a. audanti; the notidodes with the heaviest ventral pigmentation is lighter than the most lightly pigmented audanti. Compared with male abhotti from the interior highlands, male notidodes reach a larger size (21.9 vs. 18.9) and are longer snouted (naris to eye 2.2 [2.0-2.4] in notidodes, 1.9 [1.7-2.0] in abbotti). All other measurements show a great deal of overlap, although notidodes averages higher in every measurement except tympanum diameter (1.2 in both species) and femur (7.7 in both species). The tibia/snout-vent length ratio averages greater (47.9) in abbotti than in notidodes (46.5). None of these measurements is so helpful as pattern for differentiating the two species ; by means of the brown postanal triangle and the heavily banded legs, they can be distinguished without difficulty. The vomerine teeth may prove to be useful in differentiating the two subspecies of E. audanti. The teeth in a. audanti, although patch-like, seem to be arranged more horizontally than the series in notidodes; in the latter subspecies the patches appear to be more diagonal, the two series directed toward one another poste- riorly in a broad V. The small series of E. audanti from the Constanza region in the Cordillera Central may be known as: 384 bulletin: museum of comparative zoology Eleutherodactylus audanti melatrigonum ^ new subspecies Holotype. UCZ 43206, from 7 km (4 miles) north of Constanza, La Vega Province, Repiiblica Dominicana, one of a series taken 7 July 1963, by David C. Leber and Richard Thomas. Original number ASFS X8774. Paratypes. ASFS X8773, X8775-77, same data as type; AININH 71993-96, 5.1 miles north of Constanza, Valle de Culata, 5000 ft., La Vega Prov., Republica Dominicana, 8 July 1963, D. C. Leber, R. Thomas. Diagnosis. A subspecies of E. audanti characterized (in males; females unknown) by moderate size (male a. audanti to 18.4, notidodes to 21.9, melatrigonum to 20.3 mm), intermediate length of tibia, hindleg crossbars distinct and variously outlined in paler color, although never so boldly as in a. audanti, and pinkish belly with scattered dark chromatophores. Description of type. An adult male with the following measure- ments and ratio: snout-vent length, 20.3; head length, 6.9; head width, 7.3; diameter of tympanum, 1.4; diameter of eye, 3.0; naris to eye, 2.0; femur, 7.6; tibia, 8.3; fourth toe, 7.3; ratio of tibia/snout-vent length, 40.9. Head width greater than head length; snout truncate with nares conspicuous at anterior end of canthus rostralis; diameter of eye greater than distance from naris to anterior corner of eye ; interorbital space 2.8, slightly less than diameter of eye ; diameter of tympanum much less than diameter of eye, distance from tympanum to eye equal to about three- quarters diameter of tympanum. Digital discs present, that of digit three the largest and equal to about three-quarters area of tympanum. Fingers moderate in length, unwebbed, 3-4-2-1 in order of decreasing length; subarticular tubercles well developed, dark gray. Toes moderate in length, unwebbed, 4-3-5-2-1 in order of decreasing length, subarticular tubercles dark gray and prominent. Heels touch when femora are held at right angles to body axis. Dorsum very finely warty, warts most prominent on upper eyehds and upper surface of hindlimbs ; a faint, fine, raised line from snout to above vent. Throat and belly granular; vocal sac present, large, extending posteriorly to between forelimbs, heavily glandular, higuinal glands absent. Posterior surfaces of thighs with large, rounded, juxtaposed granules. Tongue small, oval, entire, free behind, its greatest width about one-half that of floor of mouth. Vomerine teeth in two oblique patches, beginning ' From the Greek ryielas, black, and irigonon, triangle. SCHWARTZ: SMALL HISPANIOLAN ELEUTHERODACTYLUS 385 well within the median border of the choanae, and separated from the choanae by a distance equal to twice the diameter of a choana, and from each other by a distance equal to slightly more than the length of one tooth row. Coloration of type in life. Dorsal ground color medium tan with a black interocular bar, a black scapular X, its two anterior arms forming with the interocular bar a slightly darker occipital tri- angle ; a dark gray pair of dorsolateral lines and two rather diffuse sacral spots; the two dorsolateral lines broken up posteriorly to form a series of three or four dark dorsolateral spots which approach the vent, forming a dark gray V above it; sides tan, heavily dotted with dark gray. Hindlimbs tan with two faint crossbars on the thigh, neither especially prominent and neither outlined with pale color; a more conspicuous crural crossbar, vaguely outlined with pale color; and a single dark crossbar indicated on the pes. Fore- limbs with one antebrachial and one wrist bar. A prominent dark brown bar from the snout through the eye to the forelimb insertion and covering the upper half of the tympanum. Ventral ground color pinkish with some dark gray stippling, especially on the yellow vocal sac and undersides of limbs. Iris silvery above. Variation. Variations in measurements and ratios are shown in Table III (p. 382). The vomerine series are widely separated and vary in angulation from almost straight to oblique, with the latter the more common condition. Three of the paratypes have median dorsal pale hairlines which were creamy in life; all have some sort of dorsolateral dark mark- ings, even if the dorsal ground color is dark brown, which separate the dorsal color from the lateral dotting or spotting. The inter- ocular bar, scapular X and sacral spots are common features, although the latter is often almost completely obscured by the dark dorsal pigmentation. The black to dark brown postanal triangle is always conspicuous, and the leg and arm bars are like- wise bold, and at times even outlined with pale color. The ventral ground color was pinkish in life, with rather uniform dark brown stipphng which never formed ventral blotches or mottling. No specimen shows any indication of the pale dorsum and asymmet- rical dark patches of the other two subspecies, although presumably this condition occurs. Comparisons. E. a. 7nelatrigonum is intermediate between a. audanti and a. notidodes in all measurements except that of tym- panum diameter. The ratio of tibia/snout-vent length is likewise intermediate. The localities whence a. 7nelatrigonum is known are, of course, not intermediate between those of a. audanti and 386 bulletin: museum of comparative zoology a. notidodes. In both dorsal and ventral pigmentation and pattern, a. melatrigonum resembles a. notidodes more closely than a. audanti. The hindlimb bars of a. melatrigonum are slightly more prominent than those of a. notidodes, and slightly less prominent than those of a. audanti. No a. melatrigonum has the venter blotched, as is usually the case in a. audanti. As noted above, there are no speci- mens of a. melatrigonum available at present showing the pale dorsum with asymmetrical blotching which occurs in both a. audanti and a. notidodes. From E. abbotti, with which E. a. melatrigonum occurs, the latter can be best distinguished by its prominent postanal triangle and more conspicuously banded limbs. Male melatrigonum reach a larger size than do male abbotti. All measurements overlap, at times rather widely; the measurement with least overlap is that of the tibia (8.5-9.3 in upland abbotti, 7.8-8.5 in a. melatrigonum). The ratio of tibia/snout- vent length averages less in a. melatrigonum (45.0) than in abbotti (47.9), although the extremes overlap. Lower ratios are consistently those of a. melatrigonum (40.9-47.4), while higher ratios are those of upland abbotti (45.5-51.1). Remarks. The discovery of two subspecies of E. audanti out- side the La Selle-La Hotte massif in southern Haiti indicates that this species probably occurs throughout the higher mountains of much of the Repiiblica Dominicana. The Sierra de Neiba appears to be a favored haven for high-mountain south island species which have been able to cross the Cul-de-Sac-Valle de Neiba plain, or to cross the intervening strait when the plain was flooded. In addition to E. audanti, E. parabates (which is a north island representative of the south island E. ventrilineatus-E . jugans group) is another species of frog which has distinctly south island affini- ties, and occurs only in the Sierra de Neiba. E. audanti, on the other hand, has been able to extend its range farther north into the Cordillera Central, and in so doing has been able to cross the rather xeric but high Valle de San Juan. It is likely that E. audanti occurs in the central Haitian mountains as well, since these are continuous with the Dominican Sierra de Neiba and Cordillera Central; the Haitian mountains are very poorly known herpe- tologically. It is, of course, possible that audanti is a north island species, having evolved in the uplands of the Cordillera, and thence expanded southward into the Sierra de Neiba, across the Cul-de-Sac into the jMassif de la Selle and thence into the Massif de la Hotte. Neither explanation is better substantiated than the other. SCHWARTZ: SMALL HISPANIOLAN ELEUTHERODACTYLUS 387 15 o'l "o ,^ e f to f^S <0 ^ .s s 0 6q ^ 0 ,.-^ — ' ^3 to +^ G -2i C ^ o QJ 1^ to • " a; ^ 'o >- 0 e,^ c Oj -0 "o 'C cS h to jS 3 ^ S:2 C 1 1 ^ 0- i O to -g 0 0) ^ IT -*:j ol C o X! CO -^ to 0 -« if s § 0 ^ 'S °. '^ c3 -1^ fci 02" c 0) -j3 w CO ^ e a; K (^ fc. 03 S s-- ^ (M' d tn -0 ^. fi^ i-. . "U •C &q S 388 bulletin: museum of comparative zoology ELEUTHERODACTYLUS MINUTUS AND ELEUTHERO- DACTYLUS HAITIANUS These two species present an even knottier problem than do E. ahhotti and E. audanti. Regarded by Cochran as two distinct species, minutus and haitianus are both known only from the Cordillera Central in the Republica Dominicana. Both seem to be small forms, with haitianus the smaller (but see further comments below). Shreve and WiUiams (1963: 322) regarded haitianus as a synonym of the prior minutus, and the latter as possibly an upland subspecies of ahhotti. Cochran (1941 : 26) considered that minutus had a smooth belly (and was in fact following Noble, 1923: 4, in his original diagnosis of minutus) and haitianus (= intermedius, sensu Cochran) had a granular belly. Her figures (pp. 47, 70) show that rnimdus is a rather long-legged species, reaching a known snout- vent length of 18 mm, and that haitianus is distinctly short- legged, reaching a known snout-vent length (in the type) of 21 mm. Cochran also commented in the text (1941: 48) that the "para- type" of minutus which she examined had a slightly granular belly. Shreve and Williams (1963: 323) later noted that the venter of minutus is coarsely granular, in contrast to the more finely granular venters of ahhotti and audanti. Perhaps the best way to unravel this complicated situation is to discuss the field situation, based on my recent collections only, and then turn to the nomenclatorial problems. In the Cordillera Central, primarily to the north of Constanza and between that town and El Rio, there occurs a small frog with a maximum snout- vent length of 17.0 mm in males and 19.4 mm in females. This Eleutherodadylus is thus smaller in both sexes than ahhotti, and males are smaller than male audanti melatrigonum from the same general area. The species occurs from 3600 to 6100 feet, in upland broadleaf forests; males call from herbaceous plants, often terres- trial bromeliads, not more than one foot above the ground surface. The call is a single, rising, high-pitched "wheep," almost a pulse, very faint and insect-like. The dorsal ground color varies from tan to brown, with darker sides, and occasionally there are reddish dorsolateral lines separating the dorsal and lateral colors. The dark postanal triangle is fairly distinct, the legs are banded, but rather inconspicuously so at times, and there are remnants of a scapular X. All dorsal pattern elements have a distinctly "muddy" appearance, with no feature being especially clear-cut or prominent (Fig. 3, left). Ventrally, the ground color varies from very pale yellow or cream to gray, and the belly may be either immaculate or have some dark gray dots scattered over it. In addition to the SCHWARTZ: SMALL HISPANIOLAN ELEUTHERODACTYLUS 389 scapular X, there may be a sacral chevron or smudge; the fore- limbs are usually somewhat reddish orange. To the southeast of Constanza, centering in the high Valle Nuevo region, there occurs a small frog with a maximum snout- vent length of 14.8 mm in males and 16.6 mm in females. This species occurs from 5600 to 8000 feet, mainly in pine woods, where it has been found under rocks and logs, and under sheets of moss along a road cut. One calling male was taken about ten inches above the ground on a herbaceous leaf near a small pool in the woods. The call is a descending scale of staccato "beeps." The dorsal ground color varies from tan to very dark brown, almost black in many cases; the brachium is pale reddish, and the ankle often has a reddish brown hue. The dorsal pattern consists of a Fig. 3. Left: Eleutherodactylus minutus, adult female, ASFS X8938, 16 km N Constanza, 6000', La Vega Province, Repiiblica Dominicana; snout-vent length 19.4 mm. Right: Eleutherodactylus haitianus, adult male, ASFS X8392, 9 km NNE Valle Nuevo, La Vega Province, Republica Dominicana; snout- vent length 13.9 mm. band usually outlined in dark brown to black, occasionally with a median pale hairline. A common variant is the "dead leaf" pattern — a series of obliterative pale and dark dorsal areas which render the frog inconspicuous against a varicolored brown back- ground. The snout is usually pale and sharply set off from the balance of the back. A series of four dark lateral bars, radiating from the sacrum, is commonplace (Fig. 3, right). The ventral ground color varies from pale yellow (especially on the vocal sac 390 bulletin: museum of comparative zoology in males) to clear white (not gray), and the belly and throat are very heavily spotted with rather large dark brown to black spots in almost all specimens. All dorsal pattern elements are sharp and distinct, including the hindUmb bands which are usually outhned with tan and stand out boldly from the ground color. There is a dark postanal triangle which may be distinct or inconspicuous, depending upon the intensity of the dorsal ground color. The two forms are almost completely separable on the tibia/ snout-vent length ratio. The larger form from north of Constanza averages 46.0 (44.5-49.7) in males and 47.3 (43.9-53.1) in females, whereas the smaller, more southern form averages 41.3 (37.2-45.0) in males and 40.4 (36.1-43.3) in females. When all aspects of these frogs are considered — the coloration and pattern, morphology, habitat, altitudinal distribution, and vocalization — there seems little doubt that we are dealing with two distinct species. The nomenclatorial problems involved in allocating names to them are somewhat more complex. I have examined the type of E. minutus, but not that of haitianus; the latter is, however, well illustrated by Cochran and some pertinent measurements are given. I have had access to the large series of paratypes of haitianus and these, although helpful, are so confusing in many ways that they require special discussion. The type locality of E. minutus is "near Paso Bajito, Jarabacoa- Constanza Trail"; Paso Bajito is to the north of Constanza. The elevation is not excessively high; we estimate it at about 4000 feet. The tj^pe of minutus (AMNH 11404) is a gravid female with a snout-vent length of 17.3 mm and a tibia length of 8.0 mm. The frog is presently much faded, but dorsally there is a pale zone with a faint scapular X and a faint sacral chevron, distinctly darker sides, only the vaguest indications of leg bars and an interocular bar, and a pale belly with very slightly darker spotting. Of the two species discussed above, the type of minutus agrees best with the larger form, which occurs to the north of Constanza, in pattern and size. Additionally, the tibia/snout-vent length ratio in the type (46.2) falls within the extremes of this ratio in the larger of the two species (females, 43.9-53.1), and not within this ratio in the smaller (females, 36.1-43.3). One other factor needs consider- ation: Noble (1923: 4) described the belly of minutus as smooth. Examination of the iypQ of minutus shows that the belly is, in fact, rather faintly granular; in this, again, it agrees with the larger of the two species discussed above. (Both the species under discus- sion have granular belhes; that of the smaller species is more SCHWARTZ: SMALL HISPANIOLAN ELEUTHERODACTYLUS 391 coarsely and heavily granular than that of the larger.) In con- sideration of the above data, I have no hesitancy in assigning the larger species of frog, described in detail above, to minutus. There remains the possibility, suggested by Shreve and Williams, that minutus is an upland race of abhotti. Against this suggestion is the fact that the two, although similar in pattern, are quite different in intensity of dorsal pigmentation and in size (largest upland abhotti male 18.9, female 25.4; largest minutus male 17.0, female 19.4). Vocally they are distinct. Finally, they occur together at two of our recent localities (9.1 miles north of Con- stanza; 3.3 miles east of El Rio), and very close together at another (9.3 miles north of Constanza), where abhotti was taken at 6000 feet, minutus at 5600 feet. There is no evidence of intermsdiate specimens, although admittedly intergrades might be extremely difficult to differentiate from the parent populations. Certainly, however, when minutus and abhotti are collected at the same locality, there is no difficulty in distinguishing one from the other. The tibia/snout-vent length ratio of the two overlaps in both sexes. The means are, however, higher in abhotti, averaging 47.9 (45.0-51.1) in males, 48.9 (45.6-51.3) in females; male minutus average 46.0 (44.5-49.7), females 47.2 (43.9-53.1). Turning now to the southern small frogs from the Valle Nuevo region, it would seem quite logical to assume that these specimens are haitianus. The type locality, Loma Rucilla, lies about 23.3 miles to the northwest of Valle Nuevo, and the elevation of the type and Loma Rucilla paratypes is expressly stated as 8000 to 10,000 feet. There are three "lots of paratypes: USNAI 107567, 107569-74, .MCZ 23469-74 (17 specimens), Loma Rucilla; USNM 107575-76, Loma Vieja; USNAI 107578-85, AICZ 23495-500 (27 specimens), Valle Nuevo. The two paratypes from Loma Vieja are clearly minutus in size, pattern, and tibia/snout- vent length ratio, and need not be further considered. Ten of the 35 ''haitianus" from Valle Nuevo are also minutus; if they are actu- ally from Valle Nuevo itself, they represent a new altitude record for minutus of 7600 feet. However, the paratypic series is labeled as coming from 6000 to 8000 feet, so that the minutus may have come from within the known altitudinal limits of that species. Twenty-four of the Valle Nuevo paratypes are like the recent Valle Nuevo material discussed above, and are remarkable only in that the series contains females (up to 18.0 snout-vent) and males (up to 15.8) which are larger than the more recently taken specimens. The remaining paratype, a male, has a broad, pale middorsal stripe, a feature not observed in the fresh specimens. 392 bulletin: museum of comparative zoology The twenty-four paratopotypes from Loma Rucilla are extremely puzzling. The type specimen was recorded by Cochran (1941 : 71) to have a snout-vent length of 21, and thus exceeds all Valle Nuevo specimens in size. Among the paratypes are large females with snout-vent lengths from 18.0 to 19.7, bridging the gap in size between the largest female from Valle Nuevo and the holotj^pe. One of the six males (USNM 107572) is clearly E. minutus (thus apparently raising the upper altitudinal limit of that species to at least 8000 feet). Of the five remaining males, four are small and the fifth very large (snout-vent 17.5), in fact larger than any other male. Loma Rucilla frogs lack the heavily spotted venters of the Valle Nuevo frogs, and have throats which are dark with some paler flecking. The larger frogs have unmarked venters and dark throats, the smaller ones spotted venters; the latter group approxi- mates the recently collected material from Valle Nuevo, although the frogs are larger and have less ventral spotting. Cochran's description of the type, "ventral surfaces . . . clouded with minute darker dots," her drawing of the specimen, and its size agree with the larger Loma Rucilla frogs in detail. The nomenclatorial problem resolves itself into how many forms (species or subspecies, if anj^) are involved. It seems rather unu- sual, for example, that there should be such a large difference in size in adult females between Valle Nuevo and Loma Rucilla. None of the female Valle Nuevo specimens is equal in size to the type or to the larger of the female paratypes from Loma Rucilla. The difference in size, correlated with a difference in ventral pig- mentation, noted among the Loma Rucilla specimens, strongly suggests that we maj' be dealing with two species of frogs (the larger species, E. haitianus). Aside from the ventral pigmentation and size, I am unable to differentiate these two "forms" from one another. The tibia/snout-vent length ratio of the larger specimens falls neatly within the extremes of this ratio in specimens from Valle Nuevo. The teeth appear identical, and there are no external characteristics which differentiate them when preserved. Literest- ingly, a single gravid female (MCZ 40813) from Paraje La Cienaga, Manabao, in La Vega Province, has a snout-vent length of only 14.7, and is clearly much more like the Valle Nuevo frogs than the larger Loma Rucilla frogs. La Cienaga hes about 7 miles east of the peak of Loma Rucilla. This single small female suggests very strongly that the larger Loma Rucilla specimens actually are specifically distinct from the more southern frogs. SCHWARTZ: SMALL HISPANIOLAN ELEUTHERODACTYLUS 393 Another possibility is that there are two subspecies of E. haiti- anus involved, one on Loma Rucilla and the other to the south and east, the Loma Rucilla frogs characterized by larger size and much less heavily pigmented venters. Living specimens may be very distinct, since many colors are evanescent in Eleutherodadylus. The least hkely solution is that the presumed differences in ventral pigmentation and size are happenstance artifacts of collecting; the large number of specimens from both Valle Nuevo and Loma Rucilla makes this very unlikely. Without further data in life on the Loma Rucilla populations, there seems only one course open, that of regarding all these frogs as E. haitianus, while acknowledging rather important differences between the northern and southern populations. 70' -20* 18'- 9 9 tfi io 40 Km. 70' L_ Fig. 4. Central Republica Dominicana, showing localities for E. minutus (open circles) and E. haitianus (solid circles). Semi-solid circles represent localities whence both species have been taken. 394 bulletin: museum of comparative zoology There is no question that E. haitianus is distinct from E. minutus. Aside from the habitat and vocal differences discussed above, the two species are easily differentiated on the basis of the tibia/snout- vent ratio; this and other meristic characters are shown in Table IV. That the two species apparently occur together is based entirely on old material with possibly less accurate and carefully taken data than we have for more recently collected specimens; in our experience they do not occur at precisely the same localities. This is also true for the upper altitudinal limits of minutus; all the higher elevations are based on old material. Until the older data can be confirmed, it seems appropriate to regard minutus as having a lower altitudinal distribution than haitianus, although there is an overlap of 500 feet, according to freshly taken and carefully documented material (see Fig. 4 for distribution). There is a need for rather precise notation of elevation in the Cordillera Central. A hike covering several thousand feet elevation may well encompass the altitudinal limits of several species of frogs; the issue may become quickly clouded if all specimens are labeled as coming from between the two extremes in elevation without due regard for more precise altitude. Table IV. INIeans and extremes of E. minutus and E. haitianus. For haitianus the ten largest of each sex from the Valle Nuevo region were used; for minutus all adults of each sex. (Tibia/snout- vent ratio computed for all specimens of each sex from Valle Nuevo, regardless of maturity.) E. minutus E. haitianus S& lOd" snout-vent length 16.4 ( 15.5-17.0) 13.5 (12.5-14.8) head length 5.8 ( 5.5- 6.1) 4.7 ( 4.4- 5.0) head width 5.6 ( 5.3- 6.0) 4.6 ( 4.2- 5.0) tympanum 1.0 ( . 0.9- 1.3) 0.9 ( 0.8- 1.3) eye 2.2 ( , 2.1- 2.4) 1.9 ( 1.7- 2.0) naris to eye 1.4 ( , 1.2- 1.5) 1.1 ( 1.0- 1.3) femur 6.6 ( 6.3- 7.1) 5.1 ( 4.8- 5.6) tibia 7.5 ( 7.3- 7.7) 5.5 ( 5.0- 5.7) fourth toe 6.5 ( , 5.8- 6.8) 4.9 ( 4.5- 5.6) tibia/snout-vent ratio 46.0 ( 44.5-49.7) 41.3 (37.2-45.0) 79 109 snout-vent length 17.9 ( 16.0-19.4) 14.8 (12.0-16.6) head length 6.4 ( 5.9- 6.9) 5.1 ( 4.7- 5.8) head width 6.1 ( 5.5- 6.6) 5.0 ( 4.2- 5.6) tympanum 1.1 : 1.0- 1.4) 1.1 ( 0.7- 1.2) eye 2.4 ( 2.1- 2.6) 1.9 ( 1.7- 2.2) naris to eye 1.5 ( 1.3- 1.8) 1.3 ( 1.1- 1.5) femur 7.6 ( 6.9- 8.0) 5.6 ( 4.7- 6.0) tibia 8.5 ( 7.9- 8.8) 5.9 ( 5.5- 6.5) fourth toe 7.2 ( , 5.7- 7.7) 5.2 ( 4.8- 5.7) tibia/snout-vent ratio 47.3 ;43.9-53.1) 40.4 (36.1-43.3) SCHWARTZ: SMALL HISPANIOLAN ELEUTHERODACTYLUS 395 E. haitianus has not been collected with E. audanti. Our col- lections also indicate that haitianus and abbotti do not occur together, although the altitudinal ranges of the two overlap by 400 feet. There are specimens of haitianus and abbotti from Loma Rucilla, the latter from an elevation of 6000 feet. The two species can be separated by the much shorter tibia of haitianus and the larger size of abbotti. The tibia/snout-vent length ratios of these two species in the Cordillera uplands are: male haitianus 41.3 (37.2-45.0), male abbotti 47.9 (45.5-51.1), female haitianus 40.4 (36.1-43.3), female abbotti 48.9 (45.6-51.3). There are four other small Eleutherodactylus specimens from the Cordillera Central which are of interest. These are two males and one female from 11 km (6.4 miles) east of Paso Bajito, 4500 feet (ASFS X8839-41), and a female from Valle Nuevo (AICZ 23498). The two males are in some ways very Uke E. abbotti, except that they are distinctly longer-legged than any male montane abbotti (tibia/snout- vent length 55.9 and 54.1). They also appear to be more broad headed. Both have leg bars of the abbotti style, and an inconspicuous postanal triangle Uke abbotti. They differ from abbotti in ventral pigmentation, since both have a series of dark spots along the lower jaw, and additionally one has some dark ventral dotting. Of the females, one is gravid and has a stiout- vent length of 18.1, which is rather small for gravid abbotti. These two frogs, from two widely separated localities (AICZ 23498 is a paratype of haitianus), are very similar dorsally in that the pattern consists of a pale snout, a very dark chestnut triangle from the interocular bar onto the sacrum where it meets, apex to apex, with another triangle which has its base across the groin. The ground color is a dull orange-tan. There is a prominent postanal triangle, but no other hindlimb markings except dark ankles. Ventrally, these frogs have some faint stippling on the throat and a series of dark spots along the lower jaw margin. Whether these two females are correctly associated with the two males above is unknown. At least the Paso Bajito female has a tibia/snout- vent length ratio of 48.1, a figure which is included within the known range of Cordillera abbotti. The Valle Nuevo female, on the other hand, has a ratio of 40.0, which is far below all Cordillera abbotti females, and within the extremes of female haitianus, a species with which it is definitely not associated. I have not assigned any of the above four specimens to any species. They may represent one or two new species of small Cordilleran frogs, or they may be aberrantly long-legged or pigmented individuals of well-known upland species. 396 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY RELATIONSHIPS OF SMALL HISPANIOLAN FROGS Of the four species discussed in the present paper, three are clearly members of the auriculatus group — i.e., E. abbotti, E. aud- anti, and E. minutus. As in other members of the group, the presence of a granular venter, enlarged digital discs, short patch- Uke vomerine series, and an external vocal sac indicate their affinities. E. haitianus probably should hkewise be considered a member of this assemblage. It differs from the others in having much smaller digital discs and a somewhat more squatty habitus. All four species lack inguinal glands; all four have the peritoneal covering of the testes completely pigmented with jet black chroma- tophores, while the peritoneal covering of the ovaries has scattered black to gray chromatophores. This dark gonadal pigmentation is a phenomenon which occurs sporadically in Antillean Eleuthero- dactylus, without apparent regard for the affinities of the species. For instance, it occurs in the dimidiatus group {jugans, parabates, ventrilineatus) , in the ricordi group {zugi), as well as in the auricu- latus group. There is a possibility that E. haitianus should be assigned to the varleyi group, in which are presently included (Shreve and Williams, 1963: 339): varleyi, glandulijeroides, and cubanus. E. glanduliferoides and varleyi, in addition to having the varleyi group characters of feebly developed discs, short vomerine series, and small size, also have prominent inguinal and popliteal glands; Shreve and Williams commented that varleyi did not possess these structures, but they are prominent and orange colored in freshly collected material. I do not know if cubanus possesses glands. E. varleyi has a pectoral vocal sac; the condition of the vocal sac is unknown in glanduliferoides, and cubanus apparently lacks a vocal sac. E. cubanus and glanduliferoides are smooth ventrally; varleyi was diagnosed as having a granular belly, although the specimens before me appear to have smooth venters. E. haitianus resembles the varleyi group members in small size, pattern, and short vomerine series. However, the venter is coarsely granular, there are no inguinal or femoral glands, and there is a vocal sac. Black testes are not found in the species varleyi, at least. On the basis of all characters, I prefer to regard haitianus as a member of the auriculatus group, somewhat more divergent than its near relatives in the Cordillera. Of the four species involved in the present discussion, abbotti and audanti are more closely related to one another than to minutus and haitianus. Likewise, minutus and haitianus are more closely related, with minutus more hke the abbotti-audanti pair. SCHWARTZ: SMALL HISPANIOLAN ELEUTHERODACTYLUS 397 As noted before, audanti is presumably an upland derivative of ahbotti (which is widespread throughout the island) ; audanti has either evolved in the Massif de la Selle and migrated thence to the La Hotte, on the one hand, and to the Sierra de Neiba- Cordillera Central, on the other, or has evolved in the Cordillera and has moved thence south into the La Selle. More mesic con- ditions in the lowlands would have faciUtated such movement. Both minutus and haitianus, occurring, as far as is known, only in the Cordillera, may well represent a sequential series abbotti- minutus-haitianus (if audanti was not developed from abbotti in the Cordillera but is an immigrant there), or abbotti-audanti- minutus-haitianus (if audanti developed in the Cordillera and is an immigrant to the La Selle). Specimens examined Eleutherodadylus abbotti: Haiti, Dept. du Sud, ASFS X2797, X2917-22, Camp Perrin; ASFS X3347-51, Carrefour Canon, 500'; MCZ 37729, Marfranc; Dept. de VOuest, ASFS X1363-80, X2006- 08, X2011, Peneau, 5000'; ASFS X1649, X1795-812, X1818, X1899, i\ICZ 33546-50, 33552-60, 33562-63, 33565, 33568-73, 33576-79, 33581, 33586-87, 34212, 34221, 34223, 34226, 34229, 34231-32, 34234, 34242-46, 34249, 34250-53, 34256, 34258, 34261-62, 34264, 34272, 34275-76, 34278-81, 34283-84, 34287-89, 34291-92, 34295, 34301-02, 34304-05, + 16 unnumbered speci- mens, 31729, 31730-32, 31797, Furcy, 5600'; ASFS X3868-69, 2.4 mi. S Kenscoff; MCZ 33280, Morne de Cayette; MCZ 36742, Thiotte; I\ICZ 31952-53, Savane Zombi, Foret des Pins; Dept. du Nord, MCZ 3100, 3526, Grande Riviere du Nord. Republica Dominicana, Barahona Prov., ASFS X9642-43, 3.3 mi. NE La Cienaga; ASFS X9791-94, 0.6 mi. N Las Auyamas, 3000'; ASFS X9809-15, 1.8 mi. N Las Auyamas, 3400'; ASFS X9914-15, 8 km NELas Auyamas, 2600'; ASFSX9919, V71-83, 10.5 mi. S Cabral, 3500'; ASFS V152, 24 km SW Barahona, 3700'; MCZ 35779-90, del Monte's finca, nr. Barahona; MCZ 35791-96, Herrmann's finca, nr. Paraiso; Dajabon Prov., ASFS V1623-24, 12 km S Loma de Cabrera, 2400'; Valverde Prov., ASFS V1241, 9 km N La Cruz de Guayacanes, 1600'; Puerto Plata Prov., ASFS V1689, 8 km E Imbert, 1100'; I\ICZ 23545-46, 25 km S Puerto Plata; Santiago Prov., MCZ 23451-55 (8 specimens), Pico Diego de Ocampo; Espaillat Prov., ASFS V1698, 6 km SE Sabaneta de Yasica; ASFS V1955, 2 km N Puesto Grande, 2200'; Maria Trinidad Sanchez Prov., ASFS V1S60, 2 km S El Factor; Samand Prov., ASFS 398 bulletin: museum of comparative zoology V1914, 11 km E Sanchez; MCZ 23530-31, Sanchez; ASFS V1976- 82, 5 km W Samana; Duarte Prov., ASFS V1823-24, 9 km NW Pimentel; El Seiho Prov., ASFS X7835-36, X7975-81, 3.5 mi. S Sabana de la Mar; ASFS X7902-07, 3.3 mi. SW Miches; ASFS X9267, 2.3 mi. SE Miches; ASFS X9337, 1.4 mi. SE Miches; La Vega Prov., ASFS X8564-601, X8880-84, 4 km SW EI Rio, 4000'; ASFS X8116, 11.5 mi. E El Rio, 3800'; ASFS X9162-63, 23 km E El Rio, 3050'; ASFS X9197-98, X9225-32, X9240, 6 km E El Rio, 3600'; ASFS V1735, 14 km SW La Vega, 1600'; ASFS V1792, 4 km NW La Vega; ASFS V2021, 12 km NE Jarabacoa, 2000'; MCZ 40812, Paraje La Cienaga, IManabao, Municipio Jarabacoa; MCZ 31129, 40815-18, Constanza; ASFS X8249-54, 1 mi. S Constanza, 4000'; ASFS X8244-48, 7.2 mi. S Constanza, 5000'; ASFS X9085, 11.5 mi. SE Constanza, 5800'; MCZ 23520-21, Loma Vieja, 6000'; ASFS X8754-64, 5.1 mi. N Constanza, Valle de Culata, 5000'; MCZ 40811, La Cienaga, Culata; MCZ 30588, Aserradero Bermiidez, Constanza; ASFS X9796, 9.1 mi. N Con- stanza, 6000'; ASFS X8949, 16 km N Constanza, 6000'; ASFS X8826, X8829-30, 6 km W. Constanza, 4250'; ASFS X8892, X8897, Tireo Abajo; MCZ 40806, 43458-65, La Raima, Con- stanza; MCZ 23481-82, Loma Rucilla; ASFS X81 26-27, 1.2 mi. SE Monsenor Nouel, 700'; San Rafael Prov., MCZ 31170-71, 40814, Rancho de la Guardia; ASFS V380-84, 14.5 km SW Hondo Valle, 4750'; ASFS V532, V536, 25 km S Elias Pifia, 5000'; ASFS V537, 19 km S Elias Pina, 4000'; ASFS V543-46, 15 km S Elias Pina, 3400'; San Juan Prov., ASFS V393, 7 km W Vallejuelo, 2600'. Eleutherodactylus audanti audanti: Haiti, Dipt, du Sud, MCZ 21551-53, foothills, IMassif de la Hotte; Dept. de VOuest, MCZ 34208-11, 34213-20, 34222, 34224-25, 34227-28, 34230, 34233, 34235-41, 34247-48, 34254-55, 34257, 34259-60, 34263, 34265-71, 34273-74, 34277, 34282, 34285-86, 34290, 34296-300, 34303, 34306-07, + 33 mmumbered specimens, 33551, 33561, 33564, 33566-67, 33574-75, 33580, 33582-85, 37728, ASFS X1813-17, X1819-98,X1900-02, Furcy, 5600'; ASFS X1362,X2009-10, X2012- 13, Peneau, 5000'; ASFS X1313-21, 2.5 mi. S Kenscoff, 5600'; ASFS X2362-64, X3870-72, 2.4 mi. S Kenscoff; MCZ 24280 (5 specimens), USNM 72595-97, Morne Cabaio, 7000'; MCZ 21576-89 + 39 unnumbered specimens, La Visite, La Selle range; MCZ 19704-08, USNM 95111-13, Mont la Selle; USNM 85009, "Morne la Selle"; ASFS X1922, X3920-27, Foret des Pins, 5800'; MCZ 31954-63, Marie Claire, Foret des Pins; MCZ 24586-88, Bois Pin, nr. Marigot. Eleutherodactylus audanti melatrigonum: Republica Dominicana, La Vega Prov., MCZ 43206, ASFS X8773, X8775-77, 7 km N Constanza; AMNH 71993-96, 5.1 mi. N Constanza, 5000'. SCHWARTZ: SMALL HISPANIOLAN ELEUTHERODACTYLUS 399 Elenfherodactylus audanti notidodes: Repuhlica Dominicana, San Rafael Prov., MCZ 43204, ASFS V372-74, AMNH 71990-92, 20 km SW Hondo Valle, 5950'; ASFS V385, 14.5 km SW Hondo Valle, 4750'; MCZ 43205, 25 km S Elias Pina, 5000'. Ele^dherodaciylus haitianus: Repuhlica Dominicana, La Vega Prov., USNM 107567, 107569-71, 107573-74, Loma Rucilla, 4000-10000'; MCZ 23469-74 + 11 untagged specimens, Loma Rucilla; USNM 107578-85, Valle Nuevo, 6000-8000'; MCZ 23499-500 + 14 untagged specimens, 31588-89, Valle Nuevo; MCZ 40813, Paraje La Cienaga, Manabao, Municipio Jarabacoa; ASFS X8294-99, X8465, 9 km NE Valle Nuevo, 7400'; ASFS X8392-94, 9 km NNE Valle Nuevo; ASFS X8339-40, 3 km NNE Valle Nuevo; ASFS X8461-62, 11 km SE Valle Nuevo, 8000'; ASFS X8994, 5.3 mi. SE Valle Nuevo, 8000'; ASFS X9070- 76, 8.4 mi. SE Valle Nuevo, 7900'; ASFS X8676, 8.9 mi. SE Valle Nuevo, 8000'; ASFS X9083, 15 km SE Constanza; ASFS X8929, 16 km SE Constanza, 5600'; ASFS X9153, 11.8 mi. SE Constanza, 5800'. Eleutherodactylus rninutus: Repiiblica Dominicana, La Vega Prov., USN]\I 107572, Loma Rucilla, 4000-7000'; AMNH 11404, MCZ 9338, Paso Bajito; ASFS X9241, 12 km E El Rio, 3600'; ASFS X8713, X8795, 9.1 mi. N Constanza, 6000'; ASFS X8790, X8938-47, 16 km N Constanza, 6000'; ASFS X9145, 12.6 mi. SE Constanza, 6100'; USNM 107575-76, Loma Vieja, 6000'; MCZ 23495-97 + 6 untagged specimens, Valle Nuevo. LITERATURE CITED Cochran, Doris M. 1941. The herpetology of Hispaniola. Bull. U. S. Nat. Mus., 177:i-vii, 1-398, 12 pis., 120 figs. Lynn, W. Gardner 1958. Some amphibians from Haiti and a new subspecies of Eleuthero- dactylus schmidti. Herpetologica, 14(3):153-57. Mertens, Robert 1939. Herpetologische Ergebnisse einer Reise nach der Insel Hispaniola, Westindien. Abh. senckenberg. naturf. Ges., 449:1-84, 10 pis. Noble, G. K. 1923. Six new batrachians from the Dominican RepubUc. Amer. Mus. Novit., No. 61:1-6. Shreve, Benjamin, and Ernest E. Williams 1963. The herpetology of the Port-au-Prince region and Gonave Islands, Haiti. Part II. The frogs. Bull. Mus. Comp. Zool., 129(5) :302-42, 5 pis. Williams, Ernest E. 1961. The evolution and relationships of the Anolis semilineatus group. Breviora, Mus. Comp. Zool., No. 136: 1-7, map. (Received 7 April 1965.) Bulletin of the Museum of Comparative Zoology HARVARD UNIVERSITY Vol. 133, No. 9 TWO NEW FISHES OF THE MYCTOPHID GENUS DIAPHUS FROI\I THE ATLANTIC OCEAN By Basil Nafpaktitis Museum of Comparative Zoology, Harvard University CAMBRIDGE, MASS., U.S.A. FEINTED FOR THE MUSEUM January 20, 1966 Publications Issued by or in Connection WITH THE MUSEUM OF COMPARATIVE ZOOLOGY HARVARD UNIVERSITY Bulletin (octavo) 1863 — The current volume is Vol. 133. Breviora (octavo) 1952 — No. 237 is current. Memoirs (quarto) 1864-1938 — Publication terminated with Vol. 55. Johnsonia (quarto) 1941 — A publication of the Department of Mollusks. Vol. 4, no. 43 is current. Occasional Papers of the Department of Mollusks (octavo) 1945 — Vol. 2, no. 31 is current. Proceedings of the Nev^ England Zoological Club (octavo) 1899-1948 — Published in connection with the Museum. Publication terminated with Vol. 24. The continuing publications are issued at irregular intervals in num- bers which may be purchased separately. Prices and lists may be obtained from the Publications Office of the Museum of Comparative Zoology, Cambridge 38, Massachusetts. Peters' Check-List of Birds of the World. Volume 1 is out of print; volumes 2 to 7, 9, 10 and 15 are currently available. Phylogeny and Evolution of Crustacea. Proceedings of a confer- ence held at Cambridge, Mass., March 6-8, 1962. H. B. Whittington and W. D. I. Rolfe, editors. Cambridge, Mass., 192 pp., 80 figs., 1963. Cloth bound, $6.75. Fishes of the Gulf of Maine, by Henry B. Bigelow and AVilliam C. Schroeder. AVashington, viii - 577 pp., 1953. Photo-offset reprint, $6.50. Publications of the Boston Society of Natural History The remaining stock of the scientific periodicals of the Boston Society of Natural History has been transferred to the Museum of Comparative Zoology for distribution. Prices for individual numbers may be had upon request. Bulletin of the Museum of Comparative Zoology HARVARD UNIVERSITY Vol. 133, No. 9 TWO NEW FISHES OF THE MYCTOPHID GENUS DIAPHUS FROM THE ATLANTIC OCEAN By Basil Nafpaktitis Museum of Comparative Zoology, Harvard University CAMBRIDGE, MASS., U.S.A. PRINTED FOR THE MUSEUM January, 1966 Bull. Mus. Comp. ZooL, Harvard Univ., 133(9): 401-424, January, 1966. No. 9 — Two New Fishes of the Mydophid Genus Diaphus from the Atlantic Ocean By Basil Nafpaktitis INTRODUCTION During the course of a revisionary study of the myctophid genus Diaphus, representatives of two undescribed Atlantic species were found. These two, Diaphus bertelseni and Diaphus lewisi, are described below. Of the few authors who have seriously studied Diaphus, a specialized myctophid offshoot, some thought it necessary to split the complex into various combinations of genera and subgenera (Goode and Bean, 1895; Fraser-Brunner, 1949; Bolin, 1939, 1959a). Others have maintained the vast array of species within the single genus Diaphus (Brauer, 1906; Taning, 1918, 1928, 1932; Parr, 1928, 1929; Kulikova, 1961). I believe that the former ap- proach has been adopted because of inadequate material and the taxonomic complexity of the group. For reasons which will be presented later in this paper, the inclusion of all the sp3cies (except two or, possibly, three) within the single genus Diaphus is followed here. The taxonomic study of the "Diaphus complex" is not simple, but a reasonable approach can be made. Some of the difficulties involved are: 1) the close similarity among several species of the group; 2) the limited number of reliable diagnostic characters; 3) the relatively high degree of intraspecific variation; 4) the frail nature and generally poor state of preservation of these meso- pelagic fishes; 5) the frequent differences between the luminous organs on the anterior part of the head (important diagnostic characters) of mature specimens and juvenile individuals of the same species; and 6) the sexual dimorphism frequently found in the size and occasionally in the number of these head luminous organs. As a result, the number of nominal species is now close to one hundred. We do not now know which of these are valid, but the number of synonyms may prove to be large. I am indebted to the Smithsonian Institution and to Dr. Robert Gibbs of the U. S. National Museum (USNM) for providing funds and facilities for the study of material deposited in that institution. Dr. E. Bertelsen of the Danish Marine Biological Laboratory (DMBL) and Dr. C. Richard Robins of the University of Miami Marine Laboratory (UMML) generously made available funds and facilities for study at their respective institutions. Dr. Richard 404 bulletin: museum of comparative zoology Backus of the Woods Hole Oceanographie Institution (WHOI), Dr. William J. Richards of the Washington Bureau of Commercial Fisheries (WBCF), and Dr. Gerhard Krefft of the Institut fiir Seefischerei, Hamburg, have kindly provided material in their care. I am grateful to Dr. Rolf L. Bolin of Stanford University and Dr. Giles W. Mead of the Museum of Comparative Zoology (MCZ), Harvard University, for their kindness in reviewing the manuscript and for offering valuable advice and criticism. Partial financial support from National Science Foundation Grant GF 147 is gratefully acknowledged. Abbreviations of names of luminous organs on the head, and of body photophores (Fig. 1), are as follows: Dn: dorsonasal; Vn: So. fiOpostenor Fig. 1. Diagram of a generalized Diaphus showing distribution of the luminous organs of the head and body photophores. Abbreviations after Tuning, 1928, slightly modified. ventronasal; So: suborbital; Ant: antorbital, a small triangular luminous organ present in several species and lying between the Dn and the anterodorsal aspect of the orbit, not shown in Figure 1; Br: branchiostegal; Op: opercular; PO: thoracic (or pectoral); PVO: subpectoral; PLO: suprapectoral ; VO: ventral; VLO: supraventral; SAO : supra-anal ; AOa: anterior anal; AOp: posterior anal; Pol: posterolateral; Pre: precaudal. Measurements were made as follows: standard length (s. 1.): shortest distance between tip of snout and end of hypural; head length: from tip of snout to extreme posterior margin of opercular flap; length of upper jaw: from anterior tip of premaxillary to its posterior end; eye diameter: horizontal distance between opposite margins of bony orbit; depth of head: vertical through posterior margin of orbit; depth of body: vertical through base of upper ray of pectoral fin; depth of caudal peduncle: least vertical depth; predorsal: shortest distance between tip of snout and origin of NAFPAKTITIS: NEW MYCTOPHID FISHES 405 dorsal fin; preventral: shortest distance between tip of snout and base of outermost ray of ventral fin; preanal: shortest distance between tip of snout and origin of anal fin. All measurements were made with a pair of dividers and re- corded in tenths of millimeters. Most measurements were made under a dissecting microscope. Initial values given in the lists of measurements are arithmetic means of all specimens measured; values within parentheses represent extremes. Methods of taking and presenting measurements conform to those used by Bolin (1939). DiAPHUS BERTELSENi new species Figures 2-5 Holotype. A 49.0 mm standard length specimen, MCZ 43121, R/V CHAIN, WHOI, cruise 17, station RHB 801, 26 April, 1961, 00°15'S, 18°35'W to 00°15'S, 18°45'W, 10-ft. Isaacs-Kidd Mid- water Trawl (IKMT), 0-85-0 m depth. Paratypes. One, 61.0 mm s.l, MCZ 43122, R/V CHAIN, WHOI, cruise 35, station RHB 971, 22 February, 1963, 02°00'S, 24°57'W to 01°48'S, 24°54'W, 10-ft. IKMT, 0-295-0 m depth. Two, 21.0-23.0 mm s.l., Dana Collections, R/V DANA station 1223 V, 1 February, 1922, 22°06'N, 84°58'W, ring-trawl, open, conical net, 300 cm in diam. at opening (E300), 600 m wire out. Additional material examined. One, 9.8 mm s.l., R/V DANA station 1191 I, 14 December, 1921, 17°49'N, 64°54'W, 600 m wire out; one, 11.7 mm s.l, R/V DANA station 1231 II, 6 February, 1922, 24°30'N, 80°00'W, 600 m wire out; one, 8.5 mm s.l., R/V DANA station 1243 III, 16 February, 1922, 21°04'N, 73°48'W, 300 m wire out; four, 9.0-10.5 mm s.l, R/V DANA station 1256 IV, 4 March, 1922, 17°43'N, 64°56'W, 300 m wire out; one, 9.0 mm s.l., R/V DANA station 1257 IV, 6 March, 1922, 17°43'N, 64°56'W, 300 m wire out; one, 9.0 mm s.l., R/V DANA station 1274 III, 27 March, 1922, 17°43'N, 64°56'W, 600 m wire out. All these specimens were captured with stramin-nets, open, conical, 200 cm in diam. at opening (S200). Description. Body deep and short, its depth 3.5-3.8 in standard length; head large, its length 2.9-3.0 in standard length; diameter of eye 3.4-3.6 in length of head, 2.3-2.5 in length of upper jaw; length of snout equal to or slightly shorter than half the diameter of the eye; upper jaw 1.4-1.5 in length of head and extending less than one diameter of the eye behind posterior margin of orbit; posterior margin of operculum moderately pointed, the point 406 bulletin: museum of comparative zoology E E 10 a o ■i-i ■4-' CO to '3 I— T O w o > c +3 OS a o Q GO MT3 O 00 NAFPAKTITIS: NEW MYCTOPHID FISHES 407 above PLO. Origin of dorsal fin somewhat anterior to base of ventral fin ; origin of anal fin behind end of base of dorsal fin ; pec- toral fin short, its length 1 .3 in distance between its base and that of ventral fin; ventral fin just reaching anus; adipose fin somewhat anterior to end of base of anal fin. Luminous organs. Dn round, in cup-shaped recess, entirely- above nostril and well separated from that of opposite side (Fig. 3) ; Fig. 3. Diaphus bertelseni, front view; semidiagrammatic representation of the luminou.s organs of the head. Vn considerably larger than Dn, widest at anteroventral aspect of orbit, extending slightly around and under the nostril anteriorly, hardly reaching vertical through anterior margin of lens posteriorly. A conspicuous strip of darkly pigmented tissue present between anterior margin of orbit and luminous organs of head. Body photophores large, close to each other (particularly so in juveniles) and all well below lateral line; distance between PLO and lateral line 3-33^ times the distance between PLO and base of upper ray of pectoral fin; distance between VLO and lateral line 21/^-23^ times the distance between VLO and base of outermost ventral fin ray; SAO equidistant from each other and on a straight, sub- vertical line ; distance between lateral line and SAO3 three times the diameter of that organ; AOai elevated, its lowest margin on or slightly above line through upper margins of next two organs of same series; AOao and AOae on a gentle curve with Pol; distance between lateral line and Pol three times the diameter of that organ; AOpi above end of base of anal fin ; Prc4 slightly detached from rest 408 bulletin: museum of comparative zoology of organs in same series and well below lateral line. Large tri- angular luminous "scale" associated with PLO (especially well preserved on paratype, MCZ 43122). Dentition. Both jaws with inner, irregular series of sharp teeth which are conspicuously larger than those in outer series; those on posterior part of premaxillary definitely curved forward (Fig. 4). \ i>.o. Fig. 4. Lateral view (right side) of posterior parts of jaws, showing den- tition in Diaphus bertelseni (from holot\pe). Counts and measurements. The data which follow are based on the holotype and the three paratypes. Data from non-type material are identified as such. Dorsal fin 14-15; anal fin 15; pectoral fin 11; ventral fin 8; gill rakers on first arch 5 + 1 + 12; scales in lateral line 34-35 ; AO 6 in anterior series and 4 in posterior series. Eight additional specimens, 8.5-11.7 mm s.l., had 6 + 4 AO, and a ninth had 7+3 on the left and 6+4 on the right side. Measurements, in per cent of standard length, are as follows: length of head 34.4 (33.3-34.8); diameter of eye 9.9 (9.6-10.2); length of upper jaw 23.8 (23.0-24.4); depth of body 27.0 (26.0- 28.5); depth of caudal peduncle 12.7 (12.2-13.0); predorsal 46.6 (44.9-47.8); preanal 69.9 (67.2-71.4); preventral 50.5 (48.4-52.4). Affinities. In general shape of body and in size and arrangement of photophores, Diaphus bertelseni (especially juveniles, Fig. 5) is quite similar to D. brachycephalus Taning. This similarity is superficial, however, and there are many striking differences, such as the lack in D. bertelseni of a So, which is so prominent in D. brachycephalus; the number and arrangement of the AOa (6, with the first raised in D. bertelseni, as opposed to 5 in a straight line) ; the presence in D. bertelseni of a large luminous "scale" at PLO, the same structure being completely absent in Taning 's species, NAFPAKTITIS: NEW MYCTOPHID FISHES 409 which also possesses strongly curved and broad-based posterior premaxillary teeth. Differences between Diaphus berielseni and D. coeruleus (Klun- zinger) are found primarily in the shape, size and position of the Dn; in the head and body dimensions (length of head 2.9-3.0 in standard length, as opposed to 3.5-3.7 in D. coeruleus; body depth 3.5-3.8 in standard length, as compared to 4.6-5.0 in D. coeruleus) ; in the length of the upper jaw which, in D. bertelseni, extends less than one eye diameter behind the posterior margin of the orbit, while in D. coeruleus "Upper jaw . . . hinder end . . . surpasses the eye by more than one eye diameter" (Weber and Beaufort, 1913, p. 169) ; in the number of AOp (4 in D. bertelseni, 5 in D. coeruleus). /:;^«'feSv?^:'::'>:,..„^., ^ff0>'-^:^ _ Fig. 5. Diaphus bertelseni, juvenile, 9.8 mm standard length; R/V DANA station 1191 I, 17°49'N, 64°54'\V, 600 m wire out. From Diaphus taaningi Norman, D. bertelseni differs mainly in the size and arrangement of photophores (larger and considerably further below lateral line in D. bertelseni than in D. taaningi), and in the number of organs in the AO series (6+4, as opposed to 5 + 5 in D. taaningi). In all cases discussed above, differences in fin ray and lateral line counts are omitted as they are considered by the writer to be of minor diagnostic importance. Distribution. Pattern and limits of distribution of this species can not now be inferred. The positions of the few stations in which Diaphus bertelseni was taken (Fig. 6) suggest that the species is a South Atlantic one. UtiUzing the South Equatorial Current, members of this group may enter the North Atlantic while remain- ing confined to the South Atlantic Central Water Mass. It is also possible that very young individuals are swept away from the northwestern boundaries of the South Atlantic Central Water Mass by the North Equatorial Current and are carried further west and north. The fact that only juveniles were taken in the Antillean and Florida waters, whereas the two south equatorial stations yielded adults, may be purely accidental. 410 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY This species is the same as that Hsted as "Diaphus (sensu Fraser- Brunner) sp. A" by Backus et al., 1965. The species is named in honor of Dr. E. Bertelsen, Director of the Danish Marine Biological Institute, Charlottenlund. Fig. 6. Chart showing the stations at which Diaphus berielseni (solid dots) and Diaphus lewisi (rectangles) were taken. The solid dot at about 17°N, 64°\V, represents four DANA stations (1191 I, 14 December, 1921; 1256 IV, 4 March, 1922; 1257 IV, 6 March, 1922; 1274 III, 27 March, 1922). Areas of high productivity are stippled (reproduced to scale from Sverdrup et al., 1942). Diaphus lewisi new species Figures 7-11 Holotype. Male, 27.5 mm standard length, Dana Collections, R/V DANA station 4005 XI, 12 Alarch, 1930, 13°31'N, 18°03'W, stramin-net, open, conical, 200 cm in diam. at opening (S200), 50 m wire out. Paratypes. Two, 20.0-28.5 mm s.L, Dana Collections, data as above for holotype. Four, 15.6-32.0 mm s.L, MCZ 44000, R/V CHAIN, WHOI, cruise 35, station RHB 972, 23 February, 1963, 00°03'N, 25°00'W to 00°15'N, 25°00'W, 10-ft. IKMT, 0-87-0 m depth. NAFPAKTITIS: NEW MYCTOPHID FISHES 411 One hundred, 18.5-31.8 mm s.L, R/V GERONIMO, WBCF, cruise 2, station 138, BCF Cat. No. 370, 19 August, 1963, 09°15'S, 07°06'E, mid- water trawl, depth (?). Eighty specimens in MCZ, MCZ 44001, 20 specimens in USNM, USNM 259155-Fl. Additional material exainined. One, 22.0 mm s.l., R/V DANA station 1159 I, 29 October, 1921, 17°55'N, 24°35'W, ring-trawl, open, conical, 300 cm in diam. at opening (E300), 5000 m wire out; 133, 11.0-20.0 mm s.L, R/V DANA station 4003 VI, 9 March, 1930, 08°26'N, 15°11'W, stramin-net, open, conical, 150 cm in diam. at opening (S150), 1000 m wire out; one, 14.5 mm s.l., R/V DANA station 4004 IV, 11 March, 1930, 10°21'N, 17°59'W, stramin-net, open, conical, 200 cm in diam. at opening (S200), 100 m wire out; two, 25.0-26.8 mm s.L, R/V CHAIN, WHOI, cruise 17, station IKT4, 1 April, 1961, 02°10'S, 17°25'W, 10-ft. IKMT, 1800 m wire out; two, 23.0-25.2 mm s.L, R/V CHAIN, WHOI, cruise 17, station IKT5, 3 April, 1961, 00°35'S, 11°30'W, 10-ft. IKMT, depth (?); three, 15.2-15.7 mm s.L, R/V CHAIN, WHOI, cruise 17, station IKT9, 20 April, 1961, 07°15'N, 14°00'W, 10-ft. IKMT, 400 m wire out; one, 20.0 mm s.L, R/V CHAIN, WHOI, cruise 17, station RHB 801, 26 April, 1961, 00°15'S, 18°40'W, 10-ft. IKMT, 0-85-0 m depth; three, 17.5-26.0 mm s.L, R/V CHAIN, WHOI, cruise 17, station RHB 803, 1 May, 1961, 09°27'N, 27°45'W, 10-ft. IKMT, 0-275-0 m depth; 94, 10.2-26.0 mm s.L, R/V CHAIN, WHOI, cruise 17, station RHB 804, 1-2 May, 1961, 10°55'N, 29°30'W, 10-ft. IKMT, 0-42-0 m depth; three, 25.5-28.7 mm s.L, R/V GERONIMO, WBCF, cruise 2, station 166, BCF Cat. No. 181, 25 August, 1963, 05°49'S, 10°00'E, mid-water trawl, depth (?); six, 21.5-29.0 mm s.L, R/V GERONIMO, WBCF, cruise 3, station 130, BCF Cat. No. 971, 26 February, 1964, 04°58'N, 00°30'W, mid-water trawl, depth (?); one, 23.5 mm s.L, R/V GERONIMO, WBCF, BCF Cat. No. 324, 03°28'S, 00°14'W; one, 30.0 mm s.L, R/V PILLSBURY, UMML, station 10, 25 May, 1964, 05°55'N, 02°52'E to 05°58'N, 02°50'E, IKMT, gear depth 0-655-1065-0 m; two, 15.0-18.0 mm s.L, R/V PILLSBURY, UMML, station 36, 29 May, 1964, 03°50'N, 02°37'W, IKMT, gear depth 0-750-0 m; one, 15.2 mm s.L, R/V PILLSBURY, UMML, station 37, 29 May, 1964, 04°00'N, 02°46'W to 04°05'N, 02°50'W, IKMT, gear depth 0-480-490-0 m; five, 23.5-26.8 mm s.L, R/V WALTHER HERWIG, Institut fiir Seefischerei, station 103, 24 March, 1964, 14°30'N, 22°45'W, IKMT, depth (?); one, 31.0 mm s.L, R/V WALTHER HERWIG, Institut fiir Seefischerei, station 109, 25 March, 1964, 14°30'N, 19°42'W, IKMT, depth (?); one, 29.2 mm s.L, R/V WALTHER 412 bulletin: museum of comparative zoology HERWIG, Institut fur Seefischerei, station 129, 4 April, 1964, 09°14'N, 16°00'W, IKMT, depth (?). Of the additional material examined, those specimens taken by R/V DANA are deposited in the Dana Collections, Charlotten- lund, Denmark; the specimens collected by R/V CHAIN and R/V GERONBIO are deposited in the AICZ; those caught by R/V PILLSBURY and R/V WALTHER HERWIG are to be found in the UMjNIL and the Institut fiir Saefischerei, Hamburg, respectively. None of the collections examined appears to have been made with closing nets. Description. Small form; depth of body 3.8-4.3 in standard length; length of head 2.9-3.1 in standard length; diameter of eye 3.4-4.2 in length of head and 2.3-2.9 in length of upper jaw; snout longer than half the diameter of the eye; anterior spine-hke end of supraorbital ridge extending forward and downward nearly reaching the posterodorsal aspect of the nostril; upper jaw 1.4-1.5 in length of head and extending one diameter of eye behind poste- rior margin of orbit ; pterotic spine conspicuous ; posterior margin of operculum rounded dorsally, moderately pointed posteriorly, the point at or slightly below PLO. Origin of dorsal fin directly above or slightly anterior to base of ventral fin; orighi of anal fin behind end of base of dorsal fin; upper rays of pectoral fin reach base of ventral fin; ventral fin nearly reaching origin of anal fin; adipose fin above last 4-5 rays of anal fin. Luminous organs. Dn apparently fused with Vn, the compound structure beginning at or somewhat higher than the dorsal margin of nostril, extending between latter and anterior margin of orbit and expanding ventrally, its posteroventral tip not reaching vertical through anterior margin of lens; a strip of black tissue separates the preorbital luminous complex from the anterior margin of orbit and extends along the ventral aspect of eye; its posterior end expands into a small pocket-like structure which protrudes into the iris behind the vertical through the middle of lens ; inside the pocket-like structure there is a small, round So. Tha suborbital organ first appears in individuals 16.0-17.0 mm s.l., and in well preserved adult specimens it appears connected with tha Dn-Vn complex by means of a very narrow isthmus of luminous tissue. Body photophores of medium size and separated from one another by an interspace at least as wide as the diameter of these organs; PLO and VLO distinctly closer to pectoral and ventral fin bases, respectively, than to lateral line; SAO series of three unequally NAFPAKTITIS : NEW MYCTOPHID FISHES 413 Ik M 1? a o o o C3 +-* < Q T3 e 03 lO ^ u l>^ c3 (M ^ C _iJ O "rt Q 0; _5 >^ r-" +J 5 _o +^ -f-' ^ ri 0 3 ■» K cc ^ -^^ S. -S e '^ ~~ U C2 ^ t^ c3 tb Q E -i-i 3 414 bulletin: museum of comparative zoology spaced photophores forming a slightly curved and steeply ascend- ing line ; distance between SAOi and SAO2 1 3^-2 times that between SAOi and VO5, distance between SAO2 and SAO3 at least 13^ times that between SAOi and SAO2, SAO3 about its own diameter below lateral hne; AOai 1-2 times its diameter anterodorsad to A0a2, last AOa in line with or slightly raised above preceding three or four organs of same series; Pol directly above last AOa and 3^^-! times its own diameter below lateral line; AOpi above posterior end of anal fin base ; Prc4 distinctly detached from rest of organs of same series and well below lateral hne. A triangular luminous "scale" present at PLO. Dentition. Upper jaw with inner series of strongly recurved, broad-based teeth (Fig. 8) ; lower jaw with inner irregular series of teeth feebler than those of upper jaw and only slightly curved forward. Sexual dimorphism. Alales are easily distinguished from females by the presence in the former of a conspicuous, roughly triangular antorbital luminous organ (Ant) immediately above and in contact with the Dn-Vn complex (Figs. 7 and 9). This sexual character first appears as a small, diffuse patch of luminous tissue in individ- uals 18.0-19.0 mm standard length, and attains its maximum size ...■■■■■ ■■ ...-;;^s#gi^P •■■■ ....--i^s^iiiilt' i!s^ V^;^^. ^1^ ■aS:^ %^^^ ^ Fig. 8. Lateral view (right side) of posterior parts of jaws, sliowing den- tition in Diaphus lewisi. NAFPAKTITIS: NEW MYCTOPHID FISHES 415 and its definitive outlines at sexual maturity which, in this small fish, is attained at a size of 27.0-28.0 mm standard length. Counts and measurements. Meristic data based on 127 speci- mens are as follows: dorsal fin 13-14 (usually 14); anal fin 14-15 (usually 15); pectoral fin 10-11; ventral fin 8; gill rakers on first arch 7-8 + 1 + 14-15, total 22-24; scales in lateral line 35-36; .^^lifet^--...... w ■:0%"---^. ■■ -■ ; -■•.•.* .i \ b.n. :iviv;.itv-<.-.v ^^"^^iW^i^m^m^^^^^s,..., ■"^?**f% ■■■■' ■" ■■■"■" ■■■ ■ '-'w^t-' :y^ ^c X ■o-n. y6 Fig. 9. Sexual dimorphism in the luminous organs of the head in Diaphus lewisi. A: male, 32.0 mm standard length; B: female, 27.0 mm standard length. 416 bulletin: museum of comparative zoology AOa 5-6 (very rarely 7), AOp 4-5 (very rarely 6). The AO series show considerable variation in number and arrangement not only among individuals but also between two sides of the same indi- vidual, the most common patterns being those shown in Figures 7 and 10. Fis;. 10. Caudal region of Diaphus leuisi, showing variation in the num- ber and arrangement of the photophores in the AOa series. Measurements, in per cent of standard length, based on 40 specimens 15.6-32.0 mm s.l., as folloAvs: length of head 33.6 (32.1-35.0); diameter of eye 8.9 (8.1-10.3); length of upper jaw 23.3 (22.0-24.2) ; depth of body 24.4 (23.0-26.2); depth of caudal peduncle 10.9 (10.0-12.6); predorsal 46.5 (44.8-48.9); preanal 65.4 (62.0-67.0); pre ventral 47.1 (44.8-48.5). J>.n. Fig. 11. Diaphus lewisi, juvenile, 11.5 mm standard length; R/V CHAIN, WHOI, cruise 17, station RHB 804, 10°55'N, 29°;^0'W, depth 42 m. Affinities. Diaphus lewisi is a small fish, probably not growing larger than 32-33 mm in standard length. It belongs to the Diaphus rafmesquei group which includes D. rafincsquei (Cocco) and D. holti Taning in the northeastern North Atlantic, D. mollis Ta,ning, D. brachycephalus Taning, and probably D. theta Eigen- mann and Eigenmann in the southwestern North Atlantic and, in NAFPAKTITIS: NEW MYCTOPHID FISHES 417 addition, several other species elsewhere. The main characters common to the members of this group are the three distinct luminous organs on the head (Dn, Vn and So) and the strongly recurved, broad-based premaxillary teeth. D. lewisi, although deviating from the typical rafinesqiiei pattern of luminous organs on the head in the fusion of the Dn and Vn, does possess the peculiar premaxillary dentition characteristic of the group. Diaphus suborbitalis Weber from the Indo-Pacific area has luminous organs on the head which seem to be intermediate between those of the rafinesquei group and those found in D. lewisi. In Weber's species the Dn and Vn tend to, but have not fused completely. In many other respects D. suborbitalis is, according to Weber (1913, p. 91), closely related to D. fulgens Brauer, a typical species of the rafinesquei group from the Pacific Ocean and one possibly identical with D. mollis Taning. Further discussion of possible relationships among the species mentioned in this para- graph must await a direct comparison of pertinent material. Distribution. Available data suggest that the area occupied by Diaphus lewisi is rather well defined (Fig. 6). Within its range this species is quite common, and the pattern of its distribution is obviously influenced by the hydrography along the west coast of Africa. The cold Benguela Current and the continuous upwelling from depths of 200-300 m account for the remarkably high pro- ductivity, the relatively low temperature, and the low salinity of the coastal surface water. These conditions stand in contrast to those of surface waters of the general South Atlantic tropical and sub-tropical areas which are warmer, more sahne and low in nutrient content (Raymont, 19G3). The belt of fertile water extends approximately 200 km offshore, forming tongues of gradually diminishing plankton density extending outward from the coast (Fig. 6). These tongues correspond to the main water movements that flow away from the coast, as indicated by the out- ward extension of the isotherms (Sverdrup et al, 1942). The high productivity of the coastal Benguela Current is reflected in the rich zooplankton with which, in turn, may be associated the abundance of marine vertebrate and invertebrate forms (Raymont, 1933). With a knowledge of the hydrography of the area concerned, the distribution of Diaphus lewisi and, no doubt, of other mesopelagic forms becomes meaningful. Relatively abundant along the con- tinental slope of the African west coast and particularly so in areas of high productivity, D. lewisi becomes less and less common towards the west. The northernmost limits of its distribution 418 bulletin: museum of comparative zoology (about 18°N) more or less coincide with the southern limits of the North Atlantic Central Water Mass. A rather marked drop in water temperature at about 18-20°S and in depths of 200-400 m possibly constitutes a barrier to the spreading of the species further south along the west coast of Africa. The species is named in honor of Dr. Robert E. Lewis, Curator of the i\Iuseum of Natural History, American University of Beirut, Lebanon. DISCUSSION Taxonomy. The taxonomy of the "Diaphus complex," hke that of other difficult groups, has been arranged and repeatedly re- arranged, especially on the generic level, with little success in achieving a stable system. Inadequate material may have been one of the main reasons for the existence in the literature of so many genera and subgenera. With the increase in the number and size of collections, some students of this group began realizing that they were dealing with the potentially unlimited inventiveness of nature as regards the size, shape and complexity of the luminous organs of the head. They also saw that such morphological features as the "theta" (Greek letter 6) configuration of the body photophores and the number and arrangement of some of the series of these organs (PVO, PO, VO and Pre) seemed, with very few exceptions, highly constant throughout the group. As a result one single genus, Diaphus, was accepted and used by them. ]\Iyctophid taxonomy has been heavily dependent on the number and distribution of photophores. It is undoubtedly true that these characters are of great importance in distinguishing the genera within the family. However, at the species level the value of these characters is often questionable. Their inadequacy, in the Diaphus group at least, becomes obvious as one grows familiar with: a) the close morphological similarity shown by several species of the group, and b) the relatively high degree of intraspecific variation which often masks interspecific differences. In contrast to the generally conservative diaphid morphology, the luminous organs of the head show an astonishing variety of sizes, shapes, patterns and numbers. In most cases the role of these organs in interspecific differentiation is decisive. Of the few diagnostically important characters available to the student of the Diaphus group, some have been consistently over- looked. To cite one example: In most studies of Diaphus very little, if any, attention has been paid to dentition, a character the NAFPAKTITIS: NEW MYCTOPHID FISHES 419 importance of which has been repeatedly emphasized by Bolin (1939, p. 124; 1959, p. 20). In his detailed description of the genus Diaphus, as exemphfied by Diaphus rafinesquei, Bolin (1939, p. 124) calls attention to the pecuUar dentition of this species and suggests that: "Probably most of the other species now placed in the genus will require re-allocation since examination of several forms has failed to reveal a dentition similar to that found in Diaphus rafinesquei." Bohn's suggestion subsequently found support in the observation that those forms with sharply recurved premaxillary teeth also had three well-defined luminous organs on the head (Dn, Vn and So). These two features, namely dentition and pattern of luminous organs, seemed important enough to justify generic distinction of the species which possessed them. However, further examination of large numbers of specimens belonging to different diaphid species has revealed that the premaxillary teeth display a series of gradu- ally changing shapes. Between the two extremes, straight and sharply recurved, there exists a series of more or less curved pre- maxillary teeth. Furthermore, the young of some species (e. g. Diaphus elucens, D. prohlematicus, etc.) have curved teeth, especi- ally on the posterior end of the premaxillary. These teeth are apparently lost and replaced by straight or less-curved ones in older individuals of the same species. Finally, a generic hne becomes even more difficult to draw if one considers the existence of forms like Diaphus lewisi and, probably, D. suborhitalis Weber. The former species (I have not personally examined D. suborhitalis, hence I cannot speak with certainty about it), while deviating considerably from the rafinesquei pattern of luminous organs on the head, does show the rafinesquei kind of dentition. The gradual change in shape displayed by the premaxillary teeth does not at all diminish the diagnostic usefulness of dentition, especially in the case of juveniles of morphologically very similar diaphid species. Young individuals of Diaphus fragilis Taning and D. elucens Brauer, for instance, are very difficult to differentiate. With the numbers and arrangement of their body photophores strikingly similar, luminous organs of the head easily confused (before attainment of their definitive pattern and size), and with horizontal distributions considerably overlapping, the juveniles of these two species can be identified on the basis of premaxillary and particularly dentary teeth shape and size (e. g. lower jaw with inner series of very large, Avidely but regularly spaced teeth in Diaphus fragilis, lower jaw with inner irregular series of denser and much smaller teeth in D. elucens). 420 bulletin: museum of comparative zoology Turning once again to the character of the luminous organs of the head, we are faced with a situation somewhat analogous to that of the teeth. These organs, in their great variety of designs, can, with some imagination, be arranged in one or more series of pro- gressively increasing complexity. This has certainly been at- tempted in the past (Parr, 1928, p. 140). However, sound infer- ences about evolutionary trends of these organs cannot be made on the basis of external morphology alone. A careful histological study of their structure and innervation could probably help us gain some insight into the evolution of the numerous patterns and thus, perhaps, lead us to a better understanding of the phylogenetic relationships among the various species in the group. Professor Rolf Bolin is of the opinion (personal communication) that there are several divergent lines within the Diaphus group. Although I fully agree with his opinion, I feel inclined to disagree with the idea that these divergent lines are, at the present state of our knowledge at least, clear enough to justify a division of the group into more than two genera. An attempt to formalize these lines would lead either to a large number of genera or, at best, to a reasonable number of genera but with several species "suspended" in between. On the basis of the arguments presented above, it is suggested here that all but the species assigned to the genus Lobianchia Gatti (e.g. Aethoprora Goode and Bean, Panthophos Jordan and Hubbs, Lamprossa Jordan and Hubbs, Cavelampus Whitley) be included within the single genus Diaphus, without formal division into subgenera (e.g. Hypcrphotops, Panthophos and Lamprossa, all three erected by Fraser-Brunner in 1949). Ecology, speciaiion and phenotypic similarity. Some speculation on the phenotypic similarity among several myctophid forms is in order here, and should start with a brief consideration of some aspects of oceanic ecology and of possible ways of speciation among high-seas forms. Ecological conditions in the sub-tropical and tropical oceans seem to have been quite stable in geological time (Marshall, 1963, p. 182). Relative differences in physico- chemical factors, i. e. temperature, salinity and density, between adjacent water masses and, below the thermocline, between bathy- metrically contiguous layers are usually very small. Repeated invasions of these adjacent environments by populations of a given species probably result in some of the more adaptable individuals becoming physiologically adjusted to the small differences. Fol- lowing this adjustment, which may not require more than slight adaptive changes in the reproductive physiology of the invaders. NAFPAKTITIS: NEW MYCTOPHID FISHES 421 colonization of the new habitat, or habitats, is under way. Subse- quently, one might expect the gene flow between the colonizers and the parent species to decrease. Finally, perhaps through the development of preferential mating and competition "... between [the] contiguous populations in the zone of contact and the subse- quent elimination of the less well adapted intermediates in this zone. . . ." (Ebeling, 1962, p. 149, from Fryer, 1959), the initial, partial isolation of the populations may be substantially reinforced. Now, if this is the approximate sequence and extent of the major evolutionary events leading to the formation of new species, then it would be reasonable to infer that speciation in many deep-sea fishes does not necessarily involve conspicuous or even readily noticeable structural modifications. In considering the great morphological similarity among several myctophid forms one should also take into account such evolu- tionary processes as convergence and parallel adaptations. Hubbs (1941, p. 190) says: "The general tendency of fishes to speciate along parallel courses in correlation with the temperature and sahnity of the water is being repeatedly indicated. ..." Finally, in his discussion of the phenomenon of sibling species, Mayr (1963, p. 57) calls attention to the evidence, found in recent work in developmental genetics, indicating that there is a selective pre- mium on the maintenance of the phenotype. "Any disturbance of the developmental process by a gene mutation will result in a selection pressure in favor of other genes that restore development along the normal, time-tested channels" (Mayr, 1963, from Lerner, 1954, and Waddington, 1956). The rather strong emphasis placed above on adaptation to apparently minor differences in physico-chemical factors seems to be in conflict with the known ability of adult myctophids to tolerate wide extremes in temperature, salinity and, perhaps, several other ecological factors. In the course of their extensive (several hundred meters) diurnal vertical migrations, these fishes "may weU be subjected to much greater differences in temperature than those marking the limits of their distribution" (Fraser- Brunner, 1949, p. 1020). This conflict may be resolved if we assume that during a certain period, or periods, of the year these fishes become, in terms of reproductive physiology, specifically stenothermal and stenohaline. It may also be that survival of the gametes, successful fertihzation, or early development fail in the absence of optimal, species-specific physico-chemical conditions. If these assumptions are correct we should expect these optimal 422 bulletin: museum of comparative zoology ecological conditions to delimit the breeding area and hence the distribution of the species. But, "certainly in the sea we have many suggestions that the distribution of the species is much wider in many instances than the distribution of the breeding population" (Ebeling, 1962, p. 139, from Bullock, 1958). This brings us to another potent environmental factor, namely the oceanic current patterns, and their role in the phenomenon of "expatriation." Expatriation. Within a breeding area the larvae are restricted to surface or near-surface waters. Just before, or at the very beginning of metamorphosis, these larval forms move to deeper layers. Based on his studies of larval fishes, Taning (1918, p. 20) writes: "There can . . . hardly be any doubt that the Scopelids [myctophids], when metamorphosis sets in, undergo a thorough change, acquiring a different specific gravity to that which they had as postlarvae, and consequently move down, either actively or passively, to water layers of a specific gravity suitable to their requirements during metamorphosis, and later, after the meta- morphosis is completed, ascend once more to the upper layers. In other words, we have here a decided instance of (passive or active) ontogenetic migration." Both as larvae prior to their "ontogenetic migration" and as young right after their ascent to the upper layers, myctophid and other forms with similar life histories are at the mercy of the prevailing current systems in the area concerned. The larvae and young may either be carried away from the breeding area or they may be kept restricted within it (current gyres, eddies, etc.). In the first case the individuals may spread over a wide range, far away from the breeding area of the species to which they belong. Sexual maturity will probably be attained while in environments which, although favorable in terms of satisfaction of purely indi\'idual needs, are far from meet- ing the species-specific reproductive requirements. These indi- viduals are thought to be "expatriated" and reproductively lost to their populations (Ebeling, 1962, p. 139). Concerning the fate of the "expatriates," Eknian (1953, p. 317) remarks that ". . . it remains questionable whether the species is able to exist independently in the unfavorable region or whether it wou!d not die out there if it were not continuously reinforced from the more favorable regions," and ". . . the unfavorable region is . . . outside the real home of the species, and it [is] . . . possible to contrast the autochthonous main mass of the species which lives in the reproductive area with an allochthonous sterile expatriated contingent in an expatriation area." Bolin (1959b, NAFPAKTITIS : NEW MYCTOPHID FISHES 423 p. 142), also, points out that "while straggling adults may exist for long periods in waters far beyond the normal range of the species, permanent populations are restricted to the proximity of the areas where spawning can be successful." During my study of the distri- bution of the diaphid species in the Atlantic Ocean I have found considerable evidence indicating that many of the diaphid speci- mens captured off the northeastern coast of the United States are "expatriates" belonging to species the breeding areas of which are well within the tropical and sub-tropical waters. I am well aware of the weaknesses in my assumptions and argu- ments. I would hke to conclude with a well-worn but nevertheless appropriate statement — a great deal of work remains to be done, especially with regard to the distribution and bionomics of this scientifically and, perhaps, economically important group of mesopelagic fishes. LITERATURE CITED Backus, R. H., G. W. Mead, R. L. Haedrich and A. W. Ebeling 1965. The mesopelagic fishes collected during cruise 17 of the R/V Chain, with a method for analyzing faunal transects. Bull. Mus. Comp. ZooL, 134(5): 139-157. BoLiN, R. L. 1939. A review of the mj'ctophid fishes of the Pacific coast of the United States and of lower California. Stanford Ichthyol. Bull., 1(4): 89-160. 1959a. Iniomi. Myctophidae from the "Michael Sars" North Atlantic deep-.sea expedition, 1910. Rep. "Michael Sars" N. Atl. Deep- sea Exped., 4(7): 1-45. 1959b. Differential bipolarity in the Atlantic and Pacific as expressed by the myctophid fishes. In International Oceanographic Congress, Preprints, 31 August - 12 September 1959, Mary Sears, ed., American Association for the Advancement of Science, Washing- ton, D. C, pp. 142-143. Brauer, a. 1903. Die Tiefsee-Fische. I. Systematischer Teil. Wiss. Ergeb. Deutsch. Tiefsee-Exped., Valdivia, 1898-99, 15, 432 pp. Ebeling, A. W. 1962. Systematica and zoogeography of the species in the bathypelagic fish genus Melamphaes Glinther. Dana-Report No. 58, 1962, Carlsberg Foundation, Copenhagen, 164 pp. Ekman, S. 1953. Zoogeography of the sea. Sidgwick and Jackson, London, 417 pp. Fraser-Brunner, a. 1949. A classification of the fishes of the family Myctophidae. Proc. Zool. Soc. London, 118: 1019-1106. 424 bulletin: museum of comparative zoology GooDE, G. B. and T. H. Bean 1895. Oceanic ichthyology. Special Bull. U. S. Nat. Mus., Washington, XXXV + 553 pp. HuBBs, Carl L. 1941. The relation of hydrological conditions to speciation in fishes. Sj^iip. Hydrobiol., Univ. Wisconsin Press, pp. 182-196. Kltjnzinger, C. B. 1870. Fische des Rothen Meeres. I. Theil. Percoiden-Mugiloiden. Wien, 248 pp. KULIKOVA, E. B. 1961. On the lantern fishes of the genus Diaphus (Scopelidae) from the western Pacific. Trudy Inst. Okeanol. Moskva, 43: 5-39. (In Russian.) Marshall, N. B. 1963. Diversit}', distribution and speciation of deep-sea fishes. In Speciation in the sea, J. P. Harding and Norman Tebble, eds., Symp. Syst. Ass., Pub. 5, London, pp. 181-195. Mayr, E. 1963. Animal species and evolution. Harvard Univ. Press, Cambridge, Massachusetts, 797 pp. Norman, J. R. 1930. Oceanic fishes and flatfishes collected in 1925-1927. Discovery Repts., Cambridge, 2: 261-370. Parr, A. E. 1928. Deep-sea fishes of the order Iniomi from the waters around the Bahama and Bermuda Islands. Bull. Bingham Oceanogr. Coll., New Haven, 3 (Myctophidae): 47-156. 1929. Notes on the species of myctophine fishes represented by type specimens in the United States National Museum. Proc. U. S. Nat. Mus., 76: 1-47. Raymont, J. E. G. 1963. Plankton and productivity in the oceans. Pergamon Press, Macmillan Co., New York, 660 pp. SvERDRUP, H. U., M. W. Johnson and R. H. Fleming 1942. The oceans, their physics, chemistry and general biology. Prentice- Hall, Inc., New Jersey, 1087 pp. Taning, a. V. 1918. Mediterranean Scopelidae. Repts. Danish Oceanogr. Exped. 1908-10, Kj0benhavn, 2: 1-154. 1928. Synopsis of the scopelids in the North Atlantic. Vidensk. Medd. Dansk. Naturh. Foren. Kbh., Kj0benhavn, 86: 49-69. 1932. Notes on scopelids from the Dana expeditions. Vidensk. Medd. Dansk. Naturh. Foren. Kbh., Kj0benhavn, 94: 125-146. Weber, M. 1913. Die Fische der Siboga-Expedition. Siboga Repts., Leyden, 57 (32): XII + 710. Weber, M. and L. F. Beaufort 1913. The fishes of the Indo-Australian archipelago. Leyden, 2: XX + 404. (Received 5 April 1965.) Bulletin of the Museum of Comparative Zoology HARVARD UNIVERSITY Vol. 133, Xo. 10 THE AMEIVA (LACERTILIA, TEIIDAE) OP HISPANIOLA. IT. GEOGRAPHIC VARIATION IN AMEIVA CHRYS0LAE3IA COPE By Albert Schwartz and Ronald F. Klinikowski CAMBRIDGE, MASS., U.S.A. PRINTED FOR THE MUSEUM March 16, 1900 Publications Issued by or in Connection WITH THE MUSEUM OF COMPARATIVE ZOOLOCY HAPtVARD UNIVERSITY Bulletin (octavo) 1863 — The current volume is Vol. 133. Breviora (octavo) 1952 — No. 242 is current. Memoirs (quarto) 1864-1938 — Publication terminated with Vol. vo. JoHNSONiA (quarto) 1941 — A publication of the Department of Mollusks. Vol. 4, no. 43 is current. Occasional Papers of the Department op Mollusks (octavo) 1945 — Vol. 2, no. 31 is current. Proceedings of the Nfav England Zoological Club (octavo) 1899-1948 — Published in connection with the Museum. Publication terminated with Vol. 24. The continuing publications are issued at irregular intervals in num- bers which may be purchased separately. Prices and lists may be obtained from the Publications Office of the Museum of Comparative Zoology, Cambridge 38, Massachusetts. Peters' Check-List of Birds of the World. Volume 1 is out of print; volumes 2 to 7, 9, 10 and 15 are currently available. Phylogeny and Evolution of Crustacea. Proceedings of a confer- ence held at Cambridge, Mass., March 6-8, 1962. H. B. Whittington and W. D. I. Rolfe, editors. Cambridge, Mass., 192 pp., 80 figs., 1963. Cloth bound, $6.75. Fishes of the Gulf of Maine, by Henry B. Bigelow and AVilliam C. Schroeder. Washington, viii - 577 pp., 1953. Photo-offset reprint, $6.50. Publications of the Boston Society of Natural History The remaining stock of the scientific periodicals of the Boston Society of Natural History has been transferred to the Museum of Comparative Zoology for distribution. Prices for individual numbers may be had upon request. Bulletin of the Museum of Comparative Zoology HARVARD UNIVERSITY Vol. 133, No. 10 THE AMEIVA (LACERTILIA, TEIIDAE) OF HISPANIOLA. II. GEOGRAPHIC VARIATION IN AMEIVA CHRYSOLAEMA COPE By Albert Schwartz and Ronald F. Klinikowski CAMBRIDGE, MASS., U.S.A. PRINTED FOR THE MUSEUM March, 1966 Bull. Mus. Comp. Zool., Harvard Univ., 133(10): 425-487, March, 1966. No. 10 — The Ameiva {Lacertilia, Teiidae) of Hispaniola II. Geographic Variation in Ameiva Chrysolaema Cope By Albert Schwartz 10,000 SW 84th Street, Miami, Florida 33143 AND Ronald F. Klinikowski 127 Spring St., Reading, Pennsylvania INTRODUCTION The largest of the three species of Hispaniolan ground lizards, Ameiva chrysolaema Cope, was described in 1868 with the type locality ''Gonave Island." Three other names were then ap- plied to this species in rapid succession: vittipunctata Cope 1871, affinis Fischer 1883, and regularis Fischer 1888. Of these, vittipunctata, as Cochran (1941: 275-276) has made clear, was apparently described by Cope from a young specimen (snout- vent length 88 mm) which was part of the same series from which he himself had taken the type of chrysolaema. Cope, how- ever, gave a different type locality — ' ' city of Santo Domingo ' ' — for vittipunctataA In her revision of the species in ' ' Herpetology of Hispaniola, ' ' Cochran (1941: 275-292) considered vittipunctata Cope a strict synonym of A. chrysolaema Cope. A. regularis Fischer was re- garded as a strict synonym of affinis Fischer, which was accepted as a valid mainland subspecies (a new status since affinis had previously been regarded either as a full species or a synonym). Prior to 1941 two subspecies of A. chrysolaema had been de- scribed from satellite islands of Hispaniola : A. c. woodi Coch- ran from He de la Tortue and A. c. ahhotti Noble from Isla Be- ata. These Cochran considered recognizable. A third subspecies, A. c. hoekeri Mertens (1938: 338), however, from the mainland at Fondo Negro, Republica Dominicana, was rejected as a syno- nym of A. taeniura. 1 Cochran (1941 : 245) noted that the type locality of Cele^tus f= Diploglos- SU8) weinlandi, also described by Cope and said by him to have been collected on He de la (ionave, was incorrect and the s|ipcimen actnallv came from the mainland within 25 miles of Port-au-Prince, Haiti. Since A. C. Younglove who collected the type of C. tveinlanrti in 1868 also collecred the type of A. rhrysn- laema, it is appropriate, as Dr. Cochran has done, to restrict the type locality of A. chryftolaema to "within 2.t miles of Pnrt-aii-I'rince." If the type of A. vittipunctata is actually part of the same series as the type of A. chrysolaema, the type locality of the former should likewise be considered the same as that of the latter, despite Cope's statement that it came from the city of Santo Domingo. 428 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY The discussion below departs radically from Cochran's revi- sion in rejectinpr affinis while recoonizing" many other mainland subspecies. In large part this has been due to the much greater amount of material available to us. Six hundred and fifty-five carefully documented and noted specimens of A. chrysolaema have been collected by ourselves and associates betvreen June 1962 and September 1964. Large fresh series of this lizard are now availal)le to us from southern and central Haiti, from He de la Tortue and Gonave, and from the whole of its range in the Republica Dominicana. These speci- mens are in the Albert Schwartz Field Series (ASFS) and the collection of Richard Thomas (RT). They have been amassed through the efforts of ]\Iiss Patricia A. Heinlein, and Messrs. David C. Leber and Richard Thomas. All deserve our commen- dations and thanks, most especially Messrs. Leber and Thomas who made special efforts on our liehalf to secure topotypes of A. c. ahboffi on Lsla Beata. Li addition, we have examined 417 specimens in the following collections : American ^Museum of Natural History (AMXH), Carnegie ]Museum (C^I), ]\Iuseum of Comparative Zoology (MCZ), Museum of Zoology, University of Michigan (U]\OIZ), and United States National Museum (USNM). To the curators and their assistants — Charles M. Bogert and Miss Grace M. Tilger, Neil D. Richmond, Ernest E. Williams, Charles F. Walker and George R. Zug, Doris M. Coch- ran and James A. Peters — we wish to express our appreciation for allowing us to examine pertinent specimens under their care. Paratypes of new forms have been deposited in the Museum of Natural History, University of Kansas (KU), and the University of Illinois Museum of Natural History (UIMNH), as well as in the above collections. The Harvard collections once again have been indispensable ; through the efforts of Dr. "Williams, large and well preserved series from northern Haiti have been made available to us; without these our interpretation of the north- western Haitian situation would have been not merely difficult, but rather impossible. In the matter of literature, Edmond V. Malnate has been most helpful and we are grateful for his co- operation. The illustrations are the work of the junior author. SYSTEMATIC ACCOUNT The Species as a Whole Amciva clirysolacma may be defined as follows: 1) a large species of the genus Amcii'a with snout-vent length to 160 mm in males and 134 in females; 2) dorsal caudal scales keeled and SCHWARTZ AND KLINIKOWSKI : AMEIVA 429 straig-lit; 3) ventrals in 10, 11, or 12 transverse rows and in 33 to 41 long-itndinal rows; 4) fourth toe subdipital seales from 66 to 101; 5) femoral pores 24 to 52; 6) fifteenth eandal verticil with 30 to 52 scales; 7) dorsal pattern consistino- of one of the following: a) a series of dorsal yellow to huffy longitudinal lines on a brown, tan, grayish tan, greenish, or lilackish ground color, the lines (straight or wavy) at times modified into dashes, dots, or fused with one another to give ultimately a median dor- sal longitudinal band, b) a rather uniform covering of brightly colored (blue, orange, yellow) spots on a dark background, c) a pale ground with dark vermiculations and tigroid vertical lateral bars, or d) completely or almost unicolor dorsally without any striking jiattern elements; and 8) hemipenis extending to about the seventh to ninth caudal verticil, sulcate surface naked, sulcus bifurcating apically, the branches ending in two poorly defined scalloped apical discs, non-sulcate surface entirely flounced, the flounces extending to the margins of the sulcate sur- face, a small smooth triangular area on the non-sulcat(> side which divides the flounces for about one-third the length of the organ into two fields of flounces which correspond to the apical discs. The center of the distriliution of A. chrysolacma is in the (Jul de Sac-Valle de Xeiba region of llispaniola; the species is rep- resented by large series and from numerous localities in this gen- eral region. From here, A. cJirysoIacDia extends westward to the vicinity of Leogane on the Tiburon Peninsula, and eastward as far as San Pedro de Macoris in the Pepublica Dominicana. From Leogane to San Pedro de ]\lacoris, there is a set of localities which implies a more or less continuous distribution. From this basic center, populations extend northAvestwarcl along the shore of the Golfe de la Gonave into the valley of the Riviere de I'Arti- lionite, and in the central portion of Haiti at least as far north as the vicinity of Mirebalais. In the Kepublica Dominicana, there is practical continuity of the main southern mass of the species north of the Sierra de Xeiba in the Valle de San Juan and thence to the Dominico-Haitian 1, order near Pedro Santana and immediately across the border at Cerca-la-Source in Haiti. A second major center lies in the northwestern portion of Haiti and extends thence eastward into the Valle de Cibao in the Republica Dominicana. Whether these two major populations are com- pletely isolated from one another is unknoAvn ; there is at least no contact in the Republica Dominicana, since the Cordillera Central stands between the two main regions, and the central valley along the eastern slopes of this range is unoccupied by chrysolaema. 430 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY There are several apparently disjunct populations in southern Haiti and the Republiea Dominicaua ; in the former country there is a single specimen from Aquin to the west along the Tiburon Peninsula, far removed from the nearest records from Leogane. In the Republiea Dominicana there are specimens from Juanillo near the eastern extremity of the island. Although there are no records from between San Pedro de Macoris and Juanillo, the species has presumably occurred in that region fairly re- cently (and may still occur as isolated populations), since it oc- curs on Isla Catalina and Isla Saona. It is at present unknown from the adjacent mainland in each case. There were two speci- mens reported (Cochran, 1941:282) from Ile-a-Vache but these cannot now be located ; although in two extended visits to this island we secured only the very abundant A. tacniura Cope, it is not altogether impossible that A. chrysolaema occurs there as well. Finally, there are populations south of the Sierra de Bao- ruco-Massif de la Selle which are completely cut off by these ranges from their more northern relatives ; this same phenomenon has now been noted in several other species of reptiles from this area. These mountains, as well as the virtually non-existent coastal plain along the eastern shore of the Peninsula de Bara- hona, form an inescapable trap for several reptiles in the south- ern portion of the Peninsula. For the remaining satellite islands, A. chrysolaema is repre- sented by abundant material from He de la Gonave and occurs also on He de la Tortue, Isla Beata, and the Siete Hermanos islands off the northwestern portion of the Republiea Domini- cana. Of the three species of Hispaniolan Anieiva, none has so broad a range as does A. chrysolaema. Nonetheless, there is geo- graphic evidence that even this species is retracting its range ; the apparent absence of chrysolaema from much of the south- eastern portion of the Republiea Dominicana, but its occurrence on off-shore islands in this region, suggests a formerly more widespread distribution in this area. A. chrysolaema is absent from the distal third of the Tiburon Peninsula in Haiti, and apparently from much of the central i^or- tion of that country as well. In the Republiea Dominicana it does not occur in the Cordillera Central nor the Sierra de Neiba, and is absent from the central and eastern portions of the country- except along the southeastern coast. Interestingly, despite its occurrence in the Valle de Cibao, it does not occur along the northcentral coast of the Republiea Dominicana. The Cordillera SCHWARTZ AND KLINIKOWSKI : AMEIVA 431 Septentrional acts as an effective barrier in this region. Much of the eastern Republica Dominicana is mesic (in fact, the area of highest rainfall in the country occurs in this region), and since A. chrysolaema is distinctly a lizard of xeric habitats, this one factor may well have prevented its expansion into this region and onto the Peninsula de Samana. However, its absence along the coast from Cabo Engaiio westward is strange, since this coast is arid and appears suitable for these lizards. Considering the disjunct nature of the populations of the species in extreme east- ern Hispaniola, it is possible that it never occurred in this region or that it has already retreated from this suitable coastal area. A. chrysolaema lives in xeric regions. It is abundant in the Cul-de-Sac plain below sea level and on the Peninsula de Bara- hona. It occurs also in the dry Valle de San Juan, at elevations of about 1000 feet. Although more tolerant of less xeric situa- tions than A. lineolaia, the two often occur together, with A. Uncolata inhibiting more open, cactus-studded, sandy regions, and A. chrysolaema preferring slightly more shady areas, such as adjacent copses or thickets of Acacia. If lowland woods are present, A. chrysolaema may invade them; the woods may not be dense nor with abundant ground cover. Maritime deciduous for- ests along the mangrove border (but usually not the mangroves themselves) offer a suitable habitat. Scrub-lands and open beaches with some cover are often adetiuate. In its relationships with the other two species, A. chrysolaema most often occurs with A. lineolata as noted above. On occasion, however, A. chrysola- ema occurs with the shade-loving A. taeniura. In such instances, chrysolaema appears to be tlie secondary invader of a habitat which is the preferred habitat of taeniura; in one such case near Oviedo on the Peninsula de Barahona, taeniura kept strictly to the open dry forest, whereas chrysolaema occurred almost exclu- sively along the edges of the woods where they abutted on a dry mangrove flat. The latter is likely the more preferred habitat of chrysolaema, but during the heat of the day this species was not averse to foraging in marginal forested situations. Characters studied We have examined a total of 1072 specimens of A. chryso- laema (in contrast to 42 examined by Barbour and Noble, 1915, and 198 by Cochran, 1941) ; of these, 655 are specimens collected by ourselves and parties at various times, and upon which we have extensive data on coloration and pattern. Of the races 432 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY recognized and described in the present paper, we have seen liv- ing or freshly killed specimens of all but two (the race from northwestern Haiti, and the race from extreme eastern Repvib- lica Dominicana). We feel that data on coloration and pattern are absolutely indispensable for any modern worker on the genus Ameiva. Old, faded, or (even worse) badly discolored specimens are completely useless except for scale counts and measure- ments, and analyses of populations must rest heavily and se- curely upon data from living or freshly killed animals. We have taken counts of rows of longitudinal and transverse ventrals, fourth toe subdigital scales, femoral pores, and scales in the fifteenth tail verticil (see Tables 1-4). Of these, only the numlier of transverse rows of ventrals can be used (partially) to characterize subspecies — i.e., having either 10 or 12 transverse rows of ventrals. Xo ]K)pulation has all specimens with 10 or all with 12 rows. However, there is most often a preponderance of one or the othei- in any particular sample, and we have used this modal number as typical of the race in question (Table 2), un- less the sample is rather small or the two categories differ by only a very few individuals. Of least value systematically is the number of longitudinal rows of ventrals. In the entire lot of A. chrysolacma examined, this figure varies from 33 to 41. The means for the 15 popula- tions described herein vary from 38.3 to 36.5. Data for longitu- dinal rows are presented in each case, but these data are not em- phasized. The counts of fourth toe scales, femoral pores, and fifteenth verticil scales show some trends, although in almost all cases the amount of overlap is rather large between most populations. Dif- ferences of means, however, between those races which rank first and last in each category may be rather striking (see Tables 2. 3 and 4). The largest difference between the highest and lowest populations is in fourth toe scales, where the high population has a mean of 91.0, and the low 77.8 — a difference of 13.2 scales. For facility Ave have combined in all cases the fourth toe scales from both feet and the femoral pore counts for both legs into one figure for each specimen ; we do not feel that this weakens the use of the data and it may well intensify slight differences, which would otherwise be almost unnoticeable. We have given the means and extremes for these three scale counts for each subspecies ; the differences, if any, obviously are mean differ- ences, since overlap of ranges is great in most cases. schwartz and klinikowski : ameiva 433 The Recognizable Subspecies Ameiva chrysolaema chrysolaema Cope, 1868 Ameiva chrysolaema Cope, 1868, Proc. Acad. Nat. Sci. Philadelphia, 20:127 (type locality — "He de la Gonave" = within 25 miles of Port-au- Prince, fide Cochran, 1941:275). Ameiva vittipunctata Cope, 1871, Proc. Acad. Nat. Sci. Philadelphia, 22:220 (type locality — "city of Santo Domingo" = within 25 miles of Port-au-Prince; see Cochran, 1941:275-76 for discussion of rationale for this restriction). Ameiva affinis Fischer, 1883, Beschreibungen neuer Eeptilien, [Separat- Abdruk aus dem] Osterprogramm des akademischen Gymnasiums, Ham- burg, p. 1 (type locality — "Haiti"). Diagnosis: A subspecies of A. chrysolaema characterized by a combination of very large size (males to 160 mm, females to 130 mm snout-vent length), usually 12 transverse rows of ventrals, moderate number of fourth toe subdigital scales, high number of femoral pores and of scales in the fifteenth tail verticil ; dorsal pattern consisting of about six longitudinal yellow lines and/or yellow dots arranged in series (Fig. 1, left), and a black gular band which may involve the chest and undersides of the arms. Distribution: From St. Marc (and including the "Artibonite Valley") on the north, southeast along the shore of the Golfe de la Gonave, east throughout the Cul-de-Sac plain to the environs of Etang Saumatre (Manneville, Ganthier, Fond Parisien), and west on the Tiburon Peninsula as far as the vicinity of Leogane (Pere) ; an isolated specimen from Aquin, Dept. du Sud, Haiti (Fig. 11). Discussion : A. c. chrysolaema is distinctly the largest and most bulky of the races of the species. Males reach a snout-vent length of 160 mm and females 130 mm. Color notes on a series from Eaux Gaillees in the Haitian Cul de Sac show the situation as far as coloration and pattern are concerned. Males from Eaux Gail- lees were noted as dark brown to reddish brown dorsally, espe- cially reddish on the shoulders and head (which may also be grayish). Lores and cheeks with gray blotches. The back has either a series of six yellow lines and yellow dots in the inter- spaces between the lines, or has six rows of lemon yellow spots. The lateral fields are black with or without a longitudinal series of yellow spots. The lower sides have large yellow spots as well, and the lateralmost belly plates are blue-spotted. The ventral ground color is dull blue-gray, the throat pale orange (Maerz and Paul, 1950, PL 9D7). There is a black gular band, 434 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Fig. 1. Left, Ameiva chrysolaema chrysolaema, ASFS X2162, 3.9 mi. NW Ganthier, Dept. de 1 'Quest, Haiti. Right, Ameiva c. umiratilis, holo- type, MCZ 77231, Baraliona, Barahona Prov., Eepublica Dominicana. which may expand posteriorly to cover the chest and under- sides of the arms. Some adult males have the shoulders and neck blackened, so that in these regions the yellow lines are much dulled and obscured. The females are like the males except that the lines are yellow and prominent anteriorly. The venter is dull blue-gray. The black gular collar is present but less pro- nounced than that of the males, and the throat is orange but paler than that of males. SCHWARTZ AND KLINIKOWSKI : AMEIVA 435 A series from Diquiiii to the ^vest of Port-au-Prince was col- ored much as the Eaiix Gaillees material. The dorsal ground color in males was brown with yellow dots arranged in lines or with yellow lines additionally present. There is a black lateral field with yellow spots. The sides of the head and axillae had vivid and prominent blue to blue-green blotches. The heads were dull reddish brown to dull orange, with orange-pink throats. The ventral ground color was grayish to orange with blue spots on the sides of the abdomen. The hindlimbs were dotted with yellow, the forelimbs with blue-green. Females resemble the males, but the dorsal lines or dots are less bright and prominent. From the above descriptions it is obvious that, despite some diiferences in details, these two populations (as well as many others throughout the range of chrysolaema) share a communitj' of dorsal markings — the longitudinal series of yellow^ lines and/or dots. Cochran (1941: pi. 8E) showed a dorsal view of the type of A. chrysolaema. The six dorsal lines, in this case partially fragmented into series of longitudinal dashes, are quite distinct. There is no doubt that the type of A. chrysolaema did indeed originate in the vicinity of Port-au-Prince, since only this subspecies occurs anywhere near that city. Specimens from He de la Gonave are much duller, less prominently marked, lack fragmentation of the dorsal lines, and are not referable to the nominate form. The longitudinal ventrals vary between 35 and 40 (mean 37.7) and these scales are most often arranged in 12 transverse rows (67.9 per cent), with 29.2 per cent having 10 transverse rows and 2.9 per cent having 11 rows. The variation in number of transverse rows depends primarily on whether the lateralmost of the enlarged ventral scales is sufficiently large to be con- sidered a ventral; we have so considered it if its length (longi- tudinally) is equal to that of the next inner adjacent row, and if its width (transversely) is equal to at least half that of the next inner adjacent row. Occasional specimens may also have one or two rows of the belly plates divided, thus attaining counts of 11 or 12 in another fashion. The fourth toe subdigital scales range from 76 to 101 (mean 86.7) and the femoral pores range from 33 to 50 (mean 43.7). The scales in the fifteenth caudal verticil vary between 37 and 51 (mean 44.4). Within the range ascribed to A. c. chrysolaema, there are vari- ous relatively minor pattern variants which we consider as 436 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY part of the normal variation of the subspecies. Specimens from St. Marc at the northwesternmost extreme of the range are very boldly lined longitudinally; the same is true of two indi- viduals from Manneville. In both cases there are specimens from adjacent localities or from the same locality which have more typical chrysolaema markings. The single individual from Aquin (USNM 72614) is unique in several (possibly significant) features. The longitudinal lines are composed of dots which also have a more or less transverse arrangement, so that the back has a rather conspicuous trans- versely banded appearance, a condition seen in no other A. c. chrysolaema. The black lateral field is obscured, and there is no black gular band, the entire throat, chest and undersides of the arms being unmarked. The specimen is an adult male with a snout-vent length of 146 mm, ventrals in 38 longitudinal and 12 transverse rows, 95 subdigital fourth toe scales, 41 femoral pores, and 43 scales in the fifteenth caudal verticil. None of these counts will distinguish the specimen from A. c. chrysolaema, although we attach no particular significance to this fact. Considering the wide gap between the known localities of A. chrysolaema between Pere near Leogane and Aquin (a distance of some 90 kilometers) and the fact that the Aquin specimen comes from the southern, in contrast to the northern, side of the Tiburon Peninsula, it is likely that this single individual comes from a population which is distinct from A. c. chrysolaema. AVithout additional material, and especially lacking careful data on col- oration and pattern in life, we are unwilling to name this single Aquin specimen as distinct from A. c. chrysolaema. The character ascribed by Cochran (1941:277) to differenti- ate A. c. affinis — i.e., the interparietal being larger than the adjacent scales — we find to be completely untenable. Dr. Coch- ran has also shown (1941:291) that the scale counts of affinis fall within the known range of A. c. chrysolaema. Specimens which she assigned to affinis were reported (1941:292) from Momance, Manneville, and Pere in Haiti. Other specimens (p. 282) from Manneville were assigned to c. chrysolaema. In a series of twenty-seven specimens from Fond Parisien and the eastern Cul de Sac, for example, six have the interparietal larger than adjacent scales and the balance have the interparietal smaller. The same situation applies to series from other locali- ties within the range of A. c. chrysolaema, and if we accept affinis as differentiated by this character alone, then the races SCHWARTZ AND KLINIKOWSKI : AMEIVA 437 chrysolaema and affinis are broadly and randomly sympatric. No features of pattern or coloration will distinguish specimens with smaller interparietals from those with larger interparietals, and we cannot detect any other constant scale feature which will distinguish two forms in this region. For this reason we con- sider affinis Fischer as a synonym of chrysolaema Cope. The reasons for considering vittipunctata Cope as a synonym of chrysolaema have been outlined in the introduction. Specimens examined: Haiti, Dept. de I'Artihonite, "Artibonite Valley" (not mapped), 1 (USNM 75921) ; St. Marc, 5 (USNM 59079, MCZ 58012-13, 65351, AMNH 49766) ; Bept. de VOuest, 2.2 mi. (3.5 km) SW Trou Forban, 1 (ASFS X1927) ; between Arcahaie and Trou Forban, 1 (MCZ 51433) ; 6.3 mi. (10.1 km) NE Arcahaie, 3 (ASFS X1928, X1930, X1946) ; 13 mi. (20.8 km) SW Arcahaie, 1 (ASFS X1938) ; Port-au-Prince, 12 (AMNH 49637-38, MCZ 13839, 59495-502, 69420) ; Carrefour Feuille, Port-au-Prince (not mapped), 1 (MCZ 65810) ; Delmas, 2 (MCZ 65808-09) ; Petionville, 1 (USNM 59078) ; 10 mi. (16 km) SW Port-au-Prince, 1 (UMMZ 92197) ; 3.5 mi. (5.6 km) E Croix des Bouquets, 15 (ASFS X2197-211) ; Eaux Gaillees, 33 (ASFS X1651-83) ; Manneville, 11 (MCZ 8621-23, 8625, 8629-33, 8614, 8618) ; 3.9 mi. NW Ganthier, 18 (ASFS X2153-70) ; 1.3 mi. (2.1 km) NW Fond Parisien, 3 (ASFS X2174-76) ; 0.4 mi. (0.6 km) SE Fond Parisien, 7 (ASFS X2189-95) ; Hatte Latham (not mapped), 1 (MCZ 51424); Diquini, 36 (ASFS X2381-407, MCZ 6292, 8706, 8649-51, 8653-54, 8658-59) ; Momance, 8 (MCZ 8634-35, 8638-41, 8649, 20875) ; Qa Ira, 2 (MCZ 64919-20) ; Pere, 3 (MCZ 13271-73) ; Dept. du Sud, Aquin, 1 (USNM 72614). Ameiva chrysolaema umbratilis,^ new subspecies Ilolotype: MCZ 77231, a subadult female, from Barahona, Barahona Province, Republica Dominicana, taken 25 July 1963, one of a series taken by native collectors. Original number X9721. Paratypes: All from the Republica Dominicana, as follows: MCZ 81000-04, USNM 152558-60, KU 79861-64, UIMNH 56886- 89, RT 738-39, same data as holotype; ASFS X9568-69, Bara- hona, Barahona Prov., 24 July 1963, native collector; AMNH 37943-49, 38133-39, Barahona, iBarahona Prov., 12 October 1922, 1 From the Latin for "remaining in the shade." 438 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY G. K. Noble; AMNH 63191, 63193, Barahona, Barahona Prov- ince, 10-19 July 1932, W. G. Hassler; MCZ 58019, Barahona, Barahona Prov., 13 July 1932, W. G. Hassler; MCZ 43813-14, Barahona, Barahona Prov., 18 July 1932, W. G. Hassler ; ASFS V199-200, 4 km NW, 1 km SW Barahona, Barahona Prov., 5 August 1963, A. Schwartz, R. Thomas; AMNH 49837-38, "Palo- mino Spring's, nr. Barahona" (not mapped), Barahona Prov., 15 August 1935, W. G. Hassler. Associated specimens : Rcpuhlica Dominicana: Independencia Prov., 6.5 mi. (10.5 km) NE Jimani, 1 (ASPS X9507) ; 4.4 mi. (7.0 km) SE Jimani, 3 (ASFS X9515-17) ; 13 km SW La Descubierta, 3 (ASPS X9364-66) ; 5 km E La Descu- bierta, 5 (ASFS X9354-58) ; Las Baitoas, 1 (MCZ 58776) ; 22 km SE Duverge, 7 (ASFS X9928-34) ; 1 km W El Naranjo, 1000 feet, 2 (ASFS X9943-44) ; northv^^est side, Laguna del Rincon, 1 (MCZ 58779) ; Guayabal, 9 km N Postrer Rio, 2 (MCZ 57732- 33) ; Baoruco Prov., Jaragua, 5 (ASFS X9469-72, RT 713) ; 0.7 mi. (1.2 km) E El Estero, 2 (ASFS X9467-68) ; 0.8 mi. (1.3 km) SW Neiba, 15 (ASFS V252-64, RT 775-76). Diagnosis: A subspecies of A. chrysolacma characterized by a combination of moderate size (males to 130 mm, females to 112 mm snout-vent length), usually 10 (but often 12) transverse rows of ventrals, moderate number of fourth toe subdigital scales, low number of femoral pores, and high number of scales in the fifteenth caudal verticil ; dorsal pattern consisting of dull grayish brown to greenish black dorsal ground color with a series of eight to ten dorsal longitudinal lines composed of small and numerous dull yellowish to tan dots (Fig. 1, right), and a black gular band which in adult males may involve the chest and undersides of the arms. Distribution : The Valle de Neiba from just east of Jimani to the vicinity of tlie city of Barahona, Republica Dominicana (Fig. 11).' Description of type: A subadult female with the following measurements and counts : snout-vent length 93 mm, tail 184 mm ; ventrals in 37 longitudinal and 10 transverse rows ; fourth toe subdigital scales 39 and 41 (total 80) ; femoral pores 18 and 17 (total 35) ; 43 scales in the fifteenth caudal verticil. Dorsal ground color grayish brown in life, head gray, shoulders greenish ; ten rows of dull yellowish dorsal dots, the dots in each series virtually confluent with one another, giving a vague wavy line ; lateral fields obscure darker gray with scattered buffy dots. SCHWARTZ AND KLINIKOWSKI : AMEIVA 439 Throat pale purplish orange, followed by a black gular band which extends slightl}' onto the chest and onto the underside of the f orelimbs ; ventral ground color grayish blue ; tail dull grayish brown above and grayish blue below, with some scattered darker scales dorsally. Variation: See tables. The characters of umhratilis are best expressed in populations from the eastern section of the Valle de Neiba ; however, even the most western specimens from the vicin- ity of Jimani are in no way comparable to Haitian A. c. chryso- laenia. The dorsal ground color was noted in life as being green- ish black (Jimani, La Descubierta, Jaragua), greenish brown (Duverge, El Naranjo), brown (El Estero), and grayish brown (Barahona). The dots in the dorsal longitudinal lines are tiny and very often confluent, giving almost a vermiculate appearance to the dorsal band; the dots vary in color from greenish (Jimani), creamy (La Descubierta), pale yellow to pale green (Duverge), yellow (El Estero), grayish yellow (Jaragua), or dull yellowish to tan (Barahona). The lateral fields are usually dull and inconspicuous, hardly darker than the lateral coloration ; they often include a row of yellow to creamy spots and are not outlined either above or below by pale and prominent longitudinal lines. The ventral coloration varies from pinkish gray and grayish orange to grayish blue, with specimens having the brighter colors known from the western extremity of the range. The throats are dull pinkish gray and grayish orange to dull purplish or dull orange. The l)lack gular band is invariably present and may, in adult males, expand to cover much of the chest and anterior abdomen and underside of the forelimbs. The upper surfaces of the limbs are usually unspotted, but if there are a few scattered dots these are blue on the forelimbs and yellow on the hindlimbs. Comparisons: The coloration and pattern of chrysolaema and xmhratilis are strikingly different; even in the western portion of the range of umhratilis, no specimen approaches closely the vivid dorsal coloration and pattern of the nominate race. The extreme condition in the eastern Valle de Xeiba contrasts strongly with the condition at Fond Parisien, for instance, and specimens from Jimani and La Descubierta are much more like individuals from Barahona in having fine dorsal dotting and generally more drab colors than they are to specimens from Fond Parisien. Umhratilis is a smaller lizard; no specimen of cither sex of this race achieves the much bulkier and larger 440 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY size of chrysolaema. This is certainly not a sample artifact since large series of both forms are at hand. The tendency for vmhra- tilis to have 10 versus 12 transverse rows of ventrals, as in chrysolaema, is of interest, although in umhratilis the specimens are almost equally divided between the 10- and 12-row condi- tions. In number of longitudinal ventral rows, the two races are comparable {chrysolaema 37.7, umhratilis 37.1). In number of femoral pores these two races differ strongly, with a mean of 43.7 in chrysolaema and 35.6 in umhratilis. Chrysolaema aver- ages slightly higher in counts of femoral pores and fifteenth verticil scales. Remarks: The occurrence of two very distinct races of A. chrysolaema in the Cul de Sac-Valle de Neiba complex is sur- prising. Aside from the more mesic eastern and western ends of this long xeric valley, the conditions throughout are quite com- parably severe. Interestingly, the Valle de Neiba is greatly con- stricted just to the east of Jimani ; it is possible that this narrow neck (7-10 km) has been effective in separating these two races. Specimens from the Republica Dominicana to the northwest of Jimani may well be assignable to A. c. chrysolaema. Ameiva chrysolaema boekeri Mertens, 1938 Amciva chrysolaema hoeTceri Mertens, 1938, Senckenbergiana, 20:338 (type locality — south of Foiido Negro, lower Rio Yaque del Sur, Bara- hona Province, Republica Dominicana). Diagnosis: A subspecies of A. chrysolaema characterized by a combination of moderate size (males to 126 mm, females to 111 mm snout-vent length), usually 10 transverse rows of ven- trals, moderate number of fourth toe subdigital scales, low num- ber of femoral pores, and high number of scales in the fifteenth verticil ; dorsal pattern of two phases : ( 1 ) back yellowish brown, grayish tan, to olive, and without pattern and often without any indication of lateral fields, or (2) dorsum colored as above but with faint paler marblings or longitudinal lines and a fairly prominent black to dark gray lateral field (Fig. 2), and a black gular band which may involve the chest and underside of the arms. Distribution: North of the Rio Yaque del Sur in extreme eastern Valle de Neiba, north and east to north of Azua and east to the vicinity of Bani, in the Llanos de Azua, Republica Dominicana ; intergrades with the next subspecies to the north- west in the vicinity of Hato Nuevo, Azua Province (Fig. 11). SCHWARTZ AND KLINIKOWSKI : AMEIVA 441 Fig. 2. Left, Ameiva c. hoekeri, ASFS X7811, 10 mi. NW Bani, Peravia Prov., Eepiibliea Dominicana. Eight, Ameiva c. boeTceri, ASFS V689, 15.2 mi. S San Jose de Ocoa, Peravia Prov., Eepubliea Dominicana. Discussion: A. c. hoekeri was described on the basis of four- teen lizards from Fondo Negro. Of these, seven males were dorsall}^ patternless (including the type), four males showed a " chriisolaema"-\ike pattern, and the final male was considered by Mertens (1939:72) to resemble the Beata race ahhotti — i.e., it was dorsally dotted. The two paratypic females were " chryso- toema "-like as well. We have examined a single paratype of hoekeri (MCZ 44757) and eighty-six other specimens from the range ascribed by us to hoekeri above. Of ten localities, only 442 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY two have "pure" or almost "pure" hoekeri populations (i.e., patternless), viz., a series of five specimens from the west side of Punta Martin Garcia, Barahoua Province, and a series of twenty-one from 10 mi. XW Bani, Peravia Province. Additional specimens (four) from the eastern side of Punta Martin Garcia show the typical duality of dorsal pattern, however, and thus the uniformity of this small series of five is not significant. The large series from Bani, on the other hand, has only a single adult male which shows any pattern; this population is thus almost completely patternless. A fresh series of five topotypes from Fondo Negro has four individuals with patterns, and one without pattern. Thus, although the patternless condition pre- dominates at some localities (Bani), elsewhere (and including the type locality) both types of dorsal pattern occur. The two types of dorsal patterns, as delimited by jMertens, grade into one another. The back may be longitudinally lined with from six to ten tan to yellowish lines of fine dots, or these may be very obscure centrally and more prominent at the sides of the dorsal field. Some individuals have the back finely marbled. The lateral fields are well developed and enclose a series of buffy to cream dots; the lateral fields are often outlined below by a longitudinal yellowish line, and a similar line may border these fields above. The lower sides may be dotted with blue or bluish green. The dotted condition of the back, considered by Mertens as being ahhotti-like, is not at all comparable to the large and brilliant sky-blue spotting on a black ground of that race. There is some similarity between patterned hoekeri and nmhratUis. No nmhratilis however is unpatterned. In patternless lizards, the dorsal ground color was recorded as yellowish brown (Fondo Negro), brown (San Jose de Ocoa), grayish tan (Punta Martin Garcia), black (Barreras), and reddish brown (Bani). The lateral fields may be completely absent or may be indicated by a somewhat grayer longitudinal lateral stripe, without any sort of included or adjacent pale dots. The ventral ground color is blue-gray, purplish blue, blue, light olive, gray, or orange-gray. The throat is likewise variable, but is some shade of dull orange; females have throats which are typically more grayish orange than males. The tails are brown to grayish tan above, and gray below. Scale counts for the series (including intergrades from Hato Nuevo) are: longitudinal ventrals 34-40 (mean 37.2), trans- verse ventrals in 10 (82.4 per cent) or 12 (17.6 per cent) rows, SCHWARTZ AND KLINIKOWSKI : AMEIVA 443 fourth toe scales 73 to 98 (mean 84.6), femoral pores 31-41 (36.5), fifteenth verticil 37-48 (mean 42.7). The much smaller size and faded pattern of those hoekeri which have patterns, as well as the patternless individuals, can easily be distinguished from A. c. chrysolaema. Chrysolaema is likewise characterized by 12 rather than 10 rows of ventrals. In counts of fourth toe scales, femoral pores, and fifteenth verti- cil scales, hoekeri averages less than chrysolaema, the most striking difference being in femoral pores {chrysolaema 43.7, hoekeri 36.5). Patternless hoekeri can be easily differentiated from all umhratilis, since this race is never patternless. Pat- terned hoekeri are much like umhratilis. In both, the dorsal pat- tern is faded and not well demonstrated. One feature is sugges- tive ; patterned hoekeri have the lateral fields prominent and often outlined both above and below, whereas the typical umhra- tilis condition is an obscure lateral field, not set off by longi- tudinal pale lines. Both umhratilis and hoekeri usually have 10 rows of ventrals, although umhratilis has a much higher fre- quency of 12-row individuals. In counts of fourth toe scales and femoral pores, hoekeri averages slightly higher than um- hratilis; the means for fifteenth verticil scales are identical. Because of the similarities of umhratilis and patterned hoekeri, we have considered the possibility that the name hoekeri should be applied to Ameiva from the Yalle de Neiba. To be counted against this conception is the fact that of 61 umhratilis, none is unpatterned, whereas 56.3 per cent of the specimens (hoekeri) from north of the Kio Yaque del Sur are patternless. There is no indication of this patternless condition in specimens from Barahona, nor from elsewhere in the range of umhratilis. We prefer to regard hoekeri as a separate entity, distinct from um- hratilis to the south. A. c. hoekeri is approached geographically by three adjacent races; of these it is known to intergrade only with the race to the northwest in the Valle de San Juan (these intergrades will be discussed later). From umhratilis the range of hoekeri is sep- arated by the lower reaches of the Rio Yaque del Sur and by the extremely mesic conditions of much of the eastern portion of the Valle de Neiba. Although hoekeri is not presently known to intergrade with the race next to the east along the southern coast of the Republiea Dominicana, it may well do so. It is cer- tainly significant that hoekeri occupies the western Llanos de 444 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Azua in the vicinity of Ban! ; just to the east of that city, con- ditions become more mesic, and the region is occupied by another race. Intergradation likely takes place where these two regions come into contact. Specimens examined: Repuhlica Dominicana, Barahona Prov., Fondo Negro, 6 (ASFS X9703-07, MCZ 44757); west side, Punta Martin Garcia, 5 (ASFS V84-88) ; Azua Prov., 3 km E Barreras, 2 (ASFS V3164-65) ; 2 km W Puerto Viejo, 2 (ASFS V3183-84) ; 22 km NW Azua, 3 (ASFS V465-67) ; 1.8 mi. (2.9 km) W, 1.1 mi. (1.8 km) N Azua, 18 (ASFS X8002-18, X8103) ; 1.8 mi. (2.9 km) W, 2.7 mi. (4.3 km) N Azua, 10 (ASFS X8019-28); Peravia Prov., 8.9 mi. (13.9 km) S San Jose de Ocoa, 1300 feet (430 m), 1 (ASFS V714) ; 15.2 mi. (24.3 km) S San Jose de Ocoa, 9 (ASFS V687-95) ; 10 mi. (16 km) NW Bani, 23 (ASFS X7801-21, RT 613-14). Intergrades between A. c. hoekeri and the race to the northwest were examined from : RepuMica Dominicana, Azua Prov., Hato Nuevo, 10 (ASFS X437-46). Ameiva chrysolaema alacris,^ new subspecies Holotype: MCZ 77232, an adult male, from 10 km SE San Juan, San Juan Province, Republica Dominicana, one of a series taken 9 August 1963 by Albert Schwartz and Richard Thomas. Original number V283. Paraiypes: All from the Republica Dominicana, as follows: ASFS V284-97, same data as holotype ; RT 778-79, 10 km S San Juan, San Juan Prov., 9 August 1963, R. Thomas ; MCZ 81005-06, USNM 152561-64, AMNH 92842-43, KU 79865-66, UIMNH 56890-93, 2.5 km W, 4.4 km S San Juan, San Juan Prov., 9 August 1963, D. C. Leber, R. F. Klinikowski ; KU 79867-68, AMNH 92844, 2.5 km W, 5.4 km S San Juan, 9 August 1963, D. C. Leber, R. F. Klinikowski ; ASFS V389-90, 10 km E Valle- juelo, San Juan Prov., 12 August 1963, R. Thomas; USNM 152565, 3 km E Las Matas, San Juan Prov., 9 August 1963, R. Thomas. Associated specimens: Haiti, Dept. du Nord, Cerca-la-Source, 1 (USNM 76780) ; Repuhlica Dominicana, San Rafael Prov., 3.8 mi. (6.1 km) SE Sabana Cruz, 1 (ASFS V329) ; Guayabal, 1 (MCZ 58672); Azua Prov., Tubano (= Padre las Casas), 3 (USNM 66729-31) ; 0.7 mi. (1.1 km) NW Villarpando, 9 (ASFS V419-27). 1 Prom the Latin for "lively." SCHWARTZ AND KLINIKOWSKI : AMEIVA 445 Diagnosis: A subspecies of A. chrysolaema characterized by a combination of moderate size (males to 126 mm, females to 109 mm snout-vent length), usually 10 transverse rows of ventrals, moderate number of fourth toe subdigital scales, very low num- ber of femoral pores, and high number of scales in fifteenth caudal verticil ; dorsal pattern consisting of five to seven bold longitudinal dorsal lines (the lateralmost forming a strong upper border to the prominent black lateral fields with their enclosed bright yellow dots), the longitudinal lines never broken into dots and lines as in c. chrysolaema and always conspicuous and discrete (Fig. 3, left), and a black gular band which rarely in- volves also the chest and undersides of the arms. Fig. 3. Left, Ameiva c. alaeris, holotype, MCZ 77232, 10 km SE San Juau, San Juan Piov., Republica Dominicana. Eight, Ameiva c. proeax, holotype, MCZ 77233, Santo Domingo, 2.2 km SW Rio Ozama, Dist. Nac, Eepublica Dominicana. 446 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY Distribution: From east central Haiti (Cerca-la-Source) south- eastward through the Valle de San Juan (Fig. 11) ; intergrading with A. c. hoekeri at Hato Nuevo, Azua Province, and with A. c. chrysolaema in the vicinity of Mirebalais, Dept. de I'Ouest, Haiti (see discussion below). Description of type: An adult male with the following meas- urements and counts: snout-vent length 116 mm, tail 200 mm; ventrals in 37 longitudinal and 12 transverse rows; fourth toe subdigital scales 42 and 44 (total 86) ; femoral pores 16 and 16 (total 32) ; 46 scales in the fifteenth caudal verticil. Dorsal ground color brown with seven longitudinal pale yellow lines, the lateralmost bordering above the black lateral fields with their isolated yelloAv dots; lateral fields bordered below by a slightly duller longitudinal line which contains a series of bright yellow dots; lower sides dotted with yellow. Throat gray, venter dull, dirtj^ orange. A black gular band, not extending onto the chest or undersides of the arms. Tail brown above, gray below, with an indistinct proximal and lateral area of yellowish dots; top of tail with some darker brown scales. Fore- and hindlimbs with pale scattered small dots, bluish on forelimbs and yellowish on hindlimbs. Variation: See tables. A. c. alacris presents a constant as- semblage of coloration and pattern elements throughout its range. The dorsal ground color is alwaj^s brown, with from five to seven pale or dull yellow longitudinal lines, these lines always forming a conspicuous pattern. The lines are entire and not fragTiiented or modified into series of longitud- inal dots, although in some specimens the more central lines, especially posteriorly, may be broken into dashes. In general, however, the integrity of the lines (even when fragmented) is maintained. The black lateral fields are bold, set off by pale longitudinal lines above and below, and enclose a single series of scattered yellow dots ; the lower line bordering the lateral field may have superimposed upon it a series of bright yel- low dots, thus rendering the black lateral field even more con- spicuous. The lower sides are dotted with yellow. The throat varies from gray to very pale orange, and the venter likewise varies between these two extremes. A. c. alacris intergrades to the southeast with A. c. hoekeri and to the southwest with A. c. chrysolaema. A series of ten specimens from Hato Nuevo, Azua Province (ASFS V437-46), SCHWARTZ AND KLINIKOWSKI : AMEIVA 447 shows the intergradatiou with hoekeri. Of this series, five are un- patterned hoekeri, and five represent the patterned phase of that race. These five patterned lizards have the lateral fields darker than most patterned hoekeri, and there is a distinct tendency to have the dorsal lines more boldly (brighter yellow) displayed anteriorly, although the posterior dorsal pattern is fainter and very like "typical" patterned hoekeri. The distance from Hato Nuevo {alacris X hoekeri intergrades) to Villarpando {alacris) is only ten kilometers, yet the series from the latter locality is typical of alacris in all waj^s and has no patternless individuals. From the vicinity of Mirebalais, Dept. de I'Ouest, Haiti, we have examined specimens from the following localities: 3.4 mi. (5.4 km) NE Barrage de Peligre, 2 (ASFS X2217-18) ; 1.1 mi. (1.8 km) S Mirebalais, 3 (ASFS X2237-39) ; Mirebalais, 1 (MCZ 68510) ; La Tombe, nr. Mirebalais, 8 (MCZ 68517-24) ; Boudou, nr. Mirebalais, 2 (MCZ 69387-88) ; Duvie, nr. Mirebalais, 1 (MCZ 68478). Of these, the last three places named. La Tombe, Boudou, and Duvie, cannot be located ; they have not been mapped. Taken as a whole, this lot of lizards is intermediate between chrysolaema and alacris, although they are closer to alacris than to chrys- olaema. Three lizards (ASFS X2237, ASFS X2217, MCZ 68510) show the disintegration of the dorsal lines into series of yellow spots, a typical chrysolaema feature. Several male specimens are larger than alacris, with snout -vent lengths of 133 to 145 mm (five lizards) ; two females have snout-vent lengths of 109 mm (the upper extreme of alacris females), and another has a snout- vent length of 110. In life, our specimens from Barrage de Peligre and Mirebalais had yellow lines and a greenish wash on the neck — the latter a chrysolaema character. Finally, some individuals have a discrete black gular band as in alacris, whereas others have the band expanded onto the chest and arms as in chrysolaema. We consider this entire lot of specimens inter- gradient between alacris and chrysolaema. Comparisons: A. c. alacris is easily distinguished from the three previously described races on the basis of dorsal pattern ; the discrete, bold, and undotted longitudinal lines of alacris contrast with the patternless or weakly patterned races hoekeri and umhratilis, and with the larger and dorsally dotted and lined chrysolaema. Alacris is a race with ten transverse rows of ventrals as are umhratilis and hoekeri, in contrast to the twelve- rowed chrysolaema. In fourth toe scales, alacris (84.8) averages close to hoekeri (84.6), slightly higher than umhratilis (83.0) 448 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY and lower than chrysolaema (86.7). In femoral pores, alacris has the lowest average (33.8) of any race of A. chrysolaema; of the described forms, it is approached by umhratiUs (35.6) and hoekeri (36.5) and is far below chrysolaema (43.7). In scales in the fifteenth caudal verticil, on the other hand, both alacris and chrysolaema are high (44.0 and 44.4), with hoekeri and unih rat ills (both 42.7) lower. Remarks: A. c. alacris occupies the high and xeric Valle de San Juan and associated upland foothills (i.e., Tubano). It is likely that it is more widespread in east-central Haiti than the present evidence of one specimen from Cerca-la-Source indicates. Presumabl}^ the race follows down the valley of the Riviere de I'Artibonite, and in the vicinity of Mirebalais has genetic con- tact with the more southern chrysolaema. Such genetic con- tinuity may come across the Montagues de Trou d'Eau from the Cul de Sac (although there is no obvious means of penetration of this mountain mass), or chrysolaema may reach Mirebalais via the valley of the Artibonite from the St. Marc area. The only evidence for the latter is the single specimen noted under A. c. chrysolaema, from the "Artibonite Valley"; this individual may have come from some undetermined locality which is inter- mediate between St. Marc and Mirebalais. Further collecting in these areas should easily reveal the precise place of contact between these two races. AmEIVA CHRYSOLAEMA PROCAX^ lieW SUbspCcics Holotype: MCZ 77233, an adult male, from Santo Domingo, 2.2 km SW of the Rio Ozama, Distrito Nacional, Repiibliea Dominicana, one of a series taken 14 June 1963 by Ronald F. Klinikowski, David C. Leber, and Richard Thomas. Original number X7714. Paratypes: All from the Republiea Dominicana, as folloAVS: ASFS X7711-13. X7724-26, MCZ 81007-10, USNM 152566-70, KU 79869-71, UIMNH 56894-97, RT 605-08, same data as holo- type ; ASFS X9254-56, Santo Domingo, old airport, Distrito Nacional, 17 July 1963, D. C. Leber, R. Thomas; AMNH 92845- 49, 5.9 km W Santo Domingo, Distrito Nacional, 20 June 1964. D. C. Leber, R. Thomas. Associated specimens: Rcpuhlica Dominicana, San Cristobal Prov., 8.4 mi. (13.6 km) NE Sabana Grande de Palenque, 2 1 From the Latin for "bold." SCHWAETZ AND KLINIKOWSKI : AMEIVA 449 (ASFS X8167-68) ; 4.2 mi. (6.7 km) NE Sabana Grande de Palemiue, 22 (ASFS X8149-66, RT 643-46). Diagnosis: A subspecies of A. chrysolaenia characterized by a combination of large size (males to 141 mm, females to 116 mm snout-vent length), usually 10 (but often 12) transverse rows of ventrals, moderate number of fourth toe subdigital scales, low number of femoral pores, and high number of scales in the fifteenth caudal verticil ; dorsal pattern a series of six or seven longitudinal yellow lines in a reddish brown field, the lines usu- ally wavy or broken into a series of longitudinal dashes (Fig. 3, right), and a black gular band which may be so expanded as to involve the entire ventral surface including the undersides of the arms. Distribution: The coastal regions of San Cristobal Province and the Distrito Nacional, from the Rio Ozama on the east to the vicinity of Sabana Grande de Palenque on the west (Fig. 11) ; presumed to occur inland, since individuals were seen crossing the road near the city of San Cristobal. Description of type: An adult female with the following measurements and counts : snout-vent length 114 mm, tail 275 mm ; ventrals in 38 longitudinal and 12 transverse rows ; fourth toe subdigital scales 46 and 46 (total 92) ; femoral pores 18 and 20 (total 38) ; 44 scales in the fifteenth caudal verticil. Dorsal ground color rich reddish brown in life, with a series of seven dull longitudinal lines, the median line rather obscure, the lateral lines broken into a series of wavy dashes, the lateralmost lines bordering above the black lateral fields, which have a series of tiny yellow dots inclosed within them. Sides of head gray with a creamy subocular patch and some pale irregular blue blotches. Throat fleshy gray. Ventral color dark blue-gray. Hindlimbs heavily spotted with yellow, forelimbs faintly spotted with blue. A dark gray gular band not involving the chest and underside of the arms. Tail reddish brown above, dark gray below. Variation: See tables. The series of A. c. procax is remark- ably uniform in both coloration and pattern. The dorsal ground color is always some shade of reddish brown, and there may be a yellowish green wash over the shoulders. The longitudinal lines are conspicuous, although the median line may be reduced or faint. Only in young and subadult individuals are the lines entire, and even in these specimens there is a strong tendency toward wavy fragmentation. The lateral fields are black and the yellow dots included therein are regularly very tiny; in 450 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY females these included dots are less well differentiated than in the males. The gular band is present, and in large males may be very extensive, including most of the belly. The venter is blue in juveniles and females, and gray to solid black in males. Comparisons: The reddish brown dorsal coloration and the wavy dorsal lines separate procax from all other races. Some specimens of chrysolacma were noted in life as reddish brown, but this is not the usual dorsal coloration of the nominate race. In- dividuals of chrysolaema with this coloration are readily differen- tiated from procax on the basis of the very different dorsal pat- terns of the two subspecies. In size, procax is closest to chrysolaema, although procax is dis- tinctly the smaller of the two. Chrysolacma is typically a subspe- cies with 12 rows of ventrals, whereas procax, although it has in- dividuals with 12 rows, has a modal condition of 10 rows. Procax averages fewer (84.8) fourth-toe scales than chrysolaema (86.7), the same as alacris, and more than hoekeri (84.6) and umhratilis (83.0). In number of femoral pores, procax (36.1) averages far lower than chrysolacma (43.7), and slightly lower than hoekeri (36.5), but slightly higher than umhratilis (35.6) and alacris (33.8) ; procax is one of the races with a low^ number of femoral pores. In fifteenth verticil scales, procax averages less (42.8) than chrysolaema (44.4) and alacris (44.0), and is about equal to hoekeri and umhratilis (42.7). Remarks: A. c. procax is presently not known to intergrade either with hoekeri to the west or the race next to the east along the southern Hispaniolan coast. The easternmost hoekeri locality (where the population incidentally is almost completely made up of non-patterned individuals) and the westernmost locality of procax are separated by onh' 35 kilometers. As noted in the dis- cussion of hoekeri, Bani lies about on the dividing line between the xeric Llanos de Azua to the west and more mesic conditions on the east. Procax inhabits these eastern more mesic regions, whereas hoekeri is restricted to the xeric Llanos de Azua. Ameiva CHRYSOLAEMA PARVORis,^ new subspecics Holotype: MCZ 77234, an adult male, from 0.9 mi. (1.4 km) E Boca Chica, Distrito Naeional, Republica Dominicana, one of a series taken 23 x\ugust 1963 by Ronald F. Klinikowski, David C. Leber, and Richard Thomas. Original number V649. 1 From the Latin parvum (small) and os, oris (mouth), a translation of Boca Chica. the type locality. SCHWARTZ AND KLINIKOWSKI : AMEIVA 451 Paratypes: All from the Republica Dominicana, as follows: MCZ 81011-14, USNM 152571-74, A:MNH 92850-55, KU 79872-74, UIMXH 55638-39, RT 789-90, same data as liolotype; ASFS V669-79, Boca Cliica, eastern edge, Distrito Nacional, 23 August 1963, R. F. Klinikowski, D. C. Leber, R. Thomas. Associated specimens: Repuhlica Dominicana, San Pedro de Macoris Pro v., 0.5 mi. S San Pedro de Macoris, at lighthouse, 11 (ASFS X8181-91) ; La RomanaProv., Isla Catalina, western end, 4 (ASFS V554-57). Diagnosis: A subspecies of A. chrysolaema characterized by a combination of large size (males to 137 mm, females to 113 mm snout-vent length), usually 12 transverse rows of ventrals, mod- erate number of fourth-toe subdigital scales and femoral pores, and high number of scales in the fifteenth tail verticil; dorsal pattern of tan to blackish brown ground color with dull yellow spots which may be either discrete or confluent, giving a reticu- late appearance (Fig. 4, left), lateral fields present and black, or broken to give a tigroid effect, and a black gular band which may be expanded to involve the chest and the undersides of the arms. Distribution: Coastal southeastern Republica Dominicana, from Boca Chica on the east to San Pedro de Macoris on the west, and including Isla Catalina ; range as here described apparently discontinuous, and A. chrysolaema unknown from the mainland opposite Isla Catalina (Fig. 11). Description of type: x\n adult male with the following meas- urements and counts : snout-vent length 129 mm, tail 282 mm ; ventrals in 38 longitudinal and 10 transverse rows; fourth-toe subdigital scales 41 and 42 (total 83) ; femoral pores 19 and 19 (total 38) ; 44 scales in the fifteenth tail verticil. Dorsal ground color dull blackish brown, with the entire back from the neck to the sacrum covered with dull yellow spots, not aligned into linear series ; lateral fields black, not bordered above or below, and in- vaded by the brown dorsolateral coloration, the entire sides spot- ted with yellow dorsally and pale blue ventrally. Throat gray, ventral ground color gray. A black gular band which extends slightly onto the chest and also the underside of the arms. Fore- limbs faintly spotted with bluish, hindlimbs spotted with dull yellow. Tail grayish brown dorsally, gray ventrally. Variation: See tables. The dorsal ground color of A. c. par- voris varies from blackish brown (type locality) to tan and brown (San Pedro de Macoris). The spotted condition of the black is typical of most specimens, including small juveniles. 452 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Fig. 4. Left, Ameiva c. parvoris, holotype, MCZ 77234, 0.9 mi. E Boca Chica, Dist. Xac, Eepublica Domiuicana. Eight, Ameiva c. jacta, holotype, MCZ 75267, Juanillo, La Eouiana Prov., Eepublica Dominicana. but others show a more lineate pattern somewhat like that described for procax — the dorsal lines broken into dashes giving a wavy appearance. The lateral fields maj^ be obscured as in the type, or may be slightly more prominent, with much encroachment of brown to give an irregular and indefinite edge ; at the same time the black lateral field pigment may extend up onto the sides of the back, thereby giving a distinctly tigroid appearance to the sides and dorsolateral regions. In some indi- viduals the dorsal spots are confluent, thereby increasing the tigroid effect by transverse dorsal pale markings. The throat is SCHWARTZ AND KLINIKOWSKI : AMEIVA 453 grayish to dull orange and the ventral ground color varies from grayish blue to dull, deep orange. The small series from Isla Catalina resembles the mainland specimens in dorsal pattern and in extent of the gular band, which in parvoris may involve the chest and undersides of the arms. The dorsal dots are conspicuously confluent, the lateral fields are obsolete, and the ventral coloration is pale bluish with an orange tint. The most obvious difference in coloration is that the dorsal surfaces of the hindlimbs are rusty — a feature found in no mainland parvoris. The Catalina series is composed of one adult male with a snout-vent length of 126 mm and thus a large lizard, and three females (one of which is a small juvenile), the largest of which has a snout-vent length of 112 mm, again a large lizard. The transverse ventrals are 10 in three specimens and 12 in one. In all other features of scalation the Catalina lot falls within the known range of mainland parvoris. Additional speci- mens from Isla Catalina ma}- well reveal that it is inhabited by still another distinctive race ; the strikingly rusty hindlimbs are indicative of at least one major color difference between Isla Catalina specimens and mainland parvoris. Comparisons: No other race of A. chrysolaema has the back with irregularly arranged and at times confluent spots, and tigroid sides. This feature alone will distinguish parvoris from the previous races. In size, parvoris is much smaller than chryso- laema and slightly smaller than procax, but larger than the re- maining subspecies. In fourth toe scales, parvoris averages less (83.2) than other races except umhratilis, which is comparable (83.0). Pari^om averages less (38.2) than chrysolaema (43.8) in femoral pores, but exceeds the remaining races, all of which have low femoral pore counts. In fifteenth verticil scales, parvoris has the same mean as hoekeri and umhratilis (42.7), less than chry- solaema and alacris (44.4 and 44.0), and almost the same as procax (42.8). Remarks: A. c. parvoris is not known to intergrade with procax on the west nor with the following subspecies to the east. The easternmost locality for procax is separated by 35 kilometers from the westernmost parvoris record. It is possible that the Rio Ozama may divide these two races of A. chrysolaema. The area occupied by parvoris does not differ in any obvious way from that inhabited by procax. 454 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Ameiva chrysolaema jacta^ new subspecies Holotype: MCZ 75267, an adult male, from Juanillo, La Ro- mana Province, Republica Dominicana, taken 29 March 1963, by Clayton E. Ray and Robert R. Allen. Paratypes: MCZ 75268-69, same data as holotype. Diagnosis: A subspecies of A. chrysolaema characterized by a combination of large size (males to 134 mm snout-vent length; females unknown), usually 12 transverse rows of ventrals, low number of fourth toe subdigital scales, moderate number of femo- ral pores, and very low number of scales in the fifteenth caudal verticil; a dorsal pattern of dark brown to black tigroid mark- ings on a grayish tan background (in preservation), the pattern extending in a diluted fashion onto the neck, a bold, checker- board-patterned tail (Fig. 4, right), and a black gular band which extends onto the chest and underside of the arms. Dist)ihidion: Known only from the type locality in extreme eastern Republica Dominicana (Fig. 11). Description of type: An adult male with the following meas- urements and counts : snout-vent length 134 mm, tail 304 mm ; ventrals in 38 longitudinal and 12 transverse rows; fourth-toe subdigital scales 41 and 41 (total 82) ; femoral pores 21 and 22 (total 43) ; 35 scales in the fifteenth caudal verticil. Dorsum (preserved) tannish gray with a dark brown, almost black pat- tern consisting of bold vertical tigroid markings on the sides and about five wide, dark, longitudinal lines on the back, the latter much confused and joined by the lateral vertical markings, giv- ing a rather complete and complex dark brown reticulum, which, although present on the neck, is much paler gray. Lateral fields completely absent, no dotting on sides or back whatsoever. Upper surface of forelimbs with obscure grayish lines and blotches, hindlimbs with a dark brown reticulum enclosing large spots which are pale centrally and darker peripherally. Tail with a bold checkerboard pattern of grayish tan, white, and dark brown, this pattern becoming obsolescent and absent on the distal half of the tail. Belly and throat presently dull grayish orange, with some lateral ventral scales with very dark gray blotches, thereby giving the belly somewhat of a faint checkerboard appearance laterally. Black gular band obsolete but indicated, and anterior ten rows of ventrals clouded with dark gray. 1 From the L.itin for "thrown," an nllusion to the far flung distribution of this subspecies. SCHWARTZ AND KLINIKOWSKI : AMEIVA 455 Variation: See tables. In coloration and pattern the two paratypes are much like the type and require little comment. The major difference is that the paratypes have a bold, black, gular band which involves the chest and the underside of the arms. Neither lizard has any indication of lateral fields, and the sides and back have the tigroid vertical bars and brown to black reticulum, as well as the checkerboard tail, just as de- scribed for the type. Comparisons: A. c. jacta does not need detailed comparison of pattern with any other described subspecies ; the boldly and viv- idly marked back with its light ground color will distinguish jacta from the remaining races. A. c. parvoris is closest to jacta in pattern, but the differences are so striking that the similarity between these two subspecies is not very great. In size, jacta is smaller than chrysolaema, procax and parvoris, and larger than the remaining forms. All other races have a higher numl^er of fourth-toe scales and fifteenth verticil scales. Jacta has a high mean of femoral pores, having less than chryso- laema, and more than the balance of the subspecies. Remarks: No intergradation is known between jacta and par- voris to the southwest; the easternmost mainland locality for parvoris is separated by 145 kilometers from that of jacta. We have attempted to secure specimens of A. chrysolaema between San Pedro de Macoris and Juanillo at several localities along the coast (La Romana, Boca de Chavon, Boca de Yuma) as well as inland in this eastern region, without success. Typical xeric chry- solaema habitats here are occupied by A. taeniura. Considering the likeness of jacta to the race from Isla Saona, described below, it is probable that this boldly marked type of lizard was at one time (and still is?) abundant locally in extreme eastern Hispan- iola. Presently, the hiatus between jacta and parvoris and the apparent absence of the species in this eastern region suggests strongly that the populations are relict wdth a disjunct distribu- tion. AmEIVA CHRYSOLAEMA RICHARDTHOMASI ^ UCW SUbspecicS Holotypc: :\ICZ 77235, an adult male, from the environs of Mano Juan, Isla Saona, Repiiblica Dominicana, taken 19 July 1964 by Richard Thomas. Original number V3018. 1 Named for the collector. 456 BULLETIN: MUSEUM OP COMPARATIVE ZOOLOGY Paratypes: ASFS V3019-30, USNM 152575-76, AMNH 92856- 58, KU 79875-77, UIMNH 56898-99, RT 935, same data as holo- type. Diagnosis: A subspecies of A. chrysolaema characterized by a combination of large size (males to 137 mm, females to 124 mm snout-vent length), usually 10 (but often 12) transverse rows of ventrals, high number of fourth-toe subdigital scales and femoral pores, and moderate number of scales in the fifteenth caudal ver- ticil; dorsal pattern of two phases: (1) back gray-green with Fig. 5. Left, Amelia c. richardthomasi, holotype, MCZ 77235, environs of Mano Juan, Isla Saona, Eepublica Dominicana. Eight, Ameiva c. ridtard- thomasi, ASFS Y3019, environs of Mano Juan, Isla Saona, Eepublica Dominicana. SCHWARTZ AND KLINIKOWSKI : AMEIVA 457 only an indistinct mottling of gray-brown in the area of the lat- eral fields or (2) back gray-green with heavy black mottling, no lateral fields, and tigroid vertical bars on the sides, the dorsal mottling not extending onto the shoulders and neck (Pig. 5), and without a black gular band. Di.strihufion: Known only from the type locality, but pre- sumed to occur throughout Isla Saona (Fig. 11). Description of type: An adult male with the following meas- urements and counts : snout-vent length 137 mm, tail 146 mm, twice broken and regenerated ; ventrals in 37 longitudinal and 12 transverse rows ; fourth-toe subdigital scales 45 on left side ; femoral pores 22 and 20 (total 42) ; 38 scales in the fifteenth tail verticil. Dorsal ground color gray-green (Maerz and Paul: pi. 22F1), becoming finely mottled laterally with a series of \Qvy faint grayish brown vertical bars in the region of the lateral fields ; lower sides putty colored. Throat and venter orange with no black gular band. Fore- and hindlimbs unicolor with back, and patternless. Tail grayish tan without obvious checkerboard pattern above, putty colored below. Variation: See tables. Sixteen specimens of richardthomasi (including the type) are patternless dorsally and have a finely filigreed, grayish brown area in the region of the lateral fields. The tails are gray to tan without a prominent checkerboard pat- tern as in the patterned lizards. The venters are orange to drab gray, generally slightly more orange in the pectoral region, and grayer posteriorly. Throat and underside of forelimbs are mottled to nearly unicolor orange, sometimes in discrete flecks. The younger specimens have light gray throats. The underside of the hindlimbs and tail is gray to putty colored with some orange on the anterior surface of the femur, and in some speci- mens on the entire underside of the hindlimb. Two small juve- niles (snout-vent length 47 and 51 mm) are in this patternless phase. Eight specimens (adults and subadults of both sexes) have patterned backs. In this phase, the dorsal ground color is gray- green with a heavy black mottling and strikingly' tigroid barred sides. The black dorsal x^attern quickly fades at the shoulders and is absent or very suppressed on the neck. The dorsal blotch- ing is much as that described for jacta, i.e., a vermiculate or at times longitudinally arranged configuration of black on a lighter ground. In no specimens are the lateral fields apparent and there is no spotting or dotting on the sides. The bellies of these 458 BULLETIN: MUSEUM OF COMPARATR^ ZOOLOGY patterned lizards were the same as those of patteruless speci- mens; females of the patterned lizards have the extent and in- tensity of the orange not so great as do the males. The checker- board tail is a common feature. As in A. c. hoekeri, we have no doubt that the two phases in richardthomasi represent two basic patterns, and in no way should be interpreted as an adult phase versus a juvenile and subadult phase. Although the only two juveniles at hand are patternless, there is an intermingling of sizes of both sexes insofar as the two phases are concerned. Just as in hoekeri, which is rep- resented by a much longer series, there are no intermediates be- tween the two conditions; each lizard is distinctly in one phase or the other . Comparisons: The patterned phase of A. c. richardthomasi re- quires comparison only with A. c. jacta to which race the former is obviously closely allied. The two can easily be differentiated in that jacta has a black gular band and the dorsal pattern con- tinues anteriorly onto the neck, whereas richardthomasi lacks a gular band and has the pattern faded anteriorly. Jacta also is not known to have a patternless phase The patternless phase of richardthomasi requires comparison with patternless hoekeri. The two are much alike, but richard- thomasi differs in having the filigreed or mottled lateral field area whereas hoekeri has an obsolete lateral field and no mottling in this region. Also the dorsal hues of hoekeri populations are usually not greenish. In size, richardthomasi is smaller than chrysolaema and procax, equal to parvoris, and larger than the remaining races. In having a mean of 87.6 fourth-toe scales, richardthomasi averages higher than all previously named races. In femoral pore counts, it is higher than all races except chrysolaema. Considering fifteenth caudal verticil scales, richardthomasi is exceeded by all forms ex- cept jacta. Remarks: The close alliance of richardthomasi with jacta is obvious. If we assume that extreme eastern Hispaniola is (was) inhabited by a population with heavy dorsal mottling and marbling, this population must also have given rise to the Saonan subspecies. We have no evidence that jacta or a related form still occurs on the adjacent Hispaniolan mainland. Certainh' richardthomasi is an insular derivative of an extreme eastern heavilv marked form which was likelv similar to jacta. SCHWARTZ AND KLINIKOWSKI : AMEIVA 459 Ameiva chrysolaema leberi^ new subspecies Holohjpe: MCZ 77236, an adult male, from 5 km E Peder- nales, Pedernales Province, Republica Dominicana, one of a series taken 25 June 1964 by David C. Leber and Richard Thomas. Original number V2509. Paratypes: All from the Republica Dominicana, Pedernales Prov., as follows: ASFS V2510-19, USNM 152577-82, RT 932, KU 79878-81, same data as holotype; MCZ 81015-18, UIMNH 56900-05, Pedernales, 3 July 1964, D. C. Leber, R. Thomas; AMNH 92859-61, 1 km E Pedernales, 25 July 1963, R. F. Klini- kowski ; AMNH 92862-63, 12 km E Pedernales, 25 June 1964, R. Thomas. Associated specimens: Haiti, Dept. d l' Quest, Tean, nr. Sal- trou (not mapped), 4 (MCZ 69389-92) ; Saltrou, 7 (AMNH 50000- 04, 50007-08). BepuMica Dominicana, Pedernales Prov., Oviedo, 3 (MCZ 58674-76). Diagnosis: A subspecies of A. chrysolaema characterized by a combination of small size (males to 111 mm, females to 104 mm snout-vent length), 10 transverse rows of ventrals, moderate num- ber of fourth-toe subdigital scales and femoral pores, and low number of scales in the fifteenth caudal verticil ; a completely pat- ternless rusty brown dorsum, no lateral fields (Pig. 6, left), a deep orange-red belly, and a black gular band which may involve the chest and underside of the arms. Distribution : To the south of the Massif de la Selle and Sierra de Baoruco, from the vicinity of Saltrou in Haiti, east onto the Peninsula de Barahona, to 12 kilometers southeast of Pedernales (Fig. 11). The record from Oviedo is discussed below. Description of type: An adult male with the following meas- urements and counts : snout-vent length 103 mm, tail 233 mm ; ventrals in 34 longitudinal and 10 transverse rows; fourth-toe subdigital scales 40 and 42 (total 82) ; femoral pores 21 and 20 (total 41) ; 40 scales in fifteenth caudal verticil. Dorsum uniform rusty brown anteriorly, becoming gray-brown posteriorly; sides of head reddish brown. Lateral fields absent. Throat orange, ven- tral ground color brick red, lower sides (and lateralmost two rows of ventral scales) blue. A black gular Imnd, extending onto the first four or five rows of ventrals and onto the undersides of the arms. Tail gray above, off-white below. Dorsum and top of tail completely unpatterned. 1 Xamed for one of the collectors. 460 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Fig. 6. Left, Ameiva c. leheri, holotype, MCZ 7720(5, '> km E Pederualcs, Pedeniales Prov., Kepubliea Dominicana. Bight, Ameiva c. ficta, holotype, MCZ 77237, 13.1 mi. SW Euriquillo, Pedernales Prov., Eepubliea Dominicana. Variation: See tables. The dorsal ground color varies from rusty brown to reddish brown, and the ventral ground color from gray with small amounts of red to orange-red, fading posteriorly, to brick red. The gular band is present in all specimens, and in only one male does it not extend onto the chest. The lower sides and lateral two rows of ventrals on each side are blue, or at least have blue flecking, but one small male has the lateral ventrals orange-red like the balance of the venter. The lores and cheeks are unspotted pinkish gray. The tails are gray, unmarked above, SCHWARTZ AND KLINIKOWSKI : AMEIVA 461 and the undersides of the tails are a grayish off-white. There is no obvious sexual dichroniatism. Comparisons: By virtue of its patternless dorsum, leheri can easily be differentiated from all races except patternless hoekeri and patternless richardthomasi. The different dorsal hues of leheri and ricJiardfhomasi (rusty brown versus greenish gray) and the presence of a gular band in the former and its absence in the latter, as well as the larger adult size of richardthomasi, all make this distinction easy. From patternless hoekeri, leheri dif- fers in smaller adult size, and in lacking any expression of the lateral fields, whereas boekeri retains the fields as obsolete grayish longitudinal smudges. The vivid venters of leheri are not found in hoekeri, and the brighter dorsal ground color of leheri con- trasts strongly with the more drab tones of hoekeri. Leheri is the smallest race of A. chrysolaema. In number of fourth-toe scales, leheri averages lower (85.5) than richard- thomasi (87.8) and chrysolaema (86.7), and is higher than the remaining subspecies. In femoral pores, leheri again averages less (41.3) than chrysolaema (43.7) and richardthomasi (42.6), the same as jacta, and more than in the other races. In fifteenth verticil scales, leheri is exceeded by all subspecies except jacta. Remarks: A. c. leheri is not known to intergrade with either chrysolaema to the north (from whose range it is completely separated by the Massif de la Selle) or with the form to the east on the Peninsula de Barahona. The three specimens from Oviedo noted in "Associated Specimens" above, will be discussed in detail in the treatment of the following subspecies. AmeIVA CHRYSOLAEMA FICTA^ UCW SUbspCcicS Holotype: MCZ 77237, an adult male, from 13.1 mi. (20.8 km) SW Enriquillo, Pedernales Province, Republica Dominicana, one of a series taken 22 July 1963 by Albert Schwartz and Richard Thomas. Original number X9401. Paratypes: All from the Repiiblica Dominicana, Pedernales Province, as follows: ASFS X9402-09, same data as holotype; ASFS X9950, same locality as holotype, 30 July 1963, R. Thomas; ASFS V197-98, same locality as holotype, 4 August 1963, D. C. Leber, R. Thomas. Associated specimens: Repuhlica Dominicana, Pedernales Prov., 30 km from Oviedo, road to Pedernales, 1 (MCZ 58673) ; 1 Prom the I-atin for "invented, devispfl," in allusion to resemblances to abbntti. 462 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Oviedo, 15 (MCZ 58677-80, 58682-90, 58692-93) ; 5 mi. (8 km) NE Oviedo, 11 (ASFS X9957-58, V273-80, RT 752) ; Barahona Prov., 3 km SW Enriquillo, 1 (ASFS V290) ; Enriquillo, 2 (MCZ 58777-78). Diagnosis: A subspecies of A. chrysolaema characterized by a combination of moderate size (males to 121 mm, females to 113 mm snout-vent length), 10 transverse rows of ventrals, low number of fourth-toe subdigital scales, moderate number of femoral pores and scales in the fifteenth tail verticil; dorsum tan to brown, spotted with pale blue, lateral fields obsolescent and often replaced by vertical blackish bars or vermiculations (Fig. 6, right), belly rust colored, a prominently checkerboarded tail, and a black gular band often extending onto the chest and underside of the arms. Distribution: The Peninsula de Barahona from (apparently), 30 km NW Oviedo in the west, east to the east coast in the vicin- ity of Oviedo, and thence north to Enriquillo (Fig. 11) ; see however Remarks below. Description of type: An adult male with the following meas- urements and counts : snout-vent length 112 mm, tail 287 mm ; ventrals in 36 longitudinal and 10 transverse rows; fourth toe subdigital scales 38 and 39 (total 77) ; femoral pores 21 and 20 (total 41) ; 38 scales in the fifteenth caudal verticil. Dorsal ground color brown, with six longitudinal series of more or less discrete pale blue spots, the lateralmost series bordering above the remnants of the lateral fields, below which is another longi- tudinal series of pale blue spots; lower sides with alternating vertical black and pale blue bars ; f orelimbs vaguely spotted with small dots, hindlimbs boldly marked Avith large rusty spots dorsally. Throat grayish orange, belly rust colored; an exten- sive black gular band which includes the first eight rows of ventrals and extends onto the undersides of the arms. Tail brown above, with blue spots on the first nine verticils dorsally, and additionally somewhat checkerboarded; tail ivory below. Variation: See tables. The tan to brown dorsum with pale blue discrete spots characterizes A. c. ficta. The spots, on occasion, may be greenish anteriorly or tan posteriorly, but in general they are pale blue. The lateral fields are obsolete or almost completely obliterated by vertical black bars alter- nating wdth blue bars on the sides. The throat and ventral ground color are grayish orange and rust, with bellies of females slightly paler than those of males. In some individuals the dorsal spots are distinctly lineate in appearance, and in a few SCHWARTZ AND KLINIKOWSKI : AMEIVA 463 the back has a more or less complete finely filigreed appear- ance, although this is not the norm. The spots themselves vary in size, distribution, and density ; they may be much smaller than in the type and much more closely appressed to one an- other, or the paramedian rows (if rows are discernible) may be fused to form a pair of paramedian pale blue lines. The checkerboard tail with blue spotting on its basal portion is a common feature. Comparisons: Only one other race thus far discussed, A. c. parvoris, has a spotted dorsum, although A. c. chrysolaema has a pattern of spots and lines. In neither of these two races are the dorsal spots pale blue, but are rather some shade of yellow. In actuality, the pattern of chrysolaema, although dotted, bears little resemblance to that of ficta; the pattern of parvoris is similar but the coloration and general aspect of the lizards of these two races are quite distinctive. Parvoris lacks a conspicu- ously checkerboarded tail. Compared to the described races, jicta is exceeded in size by all other forms except leheri, which is still smaller. In number of fourth-toe scales, ficta is exceeded by all races except jacta. Chrysolaema, richardthomasi, jacta, and leheri exceed ficta in mean number of femoral pores, and in this character ficta exceeds the balance of the races. Ficta averages higher in fifteenth verticil scales than richarclthomasi, leheri, and jacta, and lower than the other races. Remarks: The distribution of A. c. ficta encompasses the east- ern shore of the Peninsula de Barahona from Enriquillo south to Oviedo, and thence inland toward Pedernales for a distance of 30 kilometers. Ficta is not known to intergrade with either imibratilis to the north or leheri to the west (but see below). The northernmost station for ficta is 40 kilometers from the closest record of umhratilis ; we presume that these two races do not come in contact because of the, at best, narrow and intermittent nature of suitable habitats for chrysolaema along the east coast of the Peninsula de Barahona. There are three specimens (MCZ 58674-76) from Oviedo which are clearly leberi and in no way resemble ficta. Assuming that these specimens did indeed come from Oviedo, they present a problem. They are the only specimens from the entire eastern coast of the Peninsula de Barahona which are patternless ; in our considerable collecting experience in the Oviedo region, we never encountered nor collected any leheri-\ike individuals. There are 464 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY several possibilities ; all of which depend upon the assumed cor- rectness of the locality data for these three specimens: 1) they represent a Ze&en-phase of ficta; 2) leheri and ficta are not both subspecies of chrysolaema; 3) leheri and ficta are both chrysolaema derivatives but one has been so long separated from the parent stock that the two forms act as species, with a rather broad (30 km) region of sympatry. In defense of our arrangement of these two forms as subspecies of A. chrysolaema, the following comments are pertinent: 1) There is no incontrovertible evidence that these three lizards are a patternless phase of ficta. All of our own Oviedo material is pat- terned, and there is no indication that there exists a patternless phase oi ficta (although of course there is this possibility). 2) Since neither Ichcri nor ficta intergrades with any other subspecies for reasons of geography, and since both have apparently been long isolated from chrysolaema and umhraiilis to the north, one or both might be logically regarded as a distinct species (if so, then ahbotti and ficta would compose one species, or leheri could be so regarded). We feel that leheri, despite its complete isolation from chrysolaema, is so like patternless boekeri and richard- thomasi that to regard it as a distinct species would be mislead- ing and obscure its clear relationships to the balance of A. chrysolaema. A somewhat stronger case may be made for sep- arating ahhotti and ficta at the species level; here again, how- ever, the resemblance of both these spotted forms to parvoris for example (as well as the overall similarities of ahhotti- ficta to the more northern subspecies) tends in our opinion to negate removing these two forms from the species chrysolaema. 3) The most appealing interpretation is that one (leheri) of the two in- volved forms has been long separated from its parent stock (A. c. chrysolaema), and that once contact between it and another subspecies (ficta) has been re-established, the two forms do not intergrade but act as separate species. The present lack of contact between leheri and chrysolaema and between ficta and umhratilis suggests that both forms may well have had long inde- pendent histories. It is even not improbable that ficta has been derived from aMotti, rather than the reverse, and thus ahhotti may have been insularly isolated from leheri. Such a combination of situations might argue for species status for both leheri and ahhotti-ficta and we have considered this possibility. On the other hand, such a decision obscures the obvious relationships of these two forms to A. chrysolaema (in contrast, for instance, to A. taeniura or A. lineolata) . SCHWARTZ AND KLINIKOWSKI : AMEIVA 465 Finally, and probably the most important point is that the region between Oviedo and Pedernales still remains little known herpetologically ; there is alwaj^s the possibility that the pre- sumed leheri from Oviedo are in actuality from farther west and thus from within the known range 0/ leheri. We have tried to adhere in this ease to a via media, and rather than make assump- tions from inadequate data, we consider both leheri and ficta sub- species of A. chrysolaema, although admitting that the situa- tion is not completely clear. Exclusive of these three question- able lizards, the ranges of jicta and leheri approach, very closely; the distance between the nearest localities for the two races is only 15 kilometers. AmEIVA CHRYSOLAEMA ABBOTTI Noble, 1923 Ameiva ahhotti Noble, 1923, Amer. Mus. Novitates, 64:1 (type locality — Isla Beata, Eepublica Dominicana). Diagnosis: A subspecies of A. chrysolaema characterized by a combination of small size (males to 117 mm, females to 108 mm snout-vent length), usually 12 transverse rows of ventrals, high number of fourth-toe subdigital scales and femoral pores, and moderate number of scales in the fifteenth verticil; dorsum black with a pattern of isolated spots which are orange or yellowish, becoming blue anteriorly, lateral field absent, the sides spotted with sky-blue spots (Fig. 7, left), venter deep brick red to orange posteriorly, and a black gular band which expands to cover the chest and underside of the arms. Distrihution: Known only from Isla Beata, off the tip of Cabo Beata (Fig. 11). Disciissio7i: A. c. ahhotti is the most brilliantly colored and striking of the races of A. chrysolaema. The combination of black dorsal coloration, orange to yellowish spots middorsally, becom- ing blue anteriorly, and the vivid blue lateral spots provides a particularly colorful lizard. The forelimbs are black to brown (distally) with blue spots, and the hindlimbs black with prox- imally blue and distally orange spots. The venter is brick red, grading to orange or pinkish posteriorly, and the lateralmost ventrals are invaded by blue and white spotting. The heads are tan to orange with white or bluish spots on the sides. There is a bold black pectoral band which extends onto the chest and even onto the venter and the undersides of the arms. The underside of the hindlimbs is orange on the thighs and orange 466 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Fig. 7. Left, Ameiva c. abbotti, ASFS V2743, Isla Beata, Eepublica Dominicana. Eight, Ameiva c. secessa, holotype, MCZ 77238, Etroits, He de la Gonave, Haiti. to light tan on the crura. The upperside of the tail is checker- boarded black and white or cream; the underside of the tail is gray to greenish on its proximal half to two-thirds, and uniform light tan to pinkish distally. The dorsal spots are invariably discrete and there is neither a tendency for them to become lineate or to be fused into longitudinal lines. The females are colored and patterned like the males; a juvenile lacks the bright orange ventral color and the black pectoral area. A. c. ahhoiti is so very distinctive in its dorsal coloration and pattern that it is hardly necessary to compare it with any other SCHWARTZ AND KLINIKOWSKI : AMEIVA 467 subspecies. Closest, at least in pattern, are parvoris and ficta; the former, although a dorsally spotted race, is not so gaudy and lacks the discrete spotting so characteristic of abhotti. The ad- jacent race ficta, on the mainland, resembles ahhotti in basic pattern, but differs in having the back brown rather than black, in having the dorsal spots pale blue rather than orange to yellowish, in having the spots at times arranged into lines, and in having the spotting on the back quite variable in density. In contrast, ahhotti is surprisingly constant in density of dorsal spotting. Ficta is primarily a race with 10 rows of ventrals. whereas ahhotti usually has 12. Variation: See tables. A. c. ahhotti has the highest mean (89.6) of fourth-toe scales of any race described to this point, and is ap- proached only by richardthomasi (87.8) ; of all races of A. chry- solaema, ahhotti has the highest average of femoral pores (43.8), although it is closely approached by chrysolaema (43.7). AVith a mean of 40.4 fifteenth verticil scales, ahhotti exceeds ficta, richard- thomasi, leheri and jacta, and has less verticil scales than the other races. The relationships of abhotti are obviously with ficta on the mainland. Whether the latter occurs on the southern tip of the Peninsula de Barahona is unknown, but it certainly is a more likely candidate there than the drab and patternless leheri, if we assume that abhotti was derived directly from the adjacent mainland. Specimens examined: Repuhlica Dominicana, Isla Beata, 39 (ASFS V2743-69, MCZ 28571-73, 37578-79, 37581-82, 17676-77, 57049, RT 934, UMMZ 83098). Ameiva chrysolaema secessa^ new subspecies Holotype: MCZ 77238, an adult male, from Etroits, He de la Gonave, Haiti, taken 17 July 1962 by Elie Cyphale. Original number X2447. Paratypes: All from He de la Gonave, as follows: ASFS X2440-46, X2448-59, UIMNH 56906-09, USNM 152583-87, AMNH 92864-69, KU 79882-86, same data as holotype ; USNM 80377-78, Pointe Quest, 21 March 1930, L. H. Parish and W. Perrygo; USNM 77062-69, MCZ 25539-48, Pointe a Raquette, August 1927, W. J. Eyerdam; MCZ 80251-78, Pointe a Raquette, sum- mer 1964, G. Whiteman; MCZ 80231-36, Nan Palmiste, 4 km 1 From the Latin for "distant, removed." 468 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY from Pointe a Kaquette, summer 1964, G. Whiteman; MCZ 80237-50, Ti Roche, 0.5 km from Pointe a Raquette, summer 1964, G. Whiteman; USNM 80359-68, 80370-76, UMMZ, 92196, Anse a Galets, 23 March 1930, L. H. Parish and W. Perrygo; MCZ 37568-77, Anse a Galets, 9 April 1934, T. Barbour; USNM 76803, Nan Cafe, March 1929, A. J. Poole and W. Perrygo. Associated specimens: He de la Gondve (no other locality), 3 (CM 8133, MCZ 12870-71). Diagnosis: A subspecies of A. chrysolaema characterized by a combination of large size (males to 134 mm, females to 111 mm snout-vent length), usually 12 transverse rovs^s of ventrals, high number of fourth-toe subdigital scales, and moderate number of femoral pores and scales in the fifteenth caudal verticil ; dor- sum reddish brown with six or seven dull buffy longitudinal lines, grayish brown (rather than black) lateral fields with isolated buffy dots enclosed therein (Fig. 7, right), and with the black gular band usually absent, or at least very restricted, seldom involving the anterior ventrals but at times extending onto the underside of the arms. Distribution: He de la Gonave, Haiti (Fig. 11). Description of type: An adult male with the following meas- urements and counts: snout-vent length, 117 mm, tail 134 mm, broken; ventrals in 39 longitudinal and 10 transverse rows; fourth-toe subdigital scales 45 and 45 (total 90) ; femoral pores 21 and 21 (total 42) ; 42 scales in the fifteenth caudal verticil. Dorsal ground color reddish brown with seven dull buffy longi- tudinal lines, the median line somewhat broken and indistinct; head dull tan, neck greenish ; sides of head gray with whitish blotches. Lateral fields grayish brown with an enclosed series of buffy dots, more distinct posteriorly than anteriorly, the lateral fields set off above by the lateralmost dorsal lines, and below by a series of bluish spots; lower sides spotted with blue. Throat dirtj^ pinkish gray, gular band absent, belly gray. Fore- limbs with scattered pale greenish blue spots, hindlimbs pro- fusely dotted with pale yellow. Tail reddish brown above, grayish blue below, with some blue scales on the sides. Variation: See tables. The large series of A. c. secessa at hand shows little variation in pattern ; the entire animal invariably is quite dull, and the lateral fields are never conspicuous. The longi- tudinal lines do not contrast especially strongly with the dorsal ground color and in some topotypes are very obscure and are seen with some difficulty. In many specimens there are six SCHWARTZ AND KLINIKOWSKI : AMEIVA 469 (rather than seven) dorsal lines, the median line being absent. In- variably the gular band is poorly developed or completely absent ; if the band is present, it does not involve the anterior ventrals but may send some pigment onto the underside of the arms. The coloration of the venter varies from gray and bluish gray to dull orange-gray. There is no sexual dichromatism. Comparisons: A. c. secessa is so dull and drab compared to all other races that no comparison is really necessary. It differs from the spotted races parvoris, ficta, and ahhotti in being lon- gitudinally lined, and from the patternless races hoekeri, richard- thomasi and leheri in having a pattern. It is much duller pat- terned, and likewise differently patterned, from the other lined races — clirysolaema, mnhratilis, alacris, and patterned hoekeri. From jacta and richardthomasi (in the patterned phase), secessa differs in lacking the lateral tigroid markings and in having a quite different dorsal pattern. Perhaps the most cogent compari- son is with clirysolaema which occupies all the adjacent mainland about the Golfe de la Gonave. From clirysolaema, secessa can at once be differentiated by its much more drab coloration and pattern, and by the lack of dotting in combination with lines on the dorsum. Both chrysolacma and secessa are typically 12-row lizards. In fourth toe scales, secessa has a higher mean than any other race, being approached most closely by ahhotti (89.6). In num- ber of femoral pores, secessa is exceeded only by ahhotti, chryso- laema and richardthomasi; in fifteenth verticil scales, secessa, exceeds ficta, richardthomasi, leheri and jacta, and is exceeded by the means of the balance of the subspecies. Remarks: Although A. c. secessa, has presumably evolved from the adjacent A. c. clirysolaema, in dorsal pattern it grossly resembles alacris and procax, but is quite distinct in several features, notably the obscure lateral fields and the lack of a gular band. It seems likely that the nominate race has carried the pattern evolution — i.e., disintegration of the longi- tudinal lines into a series of dots — farther than has the isolated secessa which has become faded and pale in contrast to its more brightly colored neighbor. Another possible origin of secessa is discussed below. 470 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Ameiva chrysolaema DEFENSOR^ iiew subspecies Holotype: MCZ 63379, an adult male, from Mole St. Nicholas, Dept. du Nord Quest, Haiti, one of a series taken 24-29 July 1960 by A. S. Kand and J. D. Lazell, Jr. Paratypes: All from Haiti, Dept. du Nord Quest, as follows: MCZ 63368-72, 63374-78, same data as holotype; MCZ 63364- 67, Jean Rabel, 26 July 1960, A. S. Rand and J. D. Lazell, Jr. ; AMNH 49856-57, Port a I'Ecu, 1 April 1935, W. G. Hassler; USNM 59925, Bale des Moustiques, 3 May 1917, W. L. Abbott ; AMNH 49851-55, river just W Port-de-Paix, 2 April 1935, W. G. Hassler ; MCZ 58014, river just W Port-de-Paix, 2 August 1935, W. G. Hassler. Associated specimens: Haiti, Dept. du Nord Quest: Bombar- dopolis, 1 (MCZ 63381) ; Dept. de VArtibonite, Gros-Morne, 1 (MCZ 63380). Diagnosis: A subspecies of A. c. chrysolaema characterized by a combination of moderate size (males to 126 mm, females to 106 mm snout-vent length), 10 transverse rows of ventrals, low number of fourth toe subdigital scales and scales in the fifteenth caudal verticil, and moderate number of femoral pores ; dorsal pattern a series of six or seven dull longitudinal lines on a tan to brown background, lateral fields dull brown, not especially con- trasting with the dorsal ground color and often with the included light spots in the lateral field much reduced or completely absent, a checkerboard tail pattern (Fig. 8, left), and no indication of a black gular band, black on the anterior ventrals, or extension of black pigment onto the underside of the arms. Distribution: The northwest peninsula of Haiti, from Bom- bardopolis in the south to the vicinity of Port-de-Paix in the northeast, and thence south to Gros-Morne (Fig. 11). Description of type: An adult male with the following meas- urements and counts: snout-vent length 118 mm, tail 242 mm, partially regenerated; ventrals in 38 longitudinal and 12 trans- verse rows; fourth-toe subdigital scales 44 and 44 (total 88), fem- oral pores 19 and 18 (total 37) ; 38 scales in the fifteenth verticil. Dorsal ground color (in preservative) dull brown with a series of seven tan longitudinal lines, the median line the least conspic- uous, all lines disappearing on the neck. Lateral fields brown, with included tan dots only in their posterior thirds, the anterior 1 For the Latin for "defender" in allusion to the English fort at Mole St. Nicholas which guarded the Windward Passage. SCHWARTZ AND KLINIKOWSKI : AMEIVA 471 Fig. 8. Left, Ameiva c. defensor, holotype, MCZ 63379, Mole St. Nicholas, Dept. du Nord Quest, Haiti. Bight, Ameiva c. woodi, unnumbered specimen from MCZ 37583-92, He de la Tortue, Haiti. two-thirds being without dots. Forelimbs grayish tan, vaguely dotted, hindlimbs brown with large pale spots, leaving almost a reticulum of dark brown surrounding the large pale areas. Ven- tral ground color (including throat) bluish gray, no black pig- ment on throat, chest, or undersides of arms. Lower sides with gray and blue markings which are almost tigroid. Tail tan, heav- ily checkerboarded with dark brown above, blue-gray marked with cream below, and with black and some cream on sides. 472 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Variation: See tables. We are somewhat handicapped in discussing A. c. defensor since we have not seen this subspecies in life. Judging from the material at hand, especially the fresh specimens from the Museum of Comparative Zoology (old material in the American Museum from Port-de-Paix and Port a I'Ecu is so discolored that it is completely worth- less insofar as coloration is concerned but does still retain some evidences of pattern), defensor is typically a dull and drab lizard with dorsal coloration of tan to brown with six or seven longitudinal buffy lines. The lateral fields are brown and have the included dots much reduced (often absent an- teriorly) or completely absent. In the latter case the lateral field presents an unbroken brow^n lateral band. In some speci- mens, the longitudinal lines have become more obscure than in the type, due to light pigmented areas in the interline regions, and in one extreme case (IMCZ 63378 — snout-vent 119, and thus not the largest male) the entire back is marbled with dark and light and the sides are tigroid, the latter a condition noted to a slighter degree in some other individuals. No specimen has any indication of black on the chest or undersides of the arms, and the gular black band is lacking completely. The prom- inently checkerboarded tail is a constant feature and is plainly discernible even in old and discolored individuals. Comparisons: A. c. defensor most closely resembles A. c. seccssa from He de la Gonave. However, the reduced or absent dotting in the lateral fields, smaller size, and the generally somewhat brighter dorsal pattern (although defensor is nonetheless a rather drab lizard) will distinguish the two races. Secessa usu- ally has 12 rows of ventrals, defensor usually has 10. From the patternless races, def elisor can be distinguished in having a pattern, and from the spotted subspecies by having a dorsal pat- tern of longitudinal lines. From the other lined races, def elisor differs in the lack of a black lateral field with included yellow dots, and lack of a black gular band. In fourth-toe scales, defensor (82.4) exceeds only ficta (81.9) and jacia (79.7) ; in number of femoral pores, defensor (37.2) exceeds only loekeri (36.5), procax (36.1), umlratiUs (35.6), and alacris (33.8). In fifteenth verticil scales, defensor (38.0) ex- ceeds only leheri (37.9) and jacta (35.7). In counts of fourth- toe scales and fifteenth verticil scales, defensor is quite low in the series of subspecies. SCHWARTZ AND KLINIKOWSKI : AMEIVA 473 Remarks: A. c. defensor is not known to intergrade with any other subspecies ; there are two wide hiatuses, however. The closest approximation of records for defensor (Gros-Morne) and chrysolacma (St. Marc) is 83 kilometers. To the east, there are no specimens available between Port-de-Paix and Cap-Ha'itien, a distance of 70 kilometers. Although A. c. secessa is closer geographically to A. c. chryso- laema, the former is much more similar to defensor than to the nominate race. Such a similarity may be merely convergence and may not reflect direct relationships. It is possible, on the other hand, that defensor has reached the northwest peninsula from GonPive ; it seems hardly likely that the reverse is true — i.e., that Gonave has lieen colonized from the north — consider- ing the proximity of Gonave to the adjacent mainland (21 kilom- eters at its closest point) and its distance from the northwest peninsula (72 kilometers at its closest point). Ameiva chrysolaema woodi Cochran, 1923 Ameiva chrysolaemea woodi Cochran, 1923, Occ. Papers Boston Soc. Nat. Hist., 8:181 (type locality — He Tortue, Haiti). Diagnosis: A subspecies of A. chrysolaema characterized by a combination of large size (males to 141 mm, females to 126 mm snout-vent length), 10 transverse rows of ventrals, low number of fourth toe subdigital scales, moderate number of femoral pores and scales in the fifteenth verticil ; dorsum very dark brown with three to five dull yellowish to buffy stripes or a median dorsal buffy longitudinal band ; the stripes may be variously joined and modified to give rather complex dorsal figures which are derived from the simple five lines (Figs. 8, right; 9) ; sides with vertical tigroid markings, and no black gular band or extensions thereof onto the chest and underside of the arms. Distribution: He de la Tortue, Haiti (Fig. 11). Discmsion: The most strikingly patterned and at the same time most variable of the races is A. c. ivoodi. Basically the dorsal pattern is a series of three to five broad, dull yellowish to buffy longitudinal lines on a very dark brown ground. This basic pattern may be modified in that the area between the two paramedian lines may be filled in with a buffy color so that the back has a lateral pair of pale lines and a middorsal pale zone. From this condition, the balance of the back may be filled in with paler, so that the entire back is marked with a single broad pale middorsal zone. In two individuals, the pale lines have 474 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Fig. 9. Left, Ameiva c. woodi, ASFS X2270, Palmiste, He de la Tortue, Haiti. Eight, Ameiva e. woodi, ASFS X2267, Palmiste, He de la Tortue, Haiti. grossly fragmented and joined randomly, to form a bizarre, longitudinally reticulate pattern -which is quite distinctive. The heads are dull gray-green, with the lores gray. The sides are tan to grayish brown ; this lateral color invades the dark brown back to give a series of alternating dark brown and tan vertical tigroid bars, the tan bars often faintly brick colored at their dorsal points. The throats are dirty pale orange, those of the females usually lighter than those of the males (although one female has a brighter orange throat than any other specimen SCHWARTZ AND KLINIKOWSKI : AMEIVA 475 examined in life). The venter is gray, occasionally with a pale orange wash. No specimen has any indication of a black gular band or anj- black on the chest and undersides of the arms, al- though there may be some isolated black flecking on the chest. The tails are tan dorsall3^ wdth prominent black markings, giv- ing a strong checkerboard effect. The undersides of the tails are gray and often have almost as prominent checkerboarding as the upper surfaces. Variation: See tables. In pattern, no other race of A. chryso- laema is comparable to tvoodi; the three to five longitudinal lines are fewer than in the pattern regularly noted in other subspecies, and the peculiar middorsal broad band, either with or without two dorsolateral lines, and the irregular fragmentation and join- ing of the lines on a dark brown ground are all features which woodi shares with no other race. In lacking a gular band, woodi is comparable only to the adjacent defensor, and seccssa and richardtJiomasi. All other forms have the band present. A. c. ivoodi is a remarkably distinct form; it resembles none of the mainland races and is particularly unlike the adjacent mainland defensor. A. c. ivoodi is a large subspecies, being exceeded only by chrysolaema in size, although procax is equal in snout-vent length. Woodi, in having a low mean (80.5) of fourth-toe scales, exceeds only jaeta (79.7) in this count. In femoral pore count, woodi is exceeded by abhotti, chrysolaema, ricliardthomasi , and secessa, and has a mean femoral pore count equal to those of jacta and leheri. The moderate fifteenth verticil coimt (38.8) of woodi exceeds only richardthomasi, defensor, leheri, and jacta, and is equal to that of ficta. The derivation of woodi must certainly be from the adjacent defensor of the mainland. The lack of a gular band, and black on the chest and undersides of the arms, and the ten rows of ventrals are features in common between the two races. There the resemblances cease, however, since woodi is a boldly and color- fully patterned lizard, whereas defensor is dull and drab. De- fensor likewise does not exhibit any patterns which are reminis- cent of those of vwodi, although the lack of dots in the lateral fields may foreshadow the absence of these fields entirely in woodi. Specimens examined: Haiti, He de la Tortiie, Palmiste. 10 (ASFS X2267-76) ; no specific locality on the island, 29 (MCZ 37583-92 -f 19 unnumbered specimens). 476 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Ameiva chrysolaema regularis Fischer, 1888 Ameiva regularis Fischer, 1888, Jahrb. Wiss. Anst. Hamburg, 5:26 (type locality, Sans Souci, Haiti; this locality is the palace of the same name near Milot, Dept. du Nord, Haiti). Diagnosis: A subspecies of A. chrysolaema characterized by a combination of large size (males to 132 mm, females to 128 mm snout-vent length), usually 12 transverse rows of ventrals, very low number of fourth-toe subcligital scales, moderate number of femoral pores, and high number of scales in the fifteenth verticil ; dorsal pattern a series of five to seven pale yellow lines on a tan to browm ground color, occasionally with a clear tan middorsal zone, neck greenish and dorsal ground color often suffused with Fig. 10. Ameiva c. regularis. ASFS Y-1215, 9 km NW Villa Vasquez, Monte Cristi Prov., Republiea Dominicana. SCHWARTZ AND KLINIKOWSKI : AMEIVA 477 blackish, lateral fields black with an included row of yellow dots (Fig. 10), and black gular band present or absent, when present seldom invading the chest or extending onto the under- side of the arms. Distribution: North central Hispaniola, from Cap-Haitien and Grande Riviere du Nord in the west, east to Fort Liberte, Haiti, and thence to Monte Cristi and throughout the Valle de Cibao as far east as the vicinity of Santiago, Republica Domini- cana; also the Siete Hermanos islands (known from Isla Muer- tos, Toruru, Monte Chico, and Tercero) and Isla Cabras to the north of Monte Cristi (Fig. 11). Discussio7i: Cochran (1941 : pi. 8, figs. B and D) has illustrated two phases in the dorsal pattern of A. c. regularis from Cap- Ha'itien, which is near the type locality of the subspecies. Gen- erally, throughout its wide range on the mainland, regularis is fairly consistent in dorsal pattern. Specimens from the Valle de Cibao, Monte Cristi, and Pepillo Salcedo, which we have seen in life, were brown dorsally, often suffused with blackish, with a greenish wash on the neck, and had five to seven longi- tudinal pale yellow lines. At times there is a clear tan mid- dorsal zone resulting from fusion of stripes and filling in of the interspaces with tan. The lateral field is black and prom- inent with a longitudinal series of yellow dots. The lower sides are dotted with blue-green, and the sides of the belly are bright blue. The ventral ground color varies from grayish to deep dull orange, and the throat from yellowish to gray-orange. The fore- limbs are spotted with blue-green, the hindlimbs with yellow. The gular band may be either present or absent ; if present, it is not extensive and seldom encroaches upon the chest or extends onto the undersides of the arms. There are variants of the above basic pattern, such as that shown by Cochran (1941: pi. 8, fig. D), in which there are faint filigreed lines in a middorsal zone. Some specimens show acces- sory dots between the longitudinal lines, especially posteriorly; in the series from near Villa Vasquez, Monte Cristi Province, two in life clearly showed a secondary dorsal condition, similar to that typical of adult A. c. cJirysolaema, in which the dorsal lines are supplanted by a series of bright yellow dots overlying the fainter longitudinal lines. In general, specimens from near Cap-Haitien seem somewhat darker than those from the xeric Valle de Cibao, but in features of pattern they are not remark- ably different from those from the Valle de Cibao. Occasional 478 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY specimens from the western portion of the range of regularis have the yellow dots in the lateral fields very tiny, and the field thus appears, grossly, to be immaculate, as is characteristic of some defensor. A series of three adult male lizards from Isla Cabras, off the coast just north of Monte Cristi differs from mainland ma- terial in having dark brown lateral fields, and yellow dots on the lower sides. These three specimens also have extensive black gular bands involving the chest and the undersides of the arms. Neither in size nor scalation do there seem to be any differences betw^een these lizards and those from the adjacent mainland. We consider them as regularis since in most features of pattern and coloration they are very close to that race. There are 17 specimens from the Siete Hermanos, a group of seven islets off the mouth of the Kio Yaque del Norte. Of these lizards, ten are from Isla Muertos, two from Isla Tercero, two from Isla Toruru, and three from Isla Monte Chico. In colora- tion, pattern, and scalation they do not differ from mainland specimens, and we regard them as regularis. Variation : See tables. A. c. regularis may be differentiated by its lined pattern from the patternless races — hoekeri, leheri, richardtJiomasi — and those which have distinct patterns (dots, vermiculations, etc.) — parvoris, jacta, ricJiarcWwmasi, fief a, ab- hotti. From the lined races, regularis differs in lacking a dotted and lined dorsum in combination (chrysolaema), in having a dark ground color dorsally and complete and black lateral fields (defensor, secessa), in almost always lacking a pattern of a mid- dorsal tan zone and at times having a black gular band (woodi), and in having twelve rather than ten rows of ventrals (hoekeri, alacris, procax, umhratilis). Begularis most closely resembles alacris and procax; compared with alacris, regularis reaches a larger size and averages fewer fourth-toe subdigital scales (77.8 versus 84.8). From procax, regularis differs in smaller adult size and in often lacking a black gular band, which procax regularly possesses. The two subspecies differ in fourth-toe scales, with regidaris having a lower mean (77.8) than procax (84.8). The north central subspecies is not known to intergrade with either of its neighbors ; there are no specimens to the west between Cap Ha'itien (regularis) and Gros-Morne (defensor) , a distance of 70 kilometers. The closest approximation of regularis (Fort Liberte) and alacris (Cerca-la-Source) is 58 kilometers. The intervening mountains almost certainly completely separate these two subspecies. SCHWARTZ AND KLINIKOWSKI : AMEIVA 479 The distribution of A. c. rcgularis along the northeastern coast of Haiti and thence into the Valle de Cibao in the Republica Dominicana suggests that this form has evolved along the north- ern coast and thence has penetrated into the xeric cul-de-sac of the Valle de Cibao. The affinities of rcgularis with alacris sug- gest strongly that the parent stock has been the latter, yet we cannot visualize any means of dispersal of alacris to the north Haitian coast ; presently at least the Massif du Nord and the Cordillera Central form insurmountable barriers. One route of dispersal suggests itself : the valleys of the Grande Riviere du Nord and the Riviere Bouyaha (the latter a member of the Artibonite system whose upper valleys are occupied by alacris) approach each other in the Departement du Nord. These valleys and their approximation may have offered a means of ingress for Ameiva from the south into the northern Haitian littoral. Specimens examined: Haiti, Dipt, du Nord, Cap-Haitien, 14 (USNM 74075-86, MCZ 37593-94) ; Ti Guinin nr. Cap-Haitien, (not mapped), 24 (UMMZ 122819 [=12 specimens], MCZ 66527- 38) ; Grande Riviere du Nord, 46 (UMMZ 122820 [=12 speci- mens], MCZ 63353-63, 66514-26); Fort Liberte, 6 (USNM 76770-75) ; RepuMica Dominicana, Monte Cristi Prov., Laguna de Salodillo, 7 km S Pepillo Salcedo, 1 (ASFS V1430) ; 4 km E Pepillo Salcedo, 8 (ASFS V1149-55, V1166) ; Isla Cabras, 3 (ASFS V1372-74) ; Monte Cristi, 1 (MCZ 58018) ; 2 km SE Monte Cristi, 5 (ASFS V1210-12, V1284-86) ; 9 km NW Villa Vasquez, 14 (ASFS V1214-25, RT 811-12) ; 5 km W Guayubm, 15 (ASFS V1494-508) ; 7 km N Guayubin, 13 (ASFS V1471-83) ; Valverde Prov., 9 km N Los Quemados, 1 (ASFS V1766) ; 7 km E Valverde, 10 (ASFS V2931-40) ; 2 km E Esperanza, 5 (ASFS V1755-59) ; Santiago Prov., 7 km W Santiago, 2 (ASFS V2925- 26) ; Santiago and vicinity, 6 (MCZ 58665-66, 58668-71) ; Siete Hermanos, Isla Muertos, 10 (ASFS V1590-95, RT 826, USNM 76733-35) ; Isla Monte Chico, 3 (USNM 76715-17) ; Isla Ter- cero, 2 (USNM 76736-37) ; Isla Toruru, 2 (ASFS V1573-74). DISCUSSION Before proceeding to a discussion of the variation and possible history of Ameiva chrysolaema in Hispaniola, we would like to bring out several facts which seem especially worthy of mention. The distribution of patternless races, or at least races which have some patternless members (leheri, hoekeri, richardthomasi), 480 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY rH .2 00 03 a ^ •5 03 O ^^^ o • rH •2 03 a I— 1 u 00 ^^ ■4-3 ^ fl i-i ^ ^ c5 O »,, ■ rH 1— 1 • O c3 o o Ol o I— 1 O 1-1 o 03 ^ -»•> (t o o 03 c3 CO 03 n be g F3 03 f^ O J^ ZJ !=; o -^ ■TS 03 'S '^ OJ C3 rt q-( a> rt -^ 1— 1 • rH '33 o -4-^ Cv r^ 03 CO o •^ o 1—" 0) P I— 1 ft C3 ;z; j- o ty • vv r& o s: « 2 03 s in la > o ■5 O ^ cS ;S o • rH o ■— ' •^ Ci <*H r^ O) 6 •c* ,_, 03 ft 5> 03 03 o QQ S rS o "5 rO £ ^ Sh o 03 p a O] -2 03 o 03 £ 3 o r— I Si ci s "^ ^ J5 ;:^ lO •S 03 o ^ •r^ ft ft 1— 1 •S 1— 1 t-H ,,^ =c "-D u «l-l O «" lO ;^ 03 o 1-1 1— 1 g .« ^ o p ci ^^ « ^ if ~ -i 03 is ^ •«^ "x S Ph 1"' .£' o rr. o c3 a C 03 '^ ra ^ « ?5j SCHWARTZ AND KLINIKOWSKI : AMEIVA 481 is especially interesting. Of these subspecies, one (leheri) is iso- lated on the south shore of His^Daniola below the La Selle-Baoruco massifs; leheri approaches, insofar as known without intergrada- tion, the very different patterned race ficta on the Peninsula de Barahona. Another (hoekeri) has patterned individuals, and is surrounded by three patterned races, of which it intergrades with one {alacris), is separated from another (umhratilis) by the Rio Yaque del Sur, and from the third {procax) by a distinct and dramatic change of environment. Finally, richard- fhomasi on Saona has patterned and patternless individuals; this subspecies is related most closely to jacta. Three races are dorsally spotted : ahhotti and ficta on Beata and the Peninsula de Barahona, respectively, and parvoris on the southeastern coast of Hispaniola. Parvoris is separated from its neighbor to the west, procax, by the Rio Ozama, and from its eastern neighbor, jacta, by a wide gap which is apparently })resently uninhabited by A. chrjjsolaema. The occurrence of parvoris on Isla Catalina, to the east of the known mainland distribution of that race, is noteworthy. Of the striped races, chrysolaema stands alone in its large size and its style of dorsal patterning, a combination of lines and dots. Chrysolaema is known to intergrade with alacris, another striped race, in the vicinity of Mirebalais, but no intergrades are known between chrysolaema and unihratilis in the Cul de Sac- Valle de Neiba. rmhraiilis resembles the striped phase of hoekeri to some extent ; the two are not known to intergrade. The races procax, ala<;ris, and regularis are all comparably striped ; of them, alacris and procax have 10 rows of ventrals, whereas regularis has 12 rows of ventrals. The range 0/ alacris is sep- arated from that of procax by the intervening and quite different hoekeri. Defensor, by virtue of its pallid coloration and drab pat- tern, stands alone among the mainland races, but it is approached somewhat by the drab secessa from Gonave ; secessa has 12 rows of ventrals. rkfensor 10. The Tortue subspecies woodi is very distinctive, I)ut logically must have been derived from either defensor or regularis, the only two races on the north coast. The Sierra de Xeiba and the Montagues du Trou d'Eau form the northern boundary of the Cul de Sac-Valle de Neiba plain, which is in actuality a fossil strait that once separated His- paniola into two distinct islands, the north and south islands The south side of the same plain is bounded by the Massif de la Selle and its associated northern ranges (Morne I'Hopital, ]\Iont 482 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY des Enfants Perdus) and the Sierra de Baoruco. Considering only this region, one is struck by the amazing diversity of the four races of A. chrysolaema associated with it : chrysolaema in the northwest, umhratilis in the northeast, ficta in the southeast and leberi in the southwest. Here we are involved with four races whose patterns are radically different — chrysolaema dark with longitudinal lines and dots, umbratilis pale with lines, ficta with large dorsal spots, and leheri without pattern. It has become customary to catalogue, if possible, Hispaniolan reptiles and amphibians into either north or south island species — i.e., depending upon their present and presumed past distribu- tion. Although this is somewhat difficult in the present case, we feel that A. chrysolaema is a north island species. Except for the isolated occurrence of A. c. chrysolaema at Aquin on the Tiburon Peninsula (based on a single specimen), this south- western extremity of Hispaniola lacks the species; the otherwise westernmost record is from Pere, near Leogane. The occurrence of the very different race ficta on the Peninsula de Barahona (and its relative ahhotti on Beata), as well as leheri to the west along the south coast, indicates that the A. chrysolaema stock was long isolated on the southeastern portion of the south island, where ficta evolved in isolation from the northern mass of the species. The presence of leheri along this south shore is most puzzling, especially since it is geographically closest to ficta (see Remarks under A. c. ficta for additional comments), and since it also resembles hoekeri far to the northeast. Perhaps leheri is the more ancient of the two mainland south island races, and its present rather restricted distribution a mere remnant of a range which was once more extensive, especially to the west toward Jacmel. If such is the case, leheri might be regarded as a sub- species derived from (pre) chrysolaema; a possible source of colonization from the northern shore of the Tiburon Peninsula might be the Vallee de Trouin, the low pass between the north and south shores of the peninsula. It is of course quite possible that additional collecting in the vicinity of Jacmel or between that city and Saltrou will reveal the presence of leheri; it is also possible that chrysolaema or chrysolaema X leheri inter- grades may l)e found in the Vallee de Trouin as well. On the north island, and including the Cul de Sac-Valle de Neiba, we visualize the old coast inhabited by two races, chryso- laema and umhrotilis, much as today procax and parvoris occur along the south shore of the eastern Republica Dominicana. With SCHWARTZ AND KLINIKOWSKI : AMEIVA -483 the closure of the strait, each of these races has expanded into the resulting xeric plain, although to the northwest along the Golfe de la Gonave, chrysolaema still occupies the narrow coastal plain and adjacent xeric foothills much as it may formerly have done farther south. Once across the plain, chrysolaema has ex- tended its range inland to some extent (Petionville), and to the west (Pere, Momance). Vmhratilis, on the other hand, has not been able to penetrate far into the adjacent mountains, although it does reach an elevation of 1000 feet near El Naranjo. The range of nnihratilis is bounded on the north by the valley of the Rio Yaque del Sur. To the east of the Rio Yaque del Sur are a series of four, more ©r less coastal, races. Of these, hoekeri, immediately to the north and east of the Yaque, is rather like umbratilis in its patterned phase. We consider hockcri as a direct derivative of umhratilis and restricted to the Llanos de Azua. The next three races — procax, parvoris, jacta — show increasingly scattered patterns of distribution to the east, with jacta apparently the most iso- lated. As noted previously, procax and hoekeri approach one another in the vicinity of Bani, precisely in the area of rapid transition from the xeric Llanos de Azua to the more mesic coastal areas to the east. The Rio Ozama separates procax from parvoris, which is known from only two localities on the main- land and from a slightly differentiated population on Isla Cata- lina. We consider hoekeri, procax, and parvoris as a more or less sequential coastal series still maintaining its integrity in response to environmental and geographical influences. Jacta, on the other hand, along with richardthomasi, represents a very different sort of lizard. We feel that the jacta-richard- thomasi populations at one time (and perhaps still) occupied most of the extreme eastern end of the island. The presently restricted and scattered distribution and records for not only jacta but also parvoris, as well as the isolated occurrence of parvoris on Catalina, suggest strongly that the range of A. chrysolaema in this region is retracting, leaving isolated outliers which may be indicative of former populations. The absence of records of the species between San Pedro de Macoris and Juan- illo, as well as only two general localities of parvoris on the mainland, add substance to this supposition. The eastern dis- tribution of Ameiva lineolata tends also to bear out this conten- tion. 484 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Of the remaining- races from the southern part of the north island, only the strii)ed alacris is left. Alacris is so like procax in many features that it is difficult not to associate the two. On the other hand, they are separated presently by hoekeri, with which race alacris intergrades, as it does with chrijsolaema on the west. Tt is possible that alacris and procax were at one time confluent, but that bockcri has pushed between them, thereby severing- any direct genetic continuity between the two. An- other possibility is that alacris has been derived from chryso- laema, either across the Montagues de Trou d'Eau or along the valley of the Riviere de I'Artibonite. Continued evidence of intergradation between these two races near ]Mirebalais lends support to this possibility. A. c. defensor on the northwestern peninsula has obviously been long isolated from its more southern relatives. Presumably it has been derived from chrysolaema. The Gonave race secessa resembles defensor in pattern and cohn-ation, and it is possible that Gonave was colonized from tlie north (defensor) rather than from the adjacent mainland [chrysolaema) . The latter, however, seems more likely both on the basis of proximity and what is presently known of the origin of the Gonave herpetofauna. The relationships of regularis seem closest to the procax-alacris pair; possibility of origin of regularis from alacris via the Arti- bonite system and thence to the Grande Riviere du Xord has already been discussed. Other possibilities are an old origin from defensor (although this is not particularly appealing) or an origin from procax via the central valley to the east of the Cordillera Central in the Republica Dominicana. The only evi- dence against this is that A. chrysolaema does not occupy this val- ley today, although procax occurs at its southern end and regularis at its northern end uear Santiago. Finally, A. c. woodi on Tortue, although closer geographically to defensor than to regularis, seems closer in some characteristics to regularis than to defensor. There are, however, tendencies of pattern in defensor which herald the extreme peculiarities of woodi patterns. On the other hand, occasional regularis have dorsal patterns like some woodi. It seems more likely that ivoodi is a direct derivative of defensor. SCHWARTZ AND KLINIKOWSKI : AMEIVA 485 LITERATURE CITED Barbour, Thomas, and G. Kingley Noble 1915. A revision of the lizards of the genus Ameiva. Bull. Mus. Comp. Zool., 59(6):417-479. Cochran, Doris M. 1941. The herpetology of Hispaniola. Bull. U. S. Nat. Mus., 177: i-vii + 1-398, 12 pis., 120 figs. Maerz, a., and M. Rea Paul 1950. A dictionary of color. New York, McGraw-Hill Book Co., pp. i-vii, 1-23, 137-208, 56 pis. Mertens, Robert 1938. Aniphiliien und Reptilien aus Santo Domingo, gesanimelt von Dr. H. Boker. Senckenbergiana, 20: 332-42, 6 pis. 1939. Herpetologische Ergebnisse einer Reise naeh der Insel Hispan- iola, Westindien. Abh. senckenberg. naturf. Ges., 449: 1-84, 10 pis. Schwartz, Albert (in press) The Ameiva (Reptilia, Teiidae) of Hispaniola. I. Ameiva lineolata Dumeril and Bibron. Caribbean Jour. Sci. (Received February 25, 1965.) 486 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Subspecies chrysolaema woodi procax richardthomasi parvoris secessa jacta regularis umbratilis boeheri alacris defensor ficta abbotti leberi 160 141 141 137 137 135 134 132 130 126 126 126 121 117 111 130 126 116 124 113 111 128 112 111 109 106 113 108 104 Table 1. Subspecies of Ameiva chrysolaema ranked according to snout- vent length (in mm) of largest male for each race. Transverse ventrals Subspecies N Mean and extremes (mo( secessa 50 91.0 (80-99) 12 abbotti 32 89.6 (84-97) 12 richardthomasi 22 87.8 (80-92) 10 chrysolaema 106 86.7 (76-101) 12 leberi 48 85.5 (78-96) 10 alacris 48 84.8 (77-94) 10 procax 51 84.8 (76-98) 10 boelceri 86 84.6 (73-98) 10 parvoris 42 83.2 (76-92) 12 umbratilis 61 83.0 (73-100) 10 defensor 24 82.4 (77-90) 10 ficta 41 81.9 (75-92) 10 woodi 39 80.5 (66-93) 10 jacta 3 79.7 (77-82) 12 regularis 140 77.8 (67-88) 12 Table 2. Subspecies of Ameiva chrysolaema ranked according to means of number of fourth-toe subdigital scales; each race is also characterized by the modal number of transverse rows of ventral scales (but see discussions of umbratilis, procax, and richardthomasi). N ^ number of specimens ex- amined. SCHWARTZ AND KLINIKOWSKI : AMEIVA 487 Subspecies abbotti chrysolaema richardtJiomasi seoessa jacta leberi woodi ficta parvoris regularis defensor boeTceri procax umbratilis alacris Table 3. Subspecies of Ameiva chrysolaema ranked according to mean number of femoral pores; N = same as in Table 2. Femoral pores Mean and extremes 43.8 (35-52) 43.7 (33-50) 42.6 (39-47) 41.5 (36-46) 41.3 (39-43) 41.3 (35-45) 41.3 (36-46) 40.9 (33-47) 38.2 (24-45) 37.6 (32-44) 37.2 (30-41) 36.5 (31-41) 36.1 (30-43) 35.6 (28-42) 33.8 (30-39) Scales in 15th caudal verticil Subspecies Mean and extremes chrysolaema alacris regularis procax boeTceri umbratilis parvoris abbotti secessa ficta woodi richardthomasi defensor leberi jacta Table 4. Subspecies of Ameiva chrysolaema ranked according to mean num- ber of scales in fifteenth caudal verticil ; N = same as in Table 2. 44.4 (37-51) 44.0 (39-49) 43.1 (30-52) 42.8 (39-48) 42.7 (37-48) 42.7 (38-48) 42.7 (38-46) 40.4 (36-47) 40.1 (36-44) 38.8 (36-46) 38.8 (34-43) 38.6 (36-41) 38.0 (34-41) 37.9 (34-43) 35.7 (35-36) Bulletin of the Museum of Comparative Zoology HARVARD UNIVERSITY Vol. 133, No. 11 *N A NEW ATTEMPT TO CONSTRUCT LIFE TABLES FOR KENT ISLAND HERRING GULLS By Raymond A. Paynter, Jr. CAMBRIDGE, MASS., U.S.A. PRINTED FOR THE MUSEUM May 18, 1966 PUBLICATIONS ISSUED OR DISTRIBUTED BY THE MUSEUM OF COMPARATIVE ZOOLOGY HARVARD UNIVERSITY Bulletin 1863- Breviora 1952- Memoiks 1864-1938 Occasional Papers on Mollusks 194.5- JoHNSONiA, Department of Mollusks, 1941- Other Publications. Bigelow, H. B. and W. C. Schroeder, 1953. Fishes of the Gulf of Maine. Reprint, $6.50 cloth. Brues, C. T., A. L. Melander, and F. M. Carpenter, 1954. Classification of Insects. $9.00 cloth. Creighton, W. S., 1950. The Ants of North America. Reprint, $10.00 cloth. Lyman, C. P. and A. R. Dawe (eds.), 1960. Symposium on Natural Mammalian Hibernation. $3.00 paper, $4.50 cloth. Peters' Check-Hst of Birds of the World, vols. 2-7, 9, 10, 15. (Price list on request.) Turner, R. D., 1966. A Survey and Illustrated Catalogue of the Teredinidae (Mollusca: Bivalvia). $8.00 cloth. Whittington, H. B. and W. D. I. Rolfe (eds.), 1963. Phylo- geny and Evolution of Crustacea. $6.75 cloth. Proceedings of the New England Zoological Club 1899-1948. ( Complete sets only. ) Publications of the Boston Society of Natural History. Publications Office Museum of Comparative Zoology Harvard University Cambridge, Massachusetts 02138, U. S. A. Bulletin of the Museum of Comparative Zoology HARVARD UNIVERSITY Vol. 133, No. 11 A NEW ATTEMPT TO CONSTRUCT LIFE TABLES FOR KENT ISLAND HERRING GULLS By Raymond A. Paynter, Jr. CAMBRIDGE, MASS., U.S.A. PRINTED FOR THE MUSEUM May, 1966 Bull. Mus. Comp. ZooL, Harvard Univ., 133 (11): 489-528, May, 1966. No. 11 — A New Attempt to Construct Life Tables for Kent Island Herring Gulls'^ By Raymond A. Paynteb, Jr. INTRODUCTION An earlier attempt (Paynter, 1949) to construct a life table for the population of Herring Gulls {Larus argentatus sniith- sonianus) on Kent Island, New Brunswick, yielded results which seemed in conflict with the observed status of the colony. The life table indicated that the population was in a steep decline, although the size of the colony was thought to be constant or possibly even producing a population surplus which was con- tributing to the general increase in Herring Gulls that had been noted in northeastern North America for nearly half a century. To account for the discrepancy between the life table and what was believed to be the actual status of the population, it was suggested that, (1) the 1935 class (erroneously cited as "1936"), which was utilized to construct the postfledging portion of the table, had been banded too short a time (11 years) to yield all potential recoveries, thereby producing a truncated table; (2) that the 1935 year class may have suffered unusually severe post- fledging mortality or that the 1947 year class, which provided data for the egg and nestling portion of the table, may also have been unrepresentative, or possibly both situations had prevailed ; (3) that there may have been a loss of bands, particularly among older birds, that caused the calculated survival rate and life ex- pectancy to be lowered. Fifteen years have elapsed since that study. Additional birds from the 1935 class have been recovered and it is now possible to test the hypothesis that older recoveries might be sufficient to raise the calculated life expectancy and survival rates to levels commensurate with the presumed status of the population. Re- coveries from bandings in other years at Kent Island have also accumulated, allowing comparisons between several year classes. Finally, there have been published three additional life tables for the species (Paludan, 1951; Hickey, 1952; Olsson, 1958) and these provide valuable comparative data. It is the purpose of this 1 Contribution No. 32 from the Bowdoin Scientific Station, Kent Island, Grand Munau, Xew Brunswick, Canada. 492 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY paper to re-examine the dynamics of the Kent Island Herring Gull population in the light of these developments. MATERIAL AND METHODS In calculating ages of banded gulls it has been the custom at Kent Island (e.g., Paynter, 1947; 1949) to begin the year on July first, which is about the earliest chicks are large enough to band. To avoid the inclusion of prefledging mortality in the banding recoveries, bands were removed from any chicks dying before fledging and were placed on other young. In some banding studies it is necessary to make adjustments for bias resulting from the disproportionate recovery of newly fledged birds in the vicinity of their place of birth (see Earner, 1955). This correction is not required of the Kent Island data because the gulls leave the island shortly after they are able to fly and disperse over a wide area (Gross, 1940) where, presum- ably, they are no more likely to be recovered than older birds. Banding began at Kent Island in 1934. The terminal date for data used in this study is 30 June 1963. Thus, the oldest poten- tial recovery would be a bird in its twenty-eighth year. To date the oldest record of a gull banded as a fledgling is an individual caught at Kent Island in its twenty-sixth year and released bear- ing a new band. The oldest record of a bird dying while still wearing its original band is a gull in its twenty-fourth year ; the next oldest is a twenty-second year bird. For purposes of this study only young banded in the six year classes from 1934 through 1939 will be considered. The maximum potential age for birds in the most recently banded year class is twenty -three years, which seems close to the maximum age of recovery that may be expected in the Kent Island population (see p. 514). A few birds older than twenty-six almost certainly will be found in future years, but thej^ probably play an insignificant part in the dy- namics of this population. Heeding Hickey's warnings (1952) of clerical and other errors in the banding records kept by the Fish and Wildlife Service, microfilms of all Kent Island banding schedules and recovery records were obtained through the courtesy of Allen J. Duvall. These were carefully compared with the files at Kent Island and all erroneous and questionable data were eliminated. I have no illusions about the complete accuracy of the resulting material, but considering the fact that hundreds of people have been in- volved in reporting and handling these data during three decades, 3,646 125 3.43 10,748 352 3.27 6,665 254 3.81 4,652 146 3.14 2,983 77 2.58 3,000 145 4.83 HERRING GULL LIFE TABLES 493 no further refinements seem possible. It is believed that the quan- tit}^ of data is sufficient to offset whatever deficiencies in quality remain. TABLE I Kent Island Fledglings Banded and Admissible i Eecoveries Year Banded Eecoveries Per cent 1934 1935 1936 1937 1938 1939 Total 31,694 1,099 3.47 1 See below for deflnition of admissible recoveries. From 1934 through 1939 nearly 32,000 fledgling gulls were banded on Kent Island (Table I). Discrepancies between the yearly totals in Table I and those published previously (Paynter, 1947) result from the re-analysis of the Fish and Wildlife Serv- ice records. These 32.000 bandings have yielded 1,206 recoveries and re- turns. It is necessary, however, to remove from consideration records in certain categories. Because the Kent Island life table is to be constructed from a mortality series, no records of living birds ("returns") are admissible. As a consequence, those in- dividuals which were trapped in later years at Kent Island and found bearing bands are eliminated, as are those birds which were reported to the Fish and Wildlife Service as having been "captured and released." On the grounds of uncertainty, whether the banded bird was living or dead, Hickey (1952, p. 94) probably would exclude those records reporting "no informa- tion," "observed," "found ill," "caught by fisherman," etc. Some of these reports doubtless pertain to gulls not yet dead when their bands were read, but it seems unlikely that many of these records could have been obtained from birds that were not already weakened from illness or injury and soon to die. For this reason I include these records in the mortality series. Also eliminated from the study are gulls collected in connec- tion with research at Kent Island and those shot elsewhere under 494 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY gull control permits. Neither of these hazards operated con- sistently from year to year. Lack (1954, p. 91) notes that the inclusion of birds which have been recovered by means of shooting (i.e. "normal" shooting — not systematic control) may introduce bias if this category repre- sents a substantial portion of the recovery sample and if juve- niles are more easily shot than adults. In certain instances this may be correct, but if a given cohort is hunted with equal inten- sity throughout its life span, shooting must be considered to be merely another one of the hazards acting upon a population, such as the stress of weather and the toll of predators. The fact that young birds may be more easily shot than older ones does not bias the sample any more than does the fact that adverse weather or predators may claim more young than adults. On the other hand, banded birds which are shot, or for that matter killed tlirough any human agency, probably are more likely to be re- ported than birds dying of disease or other natural causes. Hu- man activity, therefore, may appear as a disproportionately high cause of death in the sample of recoveries and may also increase the total number of recoveries. But, as long as the human activity causing death operates consistently throughout the life of a co- hort, the recoveries need not be excluded from the mortality series used in preparing a life table. About five per cent of all Kent Island recoveries are of indi- viduals reported as shot ; the true rate is probably higher but, because the species is protected by law, is concealed within the reports as "found," "no information," etc. The actual percent- age is certainlj^ not nearly so high as in Europe, where the species is unprotected and where about 60 per cent of all reports are of gulls which have been shot (Paludan, 1951 ; Olsson, 1958). There is no reason to suspect that the distribution by age or year class of recoveries of Kent Island gulls shot but reported in other categories differs from that of birds accurately recorded as shot. Taking the shooting reports at face value, we find considerable variation in the percentage of these recoveries within various age classes, as well as within year classes. For example, in the 1935 cohort, about 6.5 per cent of the gulls were reported as shot ; there were no records of birds shot beyond their seventh year and the percentages of reports from the first year through the seventh year are 8.4, 2.9, 6.5, 4.1, 16.0, 0, and 5.2. In contrast, in the 1936 year class, only 3.6 per cent of the records are of birds shot ; the oldest was in its twelfth year and the percentages by age classes run 4.8, 3.0, 0, 0, 5.2, 0, 0, 0, 0, 50.0, 0, and 50.0. The HERRING GULL LIFE TABLES 495 distribution of shooting reports illustrated here is similar in all six cohorts. It is of interest that while shooting occurs only in the first half of the maximum potential life span (there are no records beyond the thirteenth year), there is no clear-cut pattern within this period. One might have expected to find shooting relatively more common among immature (and inexperienced) birds. Birds recovered after having been rebanded are excluded. If Kent Island gulls lose bands, which they doubtless do, and if the loss is a function of the age of the bands rather than a random occurrence throughout the life span of the birds, this may be evi- dent in the life table, provided it is not obscured by other phe- nomena, as a declining survival rate. The inclusion of rebanded birds in the study, while possibly giving a truer picture of age- specific survival rates, would introduce another source of bias. Unfortunately, there has been an insufficient number of reband- ings to allow an analysis of birds in this category. The Kent Island life tables are begun with the laying of the first egg, which occurs in late May, roughly one-tenth of a year earlier than the July 1 date taken as the start of the year when calculating the ages of banded birds. Because there were no studies of the survival of the eggs and chicks at Kent Island during the period 1934-1939, the records of the 1947 year class (Paynter, 1949) are used, creating composite tables. The mean incubation period is taken at 28 days, or 0.07 years. In 1947 it was found that 28.6 per cent of the eggs failed to hatch ; the mean age of the young at fledging was calculated to be 43 days (0.12 years), and 48.5 per cent of the chicks are believed to have failed to fledge, giving a production of 0.92 fledglings per breed- ing pair. KENT ISLAND LIFE TABLES Composite life tables, using the 1947 prefledging data, for the six-year classes from 1934 through 1939, as well as a table for the combined year classes, are presented in Table II. Semi-loga- rithmic survivorship curves (Ix) are shown in Figures 1 and 2. The table and graph for the combined six-year classes (Table II; Fig. 2) indicate that about 63 per cent of the eggs laid fail to produce fledged young (see Paynter, 1949, for discussion). Then there is heavy mortality (45 per cent) from the time of fledging until the following July 1, a somewhat lessened rate for the second year of life (35 per cent), and finally a lower average 496 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY rate for the bulk of the remaining life span (26.8 per cent for the years 3 through 15). This is the expected pattern for most long-lived avian populations. It suggests that young (post- fledging) gulls are less able to cope with the hazards of their environment and that birds surviving the first two years of life have either learned to avoid more of the hazards, or the less adept ones have been eliminated through natural selection. If data for the oldest age groups were more abundant one might find the mortality rate increasing because of the senility. On the other hand, it is unlikely many animals survive this long in nature and that we shall ever have sufficient data from senile individuals to document this phenomenon, if indeed it does occur. Examined individually, the life tables and survivorship curves (Table II; Fig. 1) show interesting similarities as well as dif- ferences, which one expects of natural populations in a variable environment. For example, in four of the six cohorts the mor- tality rate declined the year following fledging, but in the 1936 and 1938 year classes it rose. The first year postfledging mor- tality rate for the 1936 cohort was 36.2 per cent, which is un- usually low and as a result was slightly below the rate of the second year (39.5 per cent), which was only a little in excess of the average (35 per cent). The same pattern was displayed, in a more exaggerated manner, in the 1938 cohort. The mortality rate in the first year was a comparatively low 37.7 per cent, but in the second year it jumped to 52.1 per cent, the highest observed. Similar yearly fluctuations occur throughout the life tables. An- nual variations in the severity of the weather are presumably responsible for many of these swings, but the problem has not been studied. Next to be considered is the reproductive potentiality of the colony as indicated by the life tables. North American gulls are assumed to breed in their fourth year (year 3-4) or, in other words, three summers after hatching. This may not be an exact assumption for Drost, Focke, and Freytag (1961) found in a German colony of European Herring Gulls {L. a. argentatus) that 20 per cent bred in their third year, 25 per cent in their fourth year, and 55 per cent in their fifth year, with the mean breeding age slightly in excess of four years. Paludan (1951), however, concluded that Danish gulls regularly breed in their third 3^ear, basing this on a single observation of a bird of this age in a breeding colony. The gull had some dark areas on the tail and on the lesser wing coverts. From this Paludan assumed that fully adult plumage is attained in the third year and that HERRING GULL LIFE TABLES 497 TABLE II Composite Life Tables for Kent Island Cohorts 1934-1939, Utilizing 1947 Prefiedging Mortality Data of Paynter (1949) 1934 Cohort X 1 d l,000q^ e„ X X X X Age 1 ^0. surviving to No. dying in Mortality rate Expectation of life in start of age age interval per 1,000 alive remaining to those years interval out of out of 1,000 at start of age attaining the age 1,000 eggs laid eggs laid interval interval (in years) 0.00-0.07 1,000.0 286.0 286.0 .97 0.07-0.19 714.0 346.3 485.0 1,28 0.19-1.10 367.7 182.4 496.1 2.31 1.10-2.10 185.3 44.1 238.0 3,23 2.10-3.10 141.2 26.5 187.7 3.09 3.10-4.10 114.7 20.1 175.2 2.68 4.10-5.10 94.6 38.4 405.9 2.15 5,10-6.10 56.2 23.7 421,7 2.28 6.10-7.10 32.5 11.8 363.1 2.57 7.10-8.10 20.7 8.9 429.9 2,76 8.10-9.10 11.8 0 0 3,47 9.10-10.10 11.8 3.0 254.2 2,47 10.10-11.10 8.8 3.0 340.9 2,14 11.10-12.10 5.8 0 0 1,98 12.10-13.10 5.8 2,9 500.0 ,98 13.10-14.10 2.9 2.9 1,000.0 .50 1935 Cohort 0.00-0.07 1,000.0 286.0 286.0 .94 0.07-0.19 714.0 346.3 485.0 1,23 0.19-1.10 367.7 176.5 480.0 2.21 1.10-2.10 191.2 70.0 366.1 2.92 2.10-3.10 121.2 30.3 250,0 3.33 3.10-4.10 90.9 23.0 253.0 3.27 4.10-5.10 67.9 20.9 307.8 3.21 5.10-6.10 47.0 13.6 289.4 3.41 6.10-7.10 33.4 10.5 314.4 3.60 7.10-8.10 22.9 4.2 183.4 4.02 8.10-9,10 18.7 4.2 224.6 3.81 9.10-10.10 14.5 2.1 144.8 3.76 10.10-11.10 12.4 2.1 169.3 3.32 11.10-12.10 10.3 4.1 398.1 2.90 12.10-13.10 6.2 1.0 161.3 3.50 13.10-14.10 5.2 2.1 403.8 3.08 14.10-15.10 3.1 1.0 322.6 3.84 15.10-16.10 2.1 0 0 4.43 16.10-17.10 2.1 0 0 3.38 17.10-18.10 2.1 1.1 523.8 2.38 18.10-19.10 1.0 0 0 3.50 19.10-20.10 1.0 0 0 2.50 20.10-21.10 1.0 0 0 1,50 21.10-22.10 1.0 1.0 1,000,0 .50 498 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY 1936 Cohort X 'x d X 1,000 q^ \ 0.00-0.07 1,000.0 286.0 286.0 1.04 0.07-0.19 714.0 346.3 485.0 1.38 0.19-1.10 367.7 133.1 362.0 2.50 1.1O-2.10 234.6 92.7 395.1 2.75 2.10-3.10 141.9 39.2 276.2 3.23 3.10-4.10 102.7 23.1 224.9 3.27 4.10-5.10 79.6 27.5 345.5 3.07 5.10-6.10 52.1 18.8 360.8 3.42 6.10-7.10 33.3 8.7 261.3 4.08 7.10-8.10 24.6 4.4 178.9 4.35 8.1O-9.10 20.2 5.8 287.1 4.19 9.10-10.10 14.4 2.9 201.4 4.68 10.10-11.10 11.5 1.4 121.7 4.74 11.10-12.10 10.1 2.9 287.1 4.33 12.10-13.10 7.2 1.4 199.4 4.87 13.10-14.10 5.8 1.5 258.6 4.93 14.10-15.10 4.3 0 0 5.49 15.10-16.10 4.3 1.4 325.6 4.56 16.10-17.10 2.9 0 0 5.41 17.10-18.10 2.9 0 0 4.41 18.10-19.10 2.9 0 0 3.41 19.10-20.10 2.9 1.5 517.2 2.41 20.10-21.10 1.4 0 0 3.50 21.10-22.10 1.4 0 0 2.50 22.10-23.10 1.4 0 0 1.50 23.10-24.10 1.4 1.4 1,000.0 .50 1937 Cohort X 'x ''x 1,000 q^ % 0.00-0.07 1,000.0 286.0 286.0 .85 0.07-0.19 714.0 346.3 485.0 1.10 0.19-1.10 367.7 176.5 480.0 1.96 1.10-2.10 191.2 52.8 276.1 2.45 2.10-3. 10 138.4 37.8 273.1 2.19 3. 10-4. 10 100.6 22.6 224.6 1.83 4.10-5.10 78.0 30.2 387.2 2.24 5.10-6.10 47.8 20.1 420.5 2.34 6.10-7.10 27.7 5.0 180.5 2.68 7.10-8.10 22.7 10.1 445.0 2.16 8.10-9.10 12.6 0 0 2.50 9.10-10.10 12.6 5.0 396.8 1.50 10.10-11.10 7.6 2.5 328.9 1.16 11.10-12.10 5.1 5.1 1,000.0 .49 HERRING GULL LIFE TABLES 499 1938 Cohort X 1 d 1,000 q e X X ' ^x x 0.00-0.07 1,000.0 286.0 286.0 ,89 0.07-0.19 714.0 346.3 485.0 1.16 0.19-1.10 367.7 138.6 376.9 2.09 1.10-2.10 229.1 119.4 521.2 2.16 2.10-3.10 109.7 33.4 304.5 2.97 3.10-4.10 76.3 14.3 190.0 3.06 4.10-5.10 62.0 9.6 154.8 2.65 5.10-6.10 52.4 19.1 364.5 2.04 6.10-7.10 33.3 9.5 285.3 1.92 7.10-8. 10 23.8 9.5 399.2 1.50 8.10-9.10 14.3 4.8 335.7 rj6 9.10-10.10 9.5 9.5 1,000.0 .49 1939 Cohort x 'x "'x 1,000 q^ \ 0.00-0.07 1,000.0 286.0 286.0 1.04 0.07-0.19 714.0 346.3 485.0 1.38 0.19-1.10 367.7 182.8 497.1 2.49 1.10-2.10 184.9 71.0 384.0 3.60 2.10-3.10 113.9 32.0 280.9 4.54 3.10-4.10 81.0 10.1 124.7 5.18 4.10-5.10 70.9 12.7 179.1 4.85 5.10-6.10 58.2 7.6 130.6 4.79 6.10-7.10 50.6 7.6 150.2 4.44 7.10-8. 10 43.0 7.6 176.7 4.13 8.10-9.10 35.4 2.5 70.6 3.91 9.10-10.10 32.9 7.6 231.0 3.18 10.10-11.10 25.3 5.1 201.6 2.98 11.10-12.10 20.2 7.6 376.2 2.61 12.10-13.10 12.6 0 0 2.88 13.10-14.10 12.6 7.6 603.1 1.88 14.10-15.10 5.0 2.5 500.0 2.98 15.10-16.10 2.5 0 0 4.48 16.10-17.10 2.5 0 0 3.48 17.10-18.10 2.5 0 0 2.48 18.10-19.10 2.5 0 0 1.48 19.10-20.10 2.5 2.5 1,000.0 .48 500 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Combined 1934-1939 Cohorts X 'x "x 1/000 q^ ^ 0.0O-0.07 1,000.0 286.0 286.0 .95 0.07-0.19 714.0 346.3 485.0 1.25 0.19-1.10 367.7 165.5 450.1 2.26 1.10-2.10 202.2 71.7 354.6 2.82 2.10-3. 10 128.5 33.4 260.0 3.27 3.10-4.10 95.1 20.4 214.5 3.30 4.1O-5.10 74.7 23.7 317.3 3.07 5.10-6.10 51.0 15.3 300.0 3.26 6.10-7.10 34.7 9.0 259.4 3.56 7.10-8.10 25.7 6.4 249.0 3.63 8.10-9.10 19.3 3.3 171.0 3.67 9.10-10.10 16.0 4.0 250.0 3.33 10.10-11.10 12.0 2.4 200.0 3.27 11.10-12.10 9.6 3.6 375.0 2.95 12.10-13.10 6.0 1.0 166.7 3.43 13.10-14.10 5.0 2.3 460.0 3.02 14.10-15.10 2.7 .7 259.3 4.18 15.10-16.10 2.0 .3 150.0 4.50 16.10-17.10 1.7 0 0 4.23 17.10-18.10 1.7 .3 176.5 3.23 18.10-19.10 1.4 0 0 2.86 19.10-20.10 1.4 .7 121.4 1.86 20.10-21.10 .7 0 0 2.29 21.10-22.10 .7 .4 571.4 1.28 22.10-23.10 .3 0 0 1.33 23.10-24.10 .3 .3 1,000.0 .33 the bird was merely retarded. This appears to be an unneces- sarily complicated interpretation. It would seem more logical to conclude that this was a case of an early breeder. Lacking data for the Kent Island gulls, we shall accept the fourth year as the mean breeding age. In 1947 the average clutch size at Kent Island was nearly 2.5 eggs (Paynter, 1949). Observations (unpublished) in 1948 indi- cated a somewhat higher mean for first clutches (ea. 2.75) and a lower mean (ca. 2.00) for clutches replacing those lost through predation, but the average for all final clutches was again close to 2.5 eggs. Taking the life table for the combined years 1934-1939, we find that from an initial cohort of 1,000 eggs 95.1 birds survive to July 1 of the third summer after hatching. It will be recalled that July 1 is the time when banding of the chicks is first practi- cal and that egg-lajdng begins about one-tenth of a year earlier, HERRING GULL LIFE TABLES 501 in late May. We must, therefore, take into account birds which are alive at the time of laying but which do not survive until July 1. From the life table it is seen that 33.4 birds die during the third year. If the mortality rate is constant throughout the year, one-tenth, or 3.3 birds, die between the time of laying and the first of July. These individuals may then be added to those alive on July 1, giving a total of 98.4 birds which survive to breed for the first time from 1,000 eggs laid four summers earlier. If the sex ratio is equal, and there is no evidence that it is not, there is an average of 49.2 pairs alive at the start of the first breeding season. At this point these birds have a mean life ex- pectancy of about 3.3 years. If they breed annually, which they proba])h'- do for the greater part of their life spans, and lay a yearly average of 2.5 eggs per clutch, the 49.2 pairs are capable of producing 405.9 eggs. This is only 41 per cent of the number required to maintain a stable population. The only year class with a life table differing materially from the combined 1934-1939 table is the 1939 cohort, with its gen- erally better survival rate among older birds. The calculated number of birds at the start of the first breeding season is 84.2. The mean life expectancy is about 5.2 years. This would allow for a lifetime production of approximately 547 eggs, which again is still far short of the 1,000 eggs needed for a stable population. If these life tables accurately represent the dynamics of the Kent Island population, it is obvious that the size of the colony should have been rapidly declining in the 1930 's and early 1940 's, and that the magnitude of the decline would be so great that it would be apparent to field observers. Only constant, large- scale immigration could obscure this phenomenon. TTnfortu- nately, the colony was not accurately censused until 1940 (Crys- tal, 1941), when approximately 12,000 nests were counted, and the presence of 3,000 more was estimated. Allowing for the inclusion of "play nests" (Goethe, 1937), there must have been a breeding population well in excess of 20,000 birds. During my field Avork from 1946 through 1948 the colony was densely popu- lated and I estimated it to contain 25,000 to 30,000 individuals (Paynter, 1949). The late Ernest Joy informed me that he had noticed no change in the size of the colony during his tenure as warden from 1935 to 1948. Thus, we must conclude that the Kent Island population was stable, or at least not noticeably un- stable, from at least 1935 through 1948. The possibility remains that the colony was failing to reproduce itself and was dependent 502 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY 1,000 < to o LLl Q. o > > => to I I I I I I ' I I I ' I ' I ■ I I I.I 3.1 5.1 7.1 9.1 11,1 13.1 15,1 17.1 19.1 21.1 23. AGE IN YEARS Fig. 1. Survivorship curves for Kent Island cohorts 1934-39. HERRING GULL LIFE TABLES 503 1,000 . Kent Id., 1934-'39 Fenno-Scandia, l925-'54 N. Amer., l925-'30 Denmark, 19l7-'42 100 _ Z3 o UJ a. CO o > 1 _l AGE IN YEARS Fig. 2. Survivorship curves for four populations. 504 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY on immigrants to maintain a constant size, but this seems im- probable. The Herring Gull has been increasing along the east- ern seaboard, as well as in Europe, since the turn of the century. It would appear unlikely that the Kent Island colony was not contributing to this general increase. A more reasonable con- clusion is that the six life tables are deficient. Before speculating on the possible reasons for this, life tables for other populations of the species will be examined. OTHER LIFE TABLES Hickey (1952) has considered the first three life tables pre- pared for North American Herring Gulls (viz., Marshall, 1947 ; Paynter, 1947, 1949), commenting on their errors and discrepan- cies; they need not be reconsidered here. Then using recoveries of gulls banded at various colonies in North America from 1925 through 1930, Hiekej^ constructed an abridged life table for a TABLE III Life Table for Theoretical Population of North American Gulls Banded 1925-1930 (504 Eecoveries) (after Hickey, 1952) X 1 X d X 1,000 q X e X 0-1 1,000.0 599.2 599.2 2.00 1-2 400.8 117.1 292.1 3.25 2-3 283.7 59.5 209.7 3.39 3-4 224.2 61.5 274.3 3.15 4-5 162.7 41.7 256.2 3.16 5-6 121.0 45.6 376.8 3.08 6-7 75.4 16.1 213.5 3.63 7-8 59.3 7.9 133.2 3.48 8-9 51.4 11.9 231.5 2.95 9-10 39.5 9.9 250.6 2.69 10-11 29.6 5.9 199.3 2.42 11-12 23.7 9.9 417.7 1.90 12-13 13.8 4.0 289.8 1.91 13-14 9.8 2.0 204.1 1.48 14-15 7.8 5.9 756.4 .73 15-16 1.9 1.9 1,000.0 .47 HERRING GULL LIFE TABLES 505 theoretical population of 504 birds, beginning the table with re- ports received subsequent to 31 August of the year in which the gulls were hatched. No reports were available for gulls older than their sixteenth year. Although the table is probably fore- shortened, recoveries beyond the sixteenth year are so infrequent that the resulting table is doubtless close to what would have been found if it had not been terminated until all potential re- coveries had been obtained. Adjustments were made for yearly variations in the number of birds banded and for the fact that some birds had not been banded sufficiently long to yield recov- eries in the year 15-16 (see Hickey, 1952, p. 11). This life table, recalculated to form a cohort of 1,000, is presented in Table III, and the survivorship curve is plotted in Figure 2. The first year mortality rate is about 60 per cent, in contrast to an average of 45 per cent at Kent Island from the time of fledging in August to the following June 30, and the second year mortality is 29 per cent, compared to the Kent Island figure of 35.5 per cent. The mean annual mortality rate for the next ten years is 25.6 per cent, which is almost exactly that at Kent Island (Table IV). Because this table is begun in September, neglecting the egg, nestling, and earliest postfledging mortality, a different tech- nique from that used with the Kent Island data must be em- ployed to assess the reproductive potential of the population. TABLE IV Mean Mortality Kates Age Population 0-1 1-2 2-12 Kent Id., 1934- '39 45.01 35.5 26.0 North Anier. (Hickey, 1952) 59.9^ 29.2 25.6 Denmark (Paludan, 1951) 62.32 21.7 15.6 Fenno-Scandia (Olsson, 1958) 56.8^ 31.9 33.8 iFirst interval 0.9 year, i.e. fledging to 30 June following year; 1 July to 30 June thereafter. ^Year begins 1 September. 3 Year begins 1 August. 506 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY This is done by totaling the l^ column (survivors) from the year 3-4 (first breeding year) to the end of the table, dividing the total by two to give the number of pairs of breeding birds, and, finally, dividing the initial cohort (1,000) by the number of pairs of breeding birds (see Hickey, 1952, pp. 94-95). The result is the average number of young per pair of adults which need be raised to the first of September of the year of hatching in order to maintain a stable population. Applying this to Hickey 's table, we find that an average of 2.44 young must be fledged and survive to September 1. This is only slightly below the mean clutch size found at Kent Island (2.5) and, therefore, allows almost no egg or chick mortality. Even if the mean clutch is assumed to be three, which is the maximum number of eggs laid by Herring Gulls, except for an exceedingly rare clutch of four (about 0.6 per cent of all clutches [Paludan, 1951, p. 49]), this would also allow for but 19 per cent mortality between the laying of the first egg and September 1. Such a mortality rate seems much too low in the light of observations at Kent Island, where more than 60 per cent of the eggs failed to yield fledged young. This life table cannot, therefore, be accepted. The first life table for a European population of Herring Gulls is that constructed by Paludan (1951) for 966 recoveries, including those shot (ca. 60 per cent), from 11,689 birds banded in Denmark from 1917 through 1942 (Table V). The oldest recov- ery is that of an individual in its twenty-sixth year ; the oldest po- tential recovery at the time the table was compiled would have been a thirty-fourth year bird. Adjustments, similar to those of Hickey (1952), w^ere made to compensate for the fact that only 8 of the 26 year classes had been banded sufficiently long to yield recoveries in the twenty-sixth year. The mean recovery rate for these eight cohorts was 8.54 per cent, versus only 3.47 per cent for the six cohorts at Kent Island. The marked difference in re- coveries is doubtless due to the frequency with which Danish gulls are shot, which enhances the chances that a banded bird will be reported, and also probably due to a better retention of bands, which will be considered subsequently. From this table (Table V), which starts on September 1 following hatching, it is seen that the first year mortality is 62.3 per cent, closely ap- proximating Hickey 's finding in North America. The second year mortality drops to 21.7 per cent, which is considerably lower than that in Hickey 's table (29.2 per cent) or the Kent Island table (35.5 per cent). The mean annual mortality for the next ten years is only 15.6 per cent, in contrast to the North American HERRING GULL LIFE TABLES 507 TABLE V Life Table for Danish Gulls Banded 1917-1942 (966 Eecoveries) (Paludan, 1951) X 1 d 1,000 q e X X X X 0-1 1,000 623 623 2.72 1-2 377 82 217 5.39 2-3 295 56 190 5.75 3-4 239 37 155 5.98 4-5 202 28 129 5.98 5-6 174 30 172 5.83 6-7 144 18 125 5.95 7-8 126 21 167 5.73 8-9 105 17 162 5.79 9-10 88 14 159 5.83 10-11 74 6 81 5.84 11-12 68 15 221 5.35 12-13 53 4 75 5.68 13-14 49 6 122 5.12 14-15 43 7 163 4.72 15-16 36 6 167 4.56 16-17 30 8 267 4.42 17-18 22 2 91 4.96 18-19 20 2 100 4.37 19-20 18 3 167 3.81 20-21 15 0 0 3.37 21-22 15 3 200 2.37 22-23 12 7 583 1.83 23-24 5 0 0 2.50 24-25 5 0 0 1.50 25-26 5 5 1,000 0.50 and Kent Island rates of about 26 per cent (Table IV). This low rate gives a survivorship curve (Fig. 2) which is strikingly dif- ferent, after the second year, from those of the Kent Island and North American populations. Applying the same methods for determining the required productivity of this population as were used with Hickey's data, 508 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY it is found tliat to maintain a constant population it is necessary for a pair of gulls to raise 1.3 young- to September first follow- ing hatching. While this would necessitate about 40 per cent greater productivity than recorded (Paynter, 1949) at Kent Island (0.92 young per pair raised to fledging in early August), it may be a reasonable expectation. On the other hand, the pre- fledo'ing mortalitv in Paludan's colonv was considerably higher than that at Kent Island. In the Danish colony in 1943 there was about a 10 per cent loss of eggs (vs. 28.6 per cent at Kent Island) and an 80 per cent loss of chicks (vs. 48.5 per cent at Kent Island). With an average clutch of three eggs this heavy mor- tality would result in a net production of about 0.5 fledgling for each pair of adults or, in other words, only a maximum of 39 per cent of the required productivity as calculated from the life table. Paludan believed that fledgling production nearly as low as that observed in the Danish population would be sufficient to maintain the population if certain deficiencies in the life table were corrected. He reasoned that the number of birds which breed for the first time should equal the number that die during that year. From the life table he found the mean annual mor- tality from the second year onward to be about 15 per cent. This would mean that in a population of 1,000 birds, 150 would die during the year and 150 should begin to breed for the first time, if the population is to remain stable. Using the life table figure of 62.3 per cent mortality for the first year and 15 per cent mean annual mortality thereafter, he calculated a production of ap- proximately 600 fledglings per 1,000 adults (1.2 per pair) would be necessary to yield about 150 birds two and a half years later, when he thought breeding began. This would require more than twice as many survivors as he had observed (0.54 per pair). However, Paludan reasoned that the observed first year mor- tality rate was higher than it should be owing to bias in favor of recoveries near the breeding colonies ; a reduction in the first year mortality would increase the number of birds surviving to breeding age and, hence, reduce the required production of fledg- lings per pair. Also, he believed the prefledging mortality that he had recorded was in excess of that which is normal in Denmark. He concluded that the yearly production of between 0.5 and 1 fledgling per pair of adults would be sufficient to maintain the population, and even allow for the general increase that had been noted throughout Europe. HERRING GULL LIFE TABLES 509 One significant probable error is the assumption that the en- tire population begins to breed in the third year. As we have seen above, a few gulls (20 per cent) do breed this early, but the average age is somewhat in excess of the fourth year. Wlien Paludan's statistics are adjusted for this later breeding age, it is found that a pair of gulls would have to produce roughly 1.4 fledglings, rather than 1.2 fledglings. TABLE VI Life Table for Daaiish Gulls, Utilizing Kent Island Prefiedging Data and Assuming First Year Recoveries to be Half of Those Actually Recorded (see i>. 508 for explauation) X I d 1,000 q e XX XX O-0.07 1,000.0 286.0 286.0 1.43 0.07-0.19 714.0 346.3 485.0 1.92 0.19-1.10 367.7 166.6 453.0 3.56 1.10-2.10 201.1 43.8 217.8 5.37 2.10-3.10 157.3 30.0 190.7 5.73 3.10-4.10 127.3 19.9 156.3 5.96 4.10-5.10 107.4 14.9 138.7 5.97 5.10-6.10 92.5 16.0 173.0 5.85 6.10-7.10 76.5 9.6 125.3 5.97 7.10-8.10 66.9 11.2 167.4 5.75 8.10-9.10 55.7 9.1 163.4 5.81 9.10-10.10 46.6 7.5 160.9 5.85 10. 10-n.lO 39.1 3.2 81.8 5.88 11.10-12.10 35.9 7.5 208.9 5.36 12.10-13.10 28.4 2.2 77.5 5.64 13-10-14.10 26.2 3.2 122.1 5.07 14.10-15.10 23.0 3.8 165.2 4.71 15.10-16.10 19.2 3.2 166.7 4.54 16.10-17.10 16.0 4.3 268.7 4.35 17.10-18.10 11.7 1.1 94.0 4.77 18.10-19.10 10.6 1.1 103.8 4.22 19.10-20.10 9.5 1.6 168.4 3.65 20.10-21.10 7.9 0 0 3.29 21.10-22.10 7.9 1.6 202.5 2.29 22.10-23.10 6.3 3.7 587.3 1.73 23.10-24.10 2.6 0 0 2.50 24.10-25.10 2.6 0 0 1.50 25.10-26.10 2.6 2.6 1,000.0 .50 510 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY If normal fledgling production in the Danish population is assumed to be one per pair of adults, which is the maximum sug- gested by Paludan, it is necessary for the first year mortality to drop from the observed rate of 62.3 per cent to approximately 52 per cent, in order to replace the 15 per cent loss of breeding birds. This requirement seems reasonable if, as believed by Palu- dan, there was a disproportionately high recovery of first-year birds in the sample. It is instructive to note (Table VI) that if the production of fledglings is assumed to be the same as at Kent Island (0.92 young per pair of adults) and if the recovery of first year gulls is halved, the first year mortality rate would be 45.3 per cent, which is almost exactly that found at Kent Island.^ x\t sexual maturity there then would be nearly 64 pairs of birds, with a life expectancy of almost six years. If these birds laid an average of 2.5 eggs per clutch, there would be produced about 950 eggs, or nearly sufficient production to maintain the population. If there were three eggs per clutch there would be a net production of about 1,140 eggs, adequate productivity to permit about a five per cent annual increase in fledged birds. From these calculations one is tempted to conclude that Palu- dan's first-year recoveries were twice as abundant as they would have been had the birds dispersed as widely as they do in suc- ceeding years and that Paludan was correct in suggesting that breeding pairs need produce only about one fledged chick per year. However, there are certain peculiarities in the recovery data that require examination. Paludan (1951, Table 29, pp. 108-109) presents a convenient tally of the annual recoveries for the year classes from 1917 through 1942. These records, in contrast to the Kent Island data, show extremely wide annual variations. For example, in the first year after banding the percentage of the total banded birds recovered ranged from 1.0 per cent in 1939 to 26.7 per cent in 1929, and in the second year the range was from 0 in five year classes to 4.0 per cent in 1937. Doubtless some of these great fluctuations are caused by sampling vagaries owing to the small number of birds banded each year and also to wide variations even within these small numbers (15 to 1,291 bandings per year). Nevertheless, there would seem to be some fluctuations which can- not be dismissed as sampling errors. These are well illustrated 1 For these calculations it is necessary to assume that the Danish records begin on 1 .Tuly ami that the lirst year recoveries are made in nine-tenths of a year, and thus are comparable with the Kent Island data. HERRING GULL LIFE TABLES 511 by the second year recoveries for the year classes of 1935, 1940, 1941, and 1942, years in Avhich the number of birds banded did not differ greatly. The number of gulls banded in these years was 765, 811, 861, and 710, respectively, and the number of sec- ond j^ear recoveries, in the same order, were 0, 11, 1, and 1. Ex- pressed differently, the recovery rates were 0, 1.36, 0.12, and 0.14 per cent. We need not know the reasons for these fluctuations to ap- preciate what profound effects they have on a demographic study. The life table constructed by Paludan shows a mortality rate in the second year of nearly 22 per cent, a rate, as has been pointed out, considerably lower than that recorded for any other popu- lation. This rate was obtained from data which indicate that about 0.77 per cent of all Danish recoveries were made in the second year. This is the average for all 26 year classes. If, how- ever, we eliminate the year classes for which there were no re- coveries during the second year the average rises to 0.96 per cent. Using this figure in the life table would increase the second year mortality rate to 27.6 per cent, and if w^e employ the figure of 1.36 per cent, which is the percentage of second year re- coveries in 1940, the mortality rate rises to nearly 35.8 per cent, which is almost exactly that of the Kent Island population. Con- comitant with these increases in the second year mortality rate are decreases in that of the first year, so that in the early inter- vals the Danish life table becomes quite similar to that of Kent Island. These calculations should suffice to illustrate that the Danish life table is based on small and highly variable samples, and that even seemingly minor changes in the data may have large-scale effects. This should also warn against assuming that Paludan 's life table is more nearly correct than others merely because it "balances" when certain data are borrowed from other popu- lations. The second life table for European gulls is that prepared by Olsson (1958), for 1,222 recoveries, including those shot (60 per cent), from about 12,700 birds banded in Sweden, Norway, and Finland (=Fenno-Scandia) from the mid-1920 's to the mid- 1950 's (Table VII). The average rate of recovery was about 9.5 per cent, which is nearly three times that at Kent Island and about ten per cent higher than that in Denmark. The table starts on August 1 following hatching. The oldest recovery is a gull in its 17th year; the oldest potential recovery at the time the study ended would be an individual in its 29th year. The raw 512 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY data have been adjusted to compensate for the fact that some year classes had not been banded long enough to yield all their potential recoveries. It may be seen (Table IV) that the first year mortality of 56.8 per cent is somewhat lower than that found in Hickey's North American and Paludan's Danish populations, and that the second year rate is slightly higher than that of the North American gulls, and considerably higher than that of the Danish birds, but still not so great as that of the Kent Island population. For the next ten years the mean annual mortality is 33.8 per cent, which is more than twice the rate recorded by Paludan and moderately higher than that of the American and Kent Island populations. Olsson had no records of the fledgling production in the Fenno- Scandian colonies and did not, therefore, attempt to test the accu- racy of his life table against such data. Using the method devised by Hickey, it is found that 2.86 fledglings must be raised to TABLE VII Life Table for Fenno-Scandian Gulls Bauded 1925-1935 (1,122 Kecoveries) (Olsson, 1958) X I d 1,000 q e X X X X 0-1 1,000 568 568 1.93 1-2 432 138 319 2.80 2-3 294 81 276 2.88 3-4 213 66 310 2.79 4-5 147 45 306 2.81 5-6 102 24 235 2.83 6-7 78 26 333 2.55 7-8 52 15 289 2.58 8-9 37 11 297 2.42 9-10 26 6 231 2.23 10-11 20 n 550 1.75 11-12 9 5 556 2.28 12-13 4 0 0 3.50 13-14 4 0 0 2.50 14-15 4 2 500 1.50 15-16 2 0 0 1.50 16-17 2 2 1,000 0.50 HERRING GULL LIFE TABLES 513 August 1 by each pair of breeding gulls if the population is to remain static. Even with a mean clutch of three eggs, such a high rate of nesting success appears improbable and it must be con- cluded that the Fenno-Scandian life table is also inaccurate. POSSIBLE ERRORS IN THE KENT ISLAND TABLES From the fact that the Kent Island colony was observed to have maintained its size from at least 1935 to 1948 (see p. 501) it seems reasonable to conclude that the population was either rela- tively stable during the 1930 's and 1940 's or, even more likely, that it was increasing and its surplus overflowed the colony and added to the expanding North American population. The life tables, however, indicate that the population should have been decreasing at a catastrophic rate. If we assume that the life tables are deficient, we must look for a source, or sources, of error within the data used in constructing the tables. There are three kinds of error which might distort the life tables, causing them to indicate that the population was rapidly declining. First, banding recoveries may have been accumulated for too short a period (23 through 28 years), thereby failing to cover the full life spans of the six cohorts studied. Second, the 1947 egg and prefledging mortality rates, which were used in con- structing the initial intervals of all six tables, may have been uncharacteristically high. Third, bands may have been recovered more readily in the early postfledging years than later in the life spans, causing an apparent increase in the mortality rates for the early intervals of the tables. The first hypothesis, i.e. that a study allowing for recoveries for a maximum period of between 23 and 28 years does not cover the full life span of the Herring Gull, is correct, but probably only to a small and insignificant degree. For example, a Kent Island gull has been found alive in its twenty-sixth year, and a bird banded as an adult was recovered 24 years later meaning it was a minimum of twenty-seven years old when it died. There are also records of captive Herring Gulls which have lived nearly fifty years (Gross, 1940) ; it is conceivable that some wild individuals may attain an equally advanced age. Nevertheless, it seems that very few gulls live beyond their twenty-third year, which is the oldest potential recovery within the youngest cohort studied at Kent Island, and those birds that do live longer doubt- less have little influence on the reproductive rate of the popula- tion because of their relatively insignificant numbers and possibly (unproven) because of sterility brought about by senility. 514 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Support for the argument that the life tables adequately cover the life spans of the gulls may be found in the recovery data. Of the 954 recoveries from the five oldest cohorts, with a maxi- mum potential age ranging from 24 to 28 years, only one indi- vidual (0.16 per cent) reached the age of 24, none exceeded that, and but three lived more than 18 years. Expressed differently, less than one-fifth of one per cent of the recoveries were of birds older than 18 years. In Paludan's Danish study (1951) there were eight cohorts with a maximum potential age between 26 and 34 years. Of the 236 recoveries from these cohorts there was just one bird in its twenty-sixth year and the recovery rate between the eighteenth and twenty-sixth year was only 0.8 per cent of the total sample. The maximum potential age of birds in Olsson's (1958) Fenno-Scandian population was twenty-nine years, but the oldest recovery was a single seventeenth-year individual. There is, of course, the possibility that a loss of bands could account for the failure to recover gulls even older than those now known. But it is believed that while the frequency of re- coveries may be reduced because of band loss, there is little likeli- hood that the maximum span of life is significantly greater than that which has been recorded. We must, therefore, seek another explanation for the failure of the life tables to document the population dynamics of the Kent Island colony. The second hypothetical source of error in the life tables, i.e. that the i^refledging mortality in 1947 was unusually high, is al- most certainly void. It was found that nearly 29 per cent of the eggs failed to hatch and about 48.5 per cent of the young died before fledging, resulting in 63.2 per cent mortality, or a net production of 0.92 fledglings per breeding pair (Paynter, 1949). This falls well within the range recorded at other Herring Gull colonies. For example, Paludan (1951) who admitted that his calculations are imprecise, estimated a production of 0.5 fledged young per pair, or 83 per cent prefledging mortality. In a study on the Summer Isles in the Irish Sea, Darling (1938) found five colonies (ranging from 6 to 150 individuals) which had pre- fledging mortality rates between 58.3 and 88.9 per cent, and which yielded from 0.78 to 0.96 fledged young per nest. Drost, Focke, and Freytag's (1961) German colony, which started with two pairs of birds and in 12 years built up to 139 pairs, had an average prefledging mortality rate of about 75 per cent, resulting in the fledging of an average of 0.7 gulls from each nest. On Skokholm, an island off the coast of Wales with a colony of 300 pairs of gulls, Lockley (1947) estimated that less than one HERRING GULL LIFE TABLES 515 fledgling per adult pair was produced. In another Welsh colony of 440 pairs, Harris (1964) reported a production of about one fledgling for each two nests. Other species of gulls laying three-egg clutches show a similar range of nesting failures. A colony of Ring-billed Gulls (L. delaivarensis) in Michigan suffered 88 per cent prefledging losses, producing 0.67 young per pair of adults (Emlen, 1956). The California Gull {L. calif ornicus) ^ in a colony in Utah, was found to have unusually low prefledging mortality, suffering a loss of only about 40 per cent of its eggs and young, and fledging 1.77 chicks per nest (Behle and Goates, 1957). In British Columbia a colony of Glaucous-winged Gulls {L. glaucescens) , studied for two years, produced 1.0 and 1.7 young per nest, and had a mor- tality rate of 64 and 52 per cent, respectively (Vermeer, 1963). Four small colonies of Lesser Black-backed Gulls (L. fuscus) in the Summer Isles produced about 1.5 young from each nest, with a prefledging mortality rate of approximately 48 per cent (Dar- ling, 1938). In summary, the prefledging mortality rate at Kent Island in 1947 was below that which has been found in most other Herring Gull colonies and in a colony of L. delaivarensis, and somewhat higher than that recorded for populations of L. californicus, L. glaucescens, and L. fuscus. From this it is concluded that the 1947 prefledging survival data are in the right order of magni- tude and probably are not responsible for the failure of the composite life tables to indicate that the population was stable or increasing. The third type of error that could distort the Kent Island life tables, causing the survival rates for breeding birds to appear too low to maintain the population, is a disproportionately high re- covery of bands in early age classes. This could be brought about through some circumstance which allows young birds to be more readily recovered than older individuals, through a loss of bands among older birds which would reduce their rate of recovery, or through a combination of these factors. Both phenomena are difficult to detect but almost certainly at least one is the reason that the life tables are not reconcilable with the observed status of the population. Paludan (1951, p. Ill) believed that the mortality rate for the first year class of the Danish gulls was unrepresentatively high because newly fledged birds died near the natal colony, where they are more likely to be recovered and reported than older birds which range more widely. At Kent Island, however, the 516 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY fledglings quickly leave the colony and winter far south of New Brunswick. Each successive year the length of the migration is lessened until as adults the birds winter only a few hundred miles from the colony (Gross, 1940). This migration pattern re- duces the opportunity for any recoveries at Kent Island during the winter months. Moreover, the records indicate that very few immature gulls are recovered at the colony during any season. There seems no possibility that the recovery sample is biased in favor of immature birds because they die in the vicinity of the natal colony. There remains to be considered the possibility that the wide dispersal of young birds during the winter, and their failure to return to the colony during the breeding season, might in some way enhance their chances of recovery over those of the less wide- ranging adults. If the immature gulls were in closer proximity than the adults to urban areas, w^here dead birds are more likely to be found, this possibility would exist. However, during the winter the younger birds are scattered from the heavily popu- lated northeastern United States south to the more sparsely in- habited (tulf Coast and Central America, and in the breeding season return to the Northeast. On the other hand, the adults winter in the heavil^y populated Northeast and breed where ob- servers are constantly alert for dead birds wearing bands. Bias in favor of the recovery of immature gulls seems improbable under these circumstances. Having considered and rejected as improbable all other poten- tial sources of error, we are left with only band loss to account for the discrepancies between the life tables and the observed status of the Kent Island colony. All Kent Island gulls were marked with aluminum butt-end l)ands, rather than with the locking, clip-type bands now used in Europe. In a British study, using butt-end and clip bands on dilferent samples plus supplementary durable plastic bands, Poulding (1954) found that during the first year after banding as fledglings, Herring Gulls with locking bands were recovered with twice the frequency of birds wearing butt-end bands. Breaking down the analysis further, it was noted that when the gulls wore butt-end bands only about four per cent of the year's total recoveries were made in the second half of the year, whereas nearly 41 per cent of the recoveries occurred in the second six months when the birds wore clip-type bands. Thus the data indi- cate that 50 per cent of the butt-end bauds are lost in the first year and nearly all of this loss takes place before the bands are HERRING GULL LIFE TABLES 517 six months old. Observations of living birds confirmed the 50 per cent loss of butt-end bands and showed a complete retention of clip bands. This evidence strongly suggests that the heavy initial loss of butt-end bands may be caused by the removal of the bands by the gulls, rather than by a weakening of the bands through wear. Differences in the strength of individual bands, and possibly variations in the manner of closing them, could well account for the rapid loss of half of the bands while some of those in the remaining half are retained in good condition for many years. If this type of band loss occurs among Kent Island birds, the life tables would be significantly affected. It would mean, dis- regarding for the moment the normal reduction in the size of cohort caused by death, that at the time of fledging the banded sample would be twice the size it is when entering the second year of life. It follows that the number of recoveries in the second year, and all subsequent years, would be half of the total had there been no loss of bands and that the recoveries in the first year would be somewhat lower, but not a full 50 per cent lower because the bands are retained for part of the first half year. Viewed from a different aspect, recoveries in the first year would be disproportionately more numerous than those of the remaining year classes. The effect on the life table Avould be a reduction in the calculated number of birds surviving beyond the first year and, of course, a marked decrease in the apparent num- ber of individuals which survive to the breeding age. We have no evidence that Kent Island gulls lose half their bands during the first six months, but if we adjust the life table to compensate for such a loss and the table is then reconcilable with the apparent true status of the population, we shall have good circumstantial evidence that this is the source of error. If, for the sake of simplicity, we assume that the mortality rate is constant during the entire first year and that 50 per cent of the bands are lost at the end of the first six months and none in the second half year, one-third of the recoveries would occur in the second six-month period. Doubling this figure will give the num- ber of recoveries w^hich would have been made if the initial banded sample had been half its original size and there was no subsequent loss of bands. For the six combined cohorts there were 1,09!) recoveries, 494 of which occurred during the first year. Making the proposed adjustments for the first year, the first year recoveries would drop to 329 and the total for the en- tire life span to 934. 518 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Using these data to construct a new life table, it is found that the number of survivors reaching the breeding age of year 3-4 rises to about 116, versus 98.4 in the uncorrected table. The life expectancy at this age remains 3.3 years. AVith a breeding cohort of this size and with this life expectancj^ a total of 477 eggs could be produced, which is an increase of 71 eggs over the total (406) calculated for the uncorrected life table. Nevertheless, this is still less than half the number required to maintain a level population. If we were to assume that early band loss is the sole source of error in the life table, in order to achieve sufficient production at the breeding stage it would be necessary to adjust the recoveries in the first year so that their total would fall well below that of the second year. Such a low rate of mortality is obviously spu- rious. We must conclude that while early band loss is a distinct possibility in the Kent Island population, there must be addi- tional losses later in the life span. Band loss maj^ fall into three broad patterns. The first, and the most expected, is loss which is correlated with the age of the bands. One would expect bands to become progressively weaker through wear and as a result be lost with increasing frequency. The second type is a proportional, or constant, loss. This is most likeh" to take place if the bands are continually removed by the birds or if they merely drop off at random. The third pattern, which may be uncommon, is an inconsistent, or fluctuating, loss. This may occur because of variations in behavior, such as shifts in feeding ranges from less saline to more saline water, which might have a variable effect on the durability of the bands, or because the new bands are not of uniform strength or are not all fastened securely. The large initial loss of bands noted by Poulding (1954) is an example of an inconsistent pattern which seems attributable to the latter cause. If we are able to determine the pattern of band loss among Kent Island gulls it may then be possible, in certain instances, to adjust the raw data to compensate for the losses and to construct a life table which will document the colony's demography. Be- cause the Kent Island birds were all marked with similar bands there is no control group against which the various cohorts may be tested, as in Poulding 's study. We must, therefore, see if any evidence of band loss may be detected within the data available. Semi-logarithmic graphs of survivors (Ix) offer a means of ap- proaching the problem. HERRING GULL LIFE TABLES 519 If survivors (Is) are plotted against age on a semi-logarithmic graph, the points will form a straight line if the rate of survival is constant. If the survival rate should decrease, or if there is an accelerating loss of bands which would create an apparent de- crease in the rate of survival, the line will assume a sigmoid shape, being deflected downward. As may be seen in the plot for the six combined cohorts at Kent Island (Fig. 2), starting with the fourth year (year 3-4) of life, the survival rate is relatively constant for at least ten years (to year 12-13). There is then a downward trend, but there are so few recoveries one cannot be certain that the pattern is not an artifact. From this it is evident that if there is band loss that is correlated with age, it does not begin until sometime after age thirteen, by which time the cohort has dwindled to a small fraction of its initial size. A further test for a correlation between a loss of bands and the time which they have been worn may be made by utilizing the recoveries of birds banded as adults. Between 1936 and 1947 a total of 1,856 gulls in adult plumage were banded; 97, or 5.2 per cent, were recovered up to 30 June 1963 (Table VIII). TABLE VIII Kent Islaud Adults Banded and Admissible ^ Eecoveries Year 1936 1937 1938 1939 1941 1946 1947 Total 1,856 97 5.23 1 Set' 1). 4!);j for (li>liiiit:(iii of ailniissible recoveries. Nothing is known of the ages of these birds beyond the fact that they were in adult plumage and, therefore, were at least in their fourth year of life when banded. The maximum time available for recoveries from the youngest cohort is 16 years and from the oldest cohort it is 28 years. The oldest recovery was a bird banded 24 years earlier ; the next oldest were two gulls which had mded Recoveries Per ce 200 2 1.00 196 13 6.63 611 46 7.53 100 5 5.00 155 5 3.23 497 21 4.22 97 5 5.15 520 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY borne their bands for 14 and 17 years, respectively. A semi- logarithmic survivorship curve for the 97 recoveries from the combined seven cohorts is shown in Figure 3. Owing to the paucity of older recoveries, no adjustment has been made for the fact that after age 16 the number of potential recoveries de- creases because not all cohorts have been banded sufficiently long to yield recoveries between ages 17 and 24. In Figure 3 a comparison is made between the semi-logarithmic survivorship curve for the birds banded as adults and the curve, starting at the fourth (adult) year, for those gulls banded as fledglings. If there is a positive correlation between the age of the bands and their loss, one would expect the survival rate for the latter group to decrease, and the curve to become deflected three years sooner than for the former group, because the birds banded as fledglings had borne their bands three years longer. If band loss begins during the first year following banding, the curves w^ould diverge immediately. If it starts some years later the two lines would remain parallel until band loss begins, when the curve for the group with older bands would descend more rapidly ; the plot for the group with the newer bands would con- tinue in a straight line (a constant annual survival rate) for three additional years and then it too would begin to fall away. As may be seen in Figure 3, the two curves remain nearly parallel for ten years, or in other words, until the group banded as fledglings has reached its thirteenth year. Then, as has been discussed, the curve for the group with the older bands begins to decline at a more rapid rate. This is not follow^ed three years later by a decline in the other curve, which one would expect if bands began to be lost at an increasing rate after they had been worn thirteen years. In both groups, and particularly in the group banded as adults, there are so few recoveries in the older age categories it would be imprudent to attempt at this time to read any significance into these differences. All we may safely conclude is that if band loss is a function of age it almost cer- tainly does not begin until the bands liaA'e been worn at least thirteen years. Because this means that the birds could have bred for ten seasons prior to the beginning of accelerated band loss, there seems little likelihood that this type of band loss would have a significant effect on the life tables. This test also reveals an interesting fact regarding early band loss. It indicates that gulls banded as adults do not suffer a greater loss of bands in the first year than in later years, which is contrary to what one would have expected from Poulding's HERRING GULL LIFE TABLES 521 study (1954) of British fledglings. This is probably an indica- tion that the North American butt-end bands are more uniformly durable than the British butt-end bands. This also seems to ex- plain why the attempt to compensate for disproportionately 1,000 — 100 _ < Z3 O l/J o > > 3 I/) Banded as adults 10 _ Banded as fledglings l_ [ I I I I ' I ■ I ■ I ■ 1 ■ I M 6 8 10 12 14 16 18 20 22 ' I ■ I Adults 0 2 4 Fledglings 3.1 5.1 7.1 9.1 II. I 13.1 15.1 17.1 19.1 21.1 23.1 25.1 AGE IN YEARS Fig. 3. Survivorship curves for Kent Island gulls banded as fledglings and as adults. 522 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY heavy band loss within the first year class of Kent Island gulls, banded as fledglings, failed to produce a satisfactory life table. Nevertheless, one could argue that adults do not remove their bands with as great a frequency as fledglings, explaining this difference by assuming that the legs of fledglings are more sensi- tive than those of adults and that the irritating bands are re- moved. Or, possibly, one could reason that in the process of learning to detect what is food, and what is not, the fledglings are attracted by the shiny bands and peck at them until they are loosened and lost. However, the fact that the Kent Island life table is not appreciably improved by adjustments made to com- pensate for heavier band loss in the first year of life seems to negate such arguments. The simple explanation that British butt-end bands are of more variable durability than those used at Kent Island seems the most satisfactory explanation. The only pattern of band loss yet to be considered is that which occurs at a constant, i.e. proportional, rate. Having concluded that band loss is not positively correlated with the age of the band, and that bands are not lost with a greater frequency at any particular time during the bird's life, it is almost certain that the Kent Island population suffers a steady loss of bands during its entire life span. This would progressively reduce the number of recoveries in each successive age interval, thereby depressing the survivorship curve and reducing the calculated expectation of life throughout the life table. We do not know the rate at which bands are lost and whether this rate is sufficiently low to allow at least a portion of the population to retain its bands until it has lived its full life span. However, because a living gull was found at Kent Island which had borne a band for 26 years, and because there are records from there of four birds Avhich died between the ages of 20 and 24, it is evident that the rate of band loss must be comparatively low. For example, if the annual loss ran as high as 20 per cent, the roughly 32,000 individuals in the Kent Island sample would have dwindled to about 455 banded birds by the beginning of the twentieth year, even without considering attrition owing to mortality. If the average annual mortality were 10 per cent, making a cumulative annual reduction in the marked population of 30 per cent, there would have been just 37 banded survivors at the start of the twentieth year, certainly too few to have yielded five records of birds 20 years and older. Assuming that losses and mortality are each 10 per cent annually, there would be approximately 445 banded survivors after nineteen years, or HERRING GULL LIFE TABLES 523 about 1.4 per cent of the initial sample, which seems ample to allow for the recovery of five individuals in the next seven years. Althoug-h the Herring Gull is a long-lived species, 10 per cent average annual mortality would appear rather low; Drost et al. (1961) calculated a rate of 10 per cent for adult birds, but our calculations must include immatures within the average. If we continue to allow for 1.4 per cent survival at age 20, any increase in the mortality rate would have to be balanced by a decrease in the rate of band loss. Thus, 15 per cent annual mortality, which is about the figure found by Paludan for birds between the ages of 2 and 12, would mean a band loss of five per cent at the most. Crude as these calculations may be, it seems reasonable to con- clude that Kent Island gulls lose bands at an average rate some- where in the vicinity of 5 or 10 per cent per year and that the average annual mortality rate must range betw^een 10 and 15 per cent. These speculations are based on the premise that bands are recovered throughout the life span of the birds and are not totally lost before the oldest gulls die. There can be no proof that this is correct, but empirically it would seem that the oldest recoveries probably very nearly represent the potential natural longevity of the Kent Island Herring Gull. A larger cohort, more durable bands, and additional decades of observation surely would produce recoveries older than those now known, but it is difficult to imagine that these could be more than an insignificant fraction of the total sample. CONCLUSIONS It is now evident that a continued loss of bands accounts for the failure of the Kent Island life tables to reconcile with the observed status of the population (i.e. either a stable or increas- ing population). Unfortunately, in spite of our estimate that the loss amounts to about 5 or 10 per cent annually, there is no way by which the raw data may be adjusted to compensate for these losses. We know, for example, that there were 494 re- coveries for the combined cohorts in the first nine-tenths of a year, and that without a ten per cent loss of the banded sample there would have been about 549 recoveries. In the second year there were 220 recoveries, but this number, without band losses during the second year, would have been approximately 243, plus an unknown number of individuals which lost bands in the first vear but survived to die during the second year. The difference 524 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY between the recorded recoveries and the number of recoveries there would have been had there been no band loss, increases, of course, in each successive age interval. This also has the effect of accelerating the descent of the survivorship curve and probably accounts for the difference between the Kent Island curve and that of the Danish population (Fig. 2). From this analysis we must conclude that because of band losses the Kent Island banding records are nearly valueless as a means of investigating the dynamics of this population of Herring Gull. Their only use in studies of this sort, if one is willing to accept the premise that some bands are sufficiently durable to be retained through the life span of at least a few of the longest living gulls, is to demarcate the maximum potential longevity of the species. All North American Herring Gulls have been marked with similar butt-end bands, and band loss is with- out doubt the reason Hickey (1952) also failed in his attempt to construct an accurate life table. Because aluminum butt-end bands are used on almost all birds banded in North America, any data obtained from this source is suspect. Future workers should be particularly alert to the possibility of band loss before in- vesting time and effort in this type of research. Locking bands were used on nearly all gulls banded in Fenno- Scandia (Olsson, 1958) . However, the material used in the manu- facture of the bands evidently was not very durable, for Olsson estimated that about 5 per cent of their original weight was lost annually. Such rapid wear, and the resulting loss of bands, is presumably the reason for the apparent reduced maximum lon- gevity of the Fenno-Scandian gulls when compared with North American and Danish birds. There can be little doubt that the similarity between the survivorship curves for Fenno-Scandian and North American birds is caused by band losses. Paludan (1951) does not mention the type of band used on Danish gulls, but, according to Poulding (1954), butt-end bands were first used and later abandoned in favor of locking liands. Possibly some of the yearly variations in recovery rates that have been noted (p. 510) are attributable to changes in the type of bands employed. Nevertheless, this cannot be one of the main causes for these variations because one would expect the recovery rates for the more recent year classes to be consistently higher than those for the older year classes, but no such pattern is evident. Band loss almost certainly occurred in the Danish population. Even if locking bands eliminate losses earh* in the life span, HERRING GULL LIFE TABLES 525 wear surely accounts for band failures in the later years of this long-lived gull. Bias in favor of recoveries during the first year of life, and band losses during the latter part of the life span, must be the primary reasons for the deficient Danish life table and its dissimilarity to other life tables. Fundamental differences between the mortality rates of the North American and Fenno- Scandian populations, on one hand, and the Danish population, on the other hand, as proposed by Paludan (1951) and Olsson (1958), seem most unlikely. ACKNOWLEDGMENTS I am grateful to Charles E. Huntington for making available the banding records of the Bowdoin Scientific Station, and to Allen J. Duvall for his assistance in obtaining microfilms of the Fish and Wildlife Service files. Da\4d W. Norton was of great assistance in collating these records. I thank Barry Margolin for spending much time pondering the statistical aspects of the study, my wife, Elizabeth S. Paynter, for preparing the figures, and Charles H. Blake for his helpful reading of the manuscript. SUMMARY 1. At Kent Island, New Brunswick, 31,694 Herring Gull {Lams argentatus smithsonianiis) fledglings were banded with aluminum butt-end bands between 1934 and 1939; by 30 June 1963 there were 1,099 (3.47 per cent) recoveries suitable for use in a mortality series (Table I) . 2. The maximum potential age for a recovery in the oldest co- hort is 28 years, and in the youngest cohort it is 23 years; the two oldest birds at death were in their twenty-fourth and twenty- second years; one gull w^as captured alive in its twenty-sixth year. 3. About five per cent of all recoveries are of birds which have been shot ; rates vary within the six year classes from 3.6 to 6.5 per cent; probably more birds are shot than are reported because the species is protected by law ; no gull older than thir- teen has been reported shot ; there is no indication that immature birds are more readily shot than adults. 4. Using prefledging mortality data for 1947 (Paynter, 1949), composite life tables are constructed for each of the six cohorts and for the combined cohorts (Table II) ; survivorship curves 526 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY are also drawn (Figs. 1 and 2) ; there are minor yearly varia- tions, but the patterns are generally similar, showing heavy mor- tality the first year, lessened mortality the second year, and a lower, relatively constant, rate thereafter. 5. Assuming breeding begins in the fourth year (year 3-4), that the average clutch is 2.5 eggs, and that each nesting results in the production of 0.92 fledglings (Paynter, 1949), it is found that an average of about 98.4 gulls attain breeding age from each 1,000 eggs laid, and these have a life expectancy of 3.3 years, enabling them to produce 405.9 eggs, or about 41 per cent of the number required to maintain a stable population. 6. The population is believed to liave been stable, or possibly expanding, from at least 1935 to 1948. The life tables must, therefore, be faulty. 7. Three types of error leading to distorted life tables are possible: 1. recoveries were accumulated for too short a period (23 to 28 years) to document the potential life span; this is probably a minor and insignificant source of error. 2. The 1947 egg and prefledging mortality rates were excessively high ; this is rejected because comparable rates have been found in this and related species. 3. There is a disproportionately high re- covery of bands in the early age classes because (a) young birds are more readily recovered, (b) older birds lose bands and are lost from the sample, or (c) both factors are operative. 8. It is certain that band loss, resulting in a disproportionately high recovery of young birds, must account for the failure of the life tables to document the demography of the Kent Island population. 9. Band loss could be (a) positively correlated with the age of the band, (b) proportional (i.e. constant) with respect to the size of the sample, or (c) fluctuating; it is concluded that bands are lost at a relatively constant rate throughout the life span. 10. Assuming that some bands are retained long enough to document the maximum potential life span, mortality and band loss combined seem not to exceed 20 per cent annually ; it is sug- gested that band loss averages around 5 or 10 per cent per year and that the average annual mortality ranges between 10 and 15 per cent. 11. Hickey's (1952) life table (Table III; Fig. 2) for North American gulls proved faulty presumably because of band losses similar to those at Kent Island. 12. Olsson's (1958) life table (Table VI; Fig. 2) for Fenno- Scandian gulls is similar in pattern to those for North America HERRING GULL LIFE TABLES 527 and Kent Island; locking bands probably eliminated losses throng'h mechanical failnre, but rapid wear (five per cent an- nually) doubtless resulted in band loss correlated with age and eventually to a total loss of marked birds by the eighteenth year, foreshortening the apparent maximum life span. 13. Paludan's (1951) life table (Table V; Fig. 2) for Danish gulls, with its higher survival rate, most nearly fits what is known of the demography of the species but bias in favor of re- coveries during the first year, wide variations (of unknown cause) in yearly recovery rates, and probably band losses to- ward the end of the life span distort the data. 14. Suggestions of essential differences between the mortality rates of the various populations are considered spurious. LITEEATUEE CITED Behlb, W. H., and W. a. Goates 1957. Breeding biology of the California Gull. Condor 59:235-246. Crystal, F. H. 1941. Herring Gull census. Ann. Eep. Bowdoin Sei. Sta. 5:37-38 [mimeo.]. Darling, F. F. 1938. Bird flocks and the breeding cycle. Cambridge U. Press, Lon- don. X + 124 p. Drost, E., E. Focke and G. Freytag. 1961. Entwicklung und Aufbau einer Population der Silbermowe, Larus argentatus argentatus. J. Ornithologie 102:404-429. Emlen, J. T., Jr. 1956. Juvenile mortality in a Eing-billed Gull colony. "Wilson Bull. 68:232-238. Earner, D. S. 1955. Birdbanding in the study of population dynamics. In A. Wolf- son [ed.], Eecent studies in avian biology. U. Illinois Press, Urbana. Pp. 397-449. Goethe, F. 1937. Beobachtungen und Untersuchungen zur Biologie der Silber- mowe {Larus a. argentatus Pontopp.), auf der Vogelinsel Mem- merstand. J. Ornithologie 86:1-119. Gross, A. O. 1940. The migration of Kent Island Herring Gulls. Bird-Banding 11:129-155. Harris, M. P. 1964. Aspects of the breeding biology of the gulls Larus argentatus, L. fuscus and L. marinus. Ibis 106:432-456. 528 BULLETIN : MUSEUM OF COMPARATI\^ ZOOLOGY HiCKEY, J. J. 1952. Survival studies of banded birds. U.S. Dept. Interior, Fish and Wildlife Ser., Spec. Sci. Eep., Wildlife No. 15. 177 p. Lack, D. 1954. The natural regulation of animal numbers. Oxford XJ. Press, London, viii + 343 p. LOCKLEY, E. M. 1947. Letters from Skokholm. J. M. Dent and Sons Ltd., London. X + 243 p. Marshall, H. 1947. Longevity of the American Herring Gull. Auk 64:188-198. Olsson, V. 1958. Dispersal, migration, longevity and death causes of Strix aluco, Buteo buteo, Ardea cinerea and Larus argentatus. A study based on recoveries of birds ringed in Fenno-Scandia. Acta Vertebratica 1:91-189. Paludan, K. 1951. Contributions to the breeding biology of Larus argentatus and Larus fuscus. Vidensk. Medd. fra Dansk naturh. Foren. 114:1- 128. Paynter, E. a., Jr. 1947. The fate of banded Kent Island Herring Gulls. Bird-Banding 18:156-170. 1949. Clutch-size and the egg and chick mortality of Kent Island Herring Gulls. Ecologj' 30:146-166. POULDING, E. H. 1954. Loss of rings by marked Herring Gulls. Bird Study 1 : 37-40. Vermeer, K. 1963. The breeding ecology of the Glaucous-winged Gull {Larus glau- cescen^) on Mandarte Island, B. C. Brit. Columbia Prov. Mus. Occas. Papers 13, 104 pp. (Eeceived September 2, 1965.) Harvard MCZ LIbrai 3 2044 066 303 983 Date Due SEP 3 0 1984