OCCASIONAL PAPERS OF THE California Academy of Sciences No. 95, 31 pages, 12 figures. TRUNK LATERAL LINE NERVES, HYOID ARCH GILL RAKERS, AND OLFACTORY BULB LOCATION IN ATHERINIFORM, MUGILID, AND PERCOID FISHES1 By Warren C. Freihofer Marine i-i- logical laboratory I LIBRA' Y * AUG 2 9 1972 Wood* Hoio, wioa». SAN FRANCISCO PUBLISHED BY THE ACADEMY July 27, 1972 OCCASIONAL PAPERS OF THE California Academy of Sciences No. 95, 31 pages, 12 figures. TRUNK LATERAL LINE NERVES, HYOID ARCH GILL RAKERS, AND OLFACTORY BULB LOCATION IN ATHERINIFORM, MUGILID, AND PERCOID FISHES1 By Warren C. Freihofer California Academy of Sciences INTRODUCTION The present paper reports on the patterns of trunk lateral line nerves in atheriniform and mugilid fishes and compares the patterns in these fishes with the patterns in percopsiform and percoid fishes and in various lower-placed fishes. The forward position of the olfactory bulb and the presence of gill rakers on the hyoid arch in the mugilid Agonostomus monticola are also described. The significance of the new facts is discussed for mugilid, percoid, and atheriniform classification and relationships. Lateral line nerves on the trunk have been little util- ized in the classification of fishes.2 This is ironically ■'-Research for this paper was supported by National Science Foundation Grant GB 30551. 2The trunk region of fishes is defined here as that part of the body which lies between the hind end of the cranium and the last caudal vertebra but exclusive of the fins. The term body includes the head, trunk, and fins. CALIFORNIA ACADEMY OF SCIENCES [Occ. Papers true for atheriniform fishes, many of which have no lateral line tubed scale row or almost none; yet all of which have a complex pattern of lateral line trunk nerves. This fact happens to be of considerable systematic significance. The primary feature of these patterns is the presence of a greater or lesser number of ventral segmental lateral line branches of the main horizontal septum nerve. Until recently (Freihofer, 1970) the ventral segmental trunk lateralis branches in fishes had been reported on only once (Handrick, 1901) . Numerous kinds of fishes other than atheriniforms have rather elaborate patterns of these nerves or some modi- fication of them. The nerves were studied on specimens prepared by the Sihler technique of staining nerves in a cleared whole speci- men (Freihofer, 1966; Fraser & Freihofer, 1971). No serially sectioned preparations were made. Agonostomus monticola was examined by dissection of alcoholic specimens. ACKNOWLEDGMENTS I would like to thank H. Adair Fehlmann, Allen Freihofer, Robert Hassur, Ernest Iverson, Melvin Moriguchi, Koji Murata, Jamie Thomerson, and J. P. Wourms for providing freshly pre- served unneutralized formalin specimens necessary for the Sihler technique or for helping in field work. Thanks are due also to Leo Shapovalov of the California Department of Fish and Game and J. M. Hyde of the Fish Diseases Laboratory at Rancho Cordova, California, for supplying formalin speci- mens of trout and salmon. Leonard Compagno helped in photo- graphing specimens for making drawings. Maurice Giles made the photographs for figures 8, 9, and 11. Lillian Dempster and William N. Eschmeyer gave suggestions on the writing of the text which are much appreciated. EXPLANATION OF TERMS The different branches named in the descriptions of the trunk lateral line nerve patterns are briefly defined as follows. The horizontal septum nerve, which is the main trunk lateral line branch of the vagus nerve, courses along the horizontal septum for the length of the trunk. Dorsal and ventral segmental rami are branches of the horizontal septum nerve. They pass dorsally or ventrally in the segment of the trunk in which they arise. They do not leave that segment. The ventral segmental rami extend at least halfway towards the midventral line of the body, usually almost to it. The dorsal rami usually reach less than halfway to the middorsal line. The lateral rami, one in each body segment, are short branches of the horizontal septum lateral line nerve. They innervate lateralis organs in the tubed lateral line scales or naked lateralis organs in the skin along the horizontal septum. If the lateral line scale row lies dorsal to the horizontal septum, it is these lateral rami that No. 95] FREIHOFER: ATHERINIFORM NERVE PATTERNS extend dorsally to innervate the lateral line. For a dor sal- ly placed lateral line, the rami innervating it usually turn posteriorly upon reaching the lateral line scale row and course for a few segments thus overlapping the next few seg- mental rami. The dorsally placed longitudinal nerve formed by the dorsal longitudinal rami supplying the lateral line scale row is called, in this report, a dorsal longitudinal collector nerve. The dorsal longitudinal collector nerve is a characteristic of acanthopterygian fishes. It is not present in fishes placed lower than paracanthopterygians in the Greenwood et al. classification. Some higher paracanthop- terygians approach the collector nerve condition. The collector-like nerve seems to be a significant feature as far as is presently known. To be distinguished from the dorsal longitudinal collector nerve is the dorsal longitudinal ramus, of which two may be present. Each originates from the hori- zontal septum nerve in the area of the supracleithrum, passes dorsally some distance before turning posteriorly, and crosses numerous body segments. The dorsal longitudinal rami con- trast with the dorsal longitudinal collector nerve in that each dorsal longitudinal ramus is composed of only one ramus. In some patterns of trunk lateral line nerves the dorsal, lateral, and ventral rami of each segment detach from the horizontal septum nerve as a common ramus which upon reaching the skin divides into the three segmental rami. This is true for Umbra , Percopsis , Aphredoderus , and some other forms examined . The following descriptions are preliminary. They are presented primarily to report on their apparent systematic significance. In future studies these nerves should be studied by serially sectioned and stained preparations to check the end organs of these nerves and to determine whether or not any branches contain other than lateralis fibers. Many other genera and families should also be studied. The systematic significance of the trunk patterns is discussed for the higher taxonomic categories, but not for the intra- familial and only partly for the interfamilial categories. TRUNK LATERAL LINE NERVES IN XENOMUGIL THOBURNI (Figure 1 . ) The description which follows is made from several speci- mens of the mugilid Xenomugil thoburni collected at the Gal- apagos Islands. The trunk lateral line pattern of branching in Xenomugil consists of five main parts: the midlateral trunk nerve lying along the horizontal septum; a small first dorsal longitudinal ramus and a large second dorsal longi- tudinal ramus; a large ventral branch to the pectoral-pelvic area; and a large midlateral ventral longitudinal branch. The two dorsal longitudinal rami come off as a common trunk from the main lateral line nerve at a point medial to Baudelot's ligament. The small first dorsal longitudinal ramus is the most anterior branch and apparently is equiva- lent to the dorsal longitudinal ramus of most fishes. It 4 CALIFORNIA ACADEMY OF SCIENCES [Occ. Papers crosses the medial side of the supracleithrum, arches dorsal- ly nearly to the middorsal line, and then courses posteriorly, ending in skin a little past the base of the pectoral fin. The large second dorsal longitudinal ramus emerges from be- hind the posterior edge of the supracleithrum and passes pos- teriorly at a level halfway between the horizontal septum and the dorsal fin. It terminates in skin a little beyond the end of the spinous first dorsal fin. At the point where the main trunk lateral line nerve passes medial to the cleithrum, the large pectoral-pelvic branch arises. The pectoral-pelvic branch crosses the medial surface of the supracleithrum and cleithrum and rises to skin along the posterior edge of the second postcleithrum below the pectoral fin. From here it follows along the postcleithrum towards the pelvic fin and terminates in skin a short distance beyond the base of this fin. En route a sizable branch is detached which innervates scales over the base of the pectoral fin and scales more an- teriorly towards the isthmus. En route to the pelvic fin other branches are given off, innervating scales ventral to the pec- toral fin and scales between it and the pelvic fin. Five branches arise on caudal peduncle. DISTRIBUTION OF PORED LATERALIS SCALES IN XENOMUGIL THOBURNI Almost every trunk scale of Xenomugil thoburni has a small horizontal tube and pore apparently for a lateralis or- gan. These pored lateralis scales are present even along the base of the spinous dorsal and anal fins and part-way out on the caudal fin base. The lateralis tubes on these scales are tiny compared to the large tubes on the few lateralis scales on the head. On a few scales in front of and below the pec- toral fin, either the entire tube or only its posterior half is vertical. There are fewer tubes in the midlateral area between the soft dorsal and anal fins. A few scales near the isthmus each have two tubes. TRUNK LATERAL LINE NERVES IN ATHERINOPS AFFINIS (Figure 2.) In the atherinid Atherinops af finis, the horizontal sep- tum lateral line nerve extends from the shoulder girdle to the base of the caudal fin where it divides, each half pass- ing obliquely and distally out across the caudal fin rays. Along its length the horizontal septum nerve gives off dorsal and ventral segmental rami. There is one dorsal segmental lateralis ramus for each of the first twelve segments. In the next 22 segments there are only 3 dorsal segmental rami. Each of the last 6 segments has a dorsal segmental ramus to give a total of 21 rami. The first longitudinal dorsal ramus arches upwards from behind the supracleithrum, reaches almost to the middorsal line, and extends posteriorly nearly to the dorsal fin. There are 23 to 26 ventral segmental rami in the approximately 4 5 trunk segments. Each ventral segmental No. 95] FREIHOFER: ATHERINIFORM NERVE PATTERNS ramus nearly reaches the midventral line. There is an en- larged ventral pectoral branch, which passes down medial to the pectoral fin. It comes from behind the pectoral fin base and innervates lateral line organs ventral to the pec- toral fin. In some specimens the first two ventral segmen- tal rami may unite distally and serve the pectoral area. In a segment lacking a ventral segmental ramus, one of the ven- tral segmental rami adjacent to that segment divides and sup- plies the segment. What there is of a tubed scaled lateral line in Atherinops is described in the next section. DISTRIBUTION OF TUBED LATERALIS SCALES IN ATHERINOPS AFFINIS (Figure 3 . ) Tubed lateralis scales occurred in the following main areas on one specimen of Atherinops af finis (number of tubed scales in parentheses) : below and anterior to the pectoral fin (9) ; along base of anal fin (6) ; between base of pectoral and pelvic fins (12) , with none near the horizontal septum; along base of pelvic fin (6) ; in area halfway up from mid- ventral line and between end of anal fin and caudal peduncle (6) ; in last seven scales in the row above the midlateral red muscle band near end of caudal peduncle (3) ; above the midlateral red muscle band and just above dorsal edge of horizontal pectoral fin (6); above the red muscle band and above the pelvic area (3) . Along the red muscle band itself there is one tubed scale near the cleithrum, one on a verti- cal above the anterior end of the pelvic fin, and in the last 21 scales before the base of the caudal fin there are 4 tubed scales and 5 scales with only the pore and no tube; on the base of caudal fin there are 2 tubed scales along the midline and one tubed scale above and one below this row of two scales. There are 2 tubed scales between the end of the pelvic fin and the front end of the anal fin. These two scales lie several rows above the row of tubed scales next to the pelvic and anal fins. In the area between the pelvic row and the red muscle band there are 3 scattered tubed scales All other areas of the trunk are devoid of tubed or pored scales. A related atherinid, Leuresthes tenuis, shows a similar distribution of tubed lateralis scales. TRUNK LATERAL LINE NERVES IN PERCOPSIS TRANSMONTANA (Figure 47} In Percopsis transmontana (= Columbia transmontana ) there are approximately 42 ventral segmental rami of the hori- zontal septum nerve, one to each trunk segment. Behind the supracleithrum two branches detach more or less together from the horizontal septum nerve. One is the dorsal longitudinal ramus, which almost reaches the origin of the dorsal fin. The other branch is the first dorsal collector ramus of the incipient longitudinal collector nerve. It is followed by 6 CALIFORNIA ACADEMY OF SCIENCES [Occ. Papers three or four other main collector rami for the collector nerve. Each collector ramus passes dorsally a short distance up to the tubed lateral line scale row, which extends the length of the trunk, and then turns posteriorly and continues for a number of segments. Each dorsal collector ramus over- laps the following collector ramus for several segments. The first branch of the first collector ramus is large and passes ventrally down the posterior edge of the pectoral girdle and innervates naked lateralis organs in skin ventral and anterior to the pectoral fin and organs in the skin between the pec- toral and pelvic fins. This large branch detaching from the first collector ramus is an enlarged first ventral segmental ramus. It may be joined by the second ventral segmental ramus in some specimens. In addition to the ventral segmen- tal rami, there are also dorsal segmental rami, one to each segment, and lateral segmental rami, one to each segment. These three segmental rami, the dorsal, lateral, and ventral, may all leave the horizontal septum nerve as one common ramus in each segment. In the anterior third of the trunk these common rami detach from the incipient longitudinal collector nerve, pass to the overlying skin, and divide into the three rami per segment — the dorsal, lateral, and ventral rami. The lateral rami innervate the overlying tubed lateral line scales. In the posterior two-thirds of the trunk, a greater number of common rami detach from the horizontal septum nerve. Each may have a small connection to the longitudinal collec- tor nerve. All that seems to detach from the dwindling in- cipient longitudinal collector nerve in the posterior half of the body is a short dorsal ramus in each segment that in- nervates lateralis organs in the tubed lateral line scale row. The longitudinal collector nerve ends on the caudal peduncle. The pattern in Aphredoderus sayanus is much like that in Percopsis transmontana except that in Aphredoderus both the tubed lateral line scale row and the dorsal longitudinal col- lector nerve are more dorsally situated. In Aphredoderus the tubed lateral line scale row ends a little past the origin of the dorsal fin except for a few widely separated tubed lateral line scales along the remainder of the horizontal septum. The dorsal longitudinal collector lateral line nerve is rudimentary in Percopsis. A similar but much higher placed collector nerve is a prominent feature in acanthopterygian and higher paracanthopterygian fishes. It has not yet been found in lower fishes. The collector nerve pattern is not present in the atheriniform fishes so far examined nor is it present in mugilids. It occurs in sphyraenids and polynemids but in Sphyraena barracuda it is not much more developed than in Percopsis . LATERAL LINE NERVES IN OTHER ATHERINIFORM FISHES (Figures 5 and 6.) In Melanotaenia nigrans (Melanotaeniidae) there is a large middorsal longitudinal ramus extending halfway between the horizontal septum and the middorsal line. It passes up No. 95] FREIHOFER: ATHERINIFORM NERVE PATTERNS from behind the cleithrum and ends about midway below the soft dorsal fin. There is no pectoral ventral lateralis ramus. The first ventral segmental ramus lies between the seventh and eighth ribs and is followed by about a dozen longer ventral segmental rami. There are no tubed lateralis scales . In an exocoetid and a hemiramphid (both unidentified) there is a large pectoral ventral lateralis ramus detached from the main lateralis nerve medial to the cleithrum. It supplies the tubed lateralis scales of the lateral line as the lateral line passes vertically down behind the pectoral fin, and also supplies the extension of the lateral line nearly to the isthmus. There are about 9 subsequent ventral rami, each turning posteriorly near the ventral midline and innervating numerous lateral line scales. In another species of hemiramphid, the enlarged pectoral branch was present, and, beginning about the fifth rib, there was almost one ventral ramus to each segment. In the cyprinodontid Profundulus guatamalensis there are 2 longitudinal dorsal rami that pass up from behind the cleithrum and course back as far as the middle of the soft dorsal fin. Each detaches separately from the main lateral line nerve. The first segmental ventral ramus begins at the level of pelvic fin origin. There are about 10 subsequent ventral rami. There are no tubed lateralis scales of the lateral line system. In the oryziatid Oryzias latipes (fig. 5) there are 2 longitudinal dorsal rami coming off together from the main lateral line nerve, a large pectoral ventral lateralis ramus and 4 or 5 succeeding ventral lateralis rami. Three speci- mens were examined. In the cyprinodontid Rivulus milesi there is apparently only one dorsal ramus, no pectoral ramus, and no ventral seg- mental rami. The main horizontal septum nerve appears to give off only a short ramus for each overlying naked lateralis organ or organs . In the cyprinodontid Pachypanchax playfairi (fig. 6) there are 2 longitudinal dorsal rami coming off together above the cleithrum and about 8 subsequent ventral rami, the first of which lies near the base of the pectoral fin and is the largest. There are no tubed lateralis scales in the lateral line. TRUNK LATERAL LINE NERVES IN LOWER FISHES In osmerids (Spirinchus thaleichthys) and in clupeids (Dorosoma petenensel there is a full development of the lateralis nerves very similar to that in Percopsis and Ather- inops. In salmonids (Salmo and Oncorhynchus) there is a horizontal septum nerve with a short lateral ramus supplying each tubed lateralis scale, but no ventral segmental lateralis rami, only one dorsal longitudinal ramus, and no development of a dorsal longitudinal collector nerve. The myctophoid Scopelengys tristis has a first dorsal longitudinal ramus extending towards the dorsal fin. From near the base of the first dorsal longitudinal ramus there is 8 CALIFORNIA ACADEMY OF SCIENCES [Occ. Papers a longitudinal horizontal ramus overlying the horizontal septum nerve and innervating about the first nine segments of the lat- eral line scale row. There are no ventral lateralis rami from the horizontal septum nerve. TRUNK LATERAL LINE NERVES IN PERCA FLAVESCENS (Figure 7.) In Perca f lavescens the lateral line tubed scale row lies about halfway towards the middorsal line and forms a gentle curve from the supracleithrum back to about one-third of the way along the caudal peduncle where it returns to the horizon- tal septum. The horizontal septum lateral line nerve runs from the supracleithrum back to where it bifurcates at the base of the caudal fin; the dorsal fork passes out to the tip of the caudal fin between the fourth and fifth rays from the mid- line of the fin; the ventral fork passes out between the third and fourth rays of the ventral lobe. The first dorsal longi- tudinal ramus reaches nearly to the dorsal fin. In the ante- rior curved section of the lateral line, the 7 short dorsal rami pass some distance dorsally up to the lateral line where each ramus turns posteriorly and extends along the lateral line supplying from 3 to 10 lateral line scales, the more an- terior rami supplying the greater number of scales. These short dorsal longitudinal rami forming the dorsally placed "collector" nerve constitute the characteristic feature of the pattern in Perca. Posterior to the point of return of the tubed lateral line scale row to the horizontal septum on the caudal peduncle each lateral line scale is innervated by a short lateral ramus of the horizontal septum nerve. The pat- tern in Perca is basically present in approximately 20 acan- thopterygian families examined so far. The trunk lateral line nerves of the sparid Spondyliosoma cantharus and the skipjack tuna, Katsuwonus pelamis, have been illustrated by Suckling (1967) . Some paracanthopterygians have a similar pattern, for example Gadus (Stannius, 1847) and Otophidium, but no fishes lower than paracanthopterygians are known to have the dorsal longitudinal collector nerve pattern. RAMUS CANALIS LATERALIS FACIALIS SYSTEM The ramus canalis lateralis facialis system of nerves consists of branches that in lower fishes usually course lengthwise in the membranous roof of the cephalic lateral line canals (Freihofer, 1970) . Most of them arise in the up- per cheek region from a branch coming up from the truncus hyomandibularis at a point near the articulation of the oper- cle with the hyomandibular . The ramus canalis lateralis facialis system is well developed in percopsiforms (reduced or absent in higher paracanthopterygians) , well developed in gobioids, present in a simpler state in umbrids and clupeids, present but reduced in beryciforms, and apparently present but still more reduced in some of the most generalized per- coids such as acropomatids, scombropids, centropomids and some others; it is well developed in apogonids. The system No. 95] FREIHOFER: ATHERINIFORM NERVE PATTERNS is not present and apparently has been lost in atheriniforms Its absence sets off all of the atheriniforms from most of the other fishes having ventral segmental trunk lateral line nerves . POSITION OF OLFACTORY BULB IN AGONOSTOMUS AND PERCOPSIS (Figures 8 and 9.) Most fishes have the olfactory bulb located at the fore- brain. The olfactory bulb is located about two-thirds of the way towards the olfactory organ from the forebrain in Mugil auratus (Holl, 1967) . In Agonostomus monticola the olfactory bulb is located at, or nearly at, the olfactory organ and is large. About one-third of the olfactory bulb lies within the concavity of the ethmoid cartilage leading to the very short canal through which the olfactory nerve connects with the ol- factory organ. On one specimen examined, the length of the olfactory tract is about one-half the length of the olfactory bulb. The olfactory bulb is about two-thirds the length of both the olfactory lobes and the rest of forebrain lobe. In another specimen of Agonostomus monticola of five inches total length, the distance between the olfactory bulb and olfactory lobe is about six times the length of the olfactory bulb or equal to the length of the forebrain and midbrain. The olfac- tory bulb in Percopsis omiscomaycus (fig. 9) is also located at or next to the olfactory organ (Freihofer, 1960a; 1960b; Gosline, 1963) . In the mugilids Joturus picardi and Valamugil seheli , the olfactory bulb is also at the olfactory organ. DISCUSSION Trunk lateral line nerves lend support to the systema- tic validity of the order Atheriniformes proposed by Rosen (1964). All of the atheriniform fishes examined have ventral segmental trunk lateralis nerves that detach from the main horizontal septum lateral line nerve. These trunk lateralis nerves set apart the atheriniform fishes from the perciform and beryciform fishes which do not have these nerves or a modification of them, but which have instead a dorsally placed "collector-like" longitudinal trunk lateral line nerve formed by branches from the horizontal septum lateral line nerve. The mugilids are an exception to this statement and are dis- cussed below. The ventral segmental trunk lateral line nerves also connect the atheriniform fishes with various lower-placed soft-rayed groups. Comparison of the various patterns of these nerves within the Atheriniformes raises some interesting systematic questions. The family Atherinidae apparently has the fullest devel- opment of the trunk lateralis nerves amongst atheriniform fishes. The other atheriniform groups examined have simpler patterns in that they have fewer ventral branches; Atherinops af finis has 23 to 26 ventral segmental trunk lateralis branches, whereas Oryzias has about 5. In Oryzias there is 10 CALIFORNIA ACADEMY OF SCIENCES [Occ. Papers an enlarged first ventral branch serving the pectoral and pelvic areas. In the cyprinodont Pachypanchax playf airi there is a smaller pectoral branch. It is followed by 8 ventral rami. In the exocoetids and hemiramphids there is an enlarged pectoral branch and about 9 succeeding ventral branches, each of which in some species turns posteriorly and innervates several lateral line scales. The family Melanotaeniidae has fewer ventral branches than do atherinid fishes judging from one species examined, Melanotaenia nigrans . Comparison of these ventral segmental trunk lateral line nerves in atheriniform fishes with those in other groups gives an indication of which may be the generalized pattern of these nerves in the atheriniforms. Other groups having these nerves in a pattern similar to that in the Atherinidae are the percopsiforms , osmerids, umbrids, stomiatoids, and clupeids. Judging from comparisons with these groups, it seems most reasonable that the pattern in the Atherinidae is the generalized condition and that the patterns in the exocoe- tids, hemiramphids, and cyprinodontoids are derived, special- ized patterns. On the basis of these lateralis nerves, the Atherinidae, therefore, would be the most generalized ather- iniform family and the other atheriniforms may have evolved from an atherinid-like ancestor. Spines in the fins of ather- iniforms would then be the generalized condition and fins without spines the advanced condition. The jaws of atherinid fishes appear no more specialized, perhaps less specialized, than those of other families. The same holds for the caudal fin skeleton. These and some other systematic interpreta- tions of atheriniform, mugilid, and sphyrainid fishes differ somewhat from the interpretations in recent papers and are discussed at the end of this report. The pattern of trunk lateral line nerves suggests what fishes may be the nearest relatives of atheriniforms. As was mentioned above, other groups having a basically similar pattern of trunk lateralis nerves are the percopsiforms, osmerids, umbrids, stomiatoids, and clupeids. The clupeids are morphologically too different to be considered close to the atheriniforms. Stomiatoids evidently are specialized offshoots from an osmerid type ancestor (Weitzman, 1967). The development of the trunk lateral line nerves looks simpler in umbrids than in osmerids, while the ramus canalis later- alis facialis system in umbrids is simpler than that in Percopsis and about as developed as, but somewhat different from, that in osmerids. The percopsiforms not only have a pattern of trunk lateralis nerves more similar to that of the atherinids, but they also have in Aphredoderus (but not in Percopsis) apparently the same basic pattern of the recurrent facial nerve (RLA) . Percopsiforms have spines in their fins as do atherinids, whereas osmerids and umbrids lack both spinous fins and RLA. The caudal fin structure of atherin- ids could feasibly be derived from a percopsid type caudal fin. In reference to atherinid relationships it would be desirable to have hybridization experiments between atherinids and percopsiforms carried further. Those made by Clark Hubbs (1970) between Menidia audens and Aphredoderus sayanus No. 95] FREIHOFER: ATHERINIFORM NERVE PATTERNS 11 terminated before gastrulation and so were not deemed suc- cessful. Perhaps further attempts using other atherinids with both Percopsis and Aphredoderus would be successful. The myctophoids do not seem as likely to be as close to the origin of atheriniforms as do the percopsiforms. Mycto- phoids lack ventral segmental trunk lateral line nerves judging from Ray (1950) for Lampanyctus leucopsarus and from a Sihler nerve preparation of Scopelengys tristisl As far as is known myctophoids also lack the recurrent facial nerve (RLA) . Salmonids (Salmo gairdneri and Oncorhynchus tschawyts- cha) lack the ventral segmental trunk lateralis nerves, RLA, and the ramus canalis lateralis facialis system. It has been hypothesized (Freihofer, 1970) that the per- copsiform fishes and the myctophoids have each evolved from an osmerid-like ancestor. In this connection the interesting similarity between true smelts and atherinids may be indica- tive of an actual relationship of atherinids to osmerids by way of percopsiform-like intermediate ancestors. (In teach- ing we have had to apologize to students and to ourselves that the top smelts, Atherinops af finis, family Atherinidae, are not related to surf smelts, Hypomesus pretiosus, family Osmeridae. We still may apologize but perhaps now we will do so with less conviction that the common names are com- pletely misleading.) Artificial hybridization experiments between surf smelts and top smelts or some other osmerid- atherinid combinations should be worth trying (fig. 12). Returning to the trunk lateral line system in the Ather- iniformes, it seems evident from comparing the distribution of nerves and tubed lateral line scales in Atherinops af finis (figs. 2 and 3) with the ventrally placed tubed lateral line scale row in a hemiramphid or exocoetid that the evolutionary sequence of formation of the ventral exocoetoid lateral line may have been somewhat as follows. In the generalized hypo- thetical atherinid ancestor there was a regular midlateral tubed lateral line scale row and, in addition, naked later- alis organs scattered over much of the body but most numerous ventrally and anterodorsally as occurs in Percopsis . These naked lateralis organs would be innervated by ventral seg- mental trunk lateralis nerves. A later stage would be forma- tion of scattered lateral line tubed scales towards the ven- tral side of the trunk as well as some towards the anteromid- dorsal line together with the simultaneous beginning of dissolution of the midlateral lateral line scale row, a stage approximated in Atherinops af finis . Lastly there would be the organization of a ventral tubed scaled lateral line from the initial ventrally scattered tubed lateralis scales and the disappearance of the anterodorsal incipient lateral line and the midlateral lateral line. The ventral naked lateralis organs and the ventral segmental trunk lateralis nerves such as occur in percopsiform fishes would be a preadaptation to the formation of such a ventral lateral line scale row. A sim- ilar but not as advanced ventral lateralis system developed in some paracanthopterygians. An alternative explanation to the evolution of the ventral lateral line in exocoetoids is that the whole midlateral lateral line gradually moved more ventrally with the nerves that supply each lateral line scale 12 CALIFORNIA ACADEMY OF SCIENCES [Occ. Papers continually lengthening and keeping up with the lateral line scales, but this process would not be necessary since the atherinid ancestors apparently already had long ventral seg- mental lateral line nerves in the ventral areas. This alter- native explanation, however, may hold for the part-way ventral lateral line of many cyprinoid fishes which do not have ventral segmental lateralis rami reaching almost to the midventral line. In comparing trunk lateral lines in Xenomugil (fig. 1) with those in the Acanthopterygii, Ather inif ormes , Paracan- thopterygii (especially with Percopsis) , osmerids, salmonids, and clupeids, it is seen that the trunk lateral line nerve pattern of Xenomugil has two major features in common with patterns in the Atherinif ormes , Paracanthopterygii, osmerids, and clupeids. These features are that there is a first ven- tral segmental lateralis ramus which is usually enlarged and serves as a pectoral branch and that there are present one or more additional ventral lateralis rami. In Xenomugil the single ventral ramus in addition to the pectoral ventral ramus extends longitudinally over a number of segments thus taking the place of a series of individual ventral segmental rami. In another mugilid from Australia, Myxus elongatus, there are two ventral rami instead of one in addition to the pectoral ramus. In Xenomugil the pectoral branch is large and serves the ventral region back to and for a short dis- tance past the pelvic fin. Xenomugil also has two dorsal longitudinal rami as has also Ather inops, Oryzias, and Pachypanchax . Xenomugil also has pored lateralis scales over much of the body, a feature which is not found in per- coidean fishes but which is a lateralis feature similar to the scattered lateral line scales in atherinids. Thus in pattern of trunk lateral line nerves, Xenomugil is like the paracanthopterygians and atheriniforms and unlike the acan- thopterygians . These basic similarities in the pattern of trunk lateral line nerves plus the general morphological similarities mugilids have to atherinoid fishes suggest that the origin of mugilid fishes might be looked for either in an ather inomorph or a percopsiform type of fish. There are several additional morphological features that point more to the percopsiforms as being closer to ancestors of mugilids. Agonostomus monticola, a mugilid occurring in fresh water in Central America has, as described above, the olfactory bulb located far forward, very close to the olfactory organ and not at the forebrain, where it is located in most teleosts Some species at least of the mugilid genera Joturus and Mugil also have the olfactory bulb far forward. The only other teleosts at all close to mugilids that also have it located at or near the olfactory organ are Percopsis and the gadid fishes. The olfactory bulb is at the forebrain in Aphre- doderus. A fact that by itself is not important is that in Agonostomus there are small, toothless fleshy structures re- sembling gill rakers on the hyoid arch (fig. 10). It is sug- gested here that these structures in Agonostomus monticola are rudimentary gill rakers. Gill rakers are not present as far as is known on the hyoid arch of atheriniforms, acanthop- terygians, or any paracanthopterygians except Percopsis, No. 95] FREIHOFER: ATHERINIFORM NERVE PATTERNS 13 which has a series of toothed plates along the hyoid arch that apparently are the remnants of gill rakers. Neither the toothed plates nor the rakers are present in Aphredoderus . Amia calva has hyoid gill rakers and so do a few other lower fishes. Toothed plates that undoubtedly are remnants from gill rakers that were once present occur on the hyoid arch of the beryciform fish, Polymixia lowei, which is another simi- larity it has to Percopsis. Agonostomus monticola was the only species among those of several genera of mugilids ex- amined that had gill raker-like structures on the hyoid arch. Rather tall and long-based projections tapering to blunt points are present on the ceratohyal of the maenid Smaris macrophthalmus . Structures are present on the prominent fold of skin lying along the edge of the hyoid arch of the hexa- grammid Hexagrammos decagrammus that strongly resemble the toothless or nearly toothless gill rakers of the first gill arch. Generalized scorpaenoids examined such as species of Sebastes, Helicolenus, Ectreposebastes , and Setarches lack gill raker-like structures on the hyoid arch. Generalized percoids examined lack such structures also. It would appear that the structures noted on the hyoid arch in the percoid Smaris macrophthalmus and in the scorpaenoid Hexagrammos decagrammus are independently derived and may not be gill rakers at all. The structures on the hyoid arch of the mugil- id Agonostomus monticola may not be rudimentary gill rakers either but in light of other similarities of mugilids to atheriniforms and of atheriniforms to percopsiforms, it seems probable that they are remnants of hyoid arch gill rakers. Histological study of the structures are needed. Mugilids have an A-^ jaw muscle that appears to be equiva- lent to the A-, of the gadid Microgadus proximus and one that is similar to but not as developed as the A-|_ of percoids. The caudal fin skeleton of mugilids with its two epurals could be derivable from that of a percopsiform type. The mugilid pelvic girdle falls short of being fully thoracic (Gosline, 1962) . The mugilid anal fin is in an interesting transitional state. It shows a third anal spine developing with age from an articulated soft ray (Jacot, 1920; Hubbs, 1944) . Specimens of Xenomugil thoburni of the same size (three inches) exhibit the whole range from soft ray to pun- gent spine for the third anal ray. The first ray of the second dorsal fin in mugilids is also articulated but shows considerable fusion towards becoming a spine. Either mugilids are still in the early stages of development of spines from soft rays or their spines are changing into soft rays. The former alternative supports the hypothesis that mugilids are offshoots from an ancestor that was in the early stages of evolving spinous fins from soft-rayed fins. Two other similarities of mugilids with percopsiforms are worth noting. One is that the cephalic lateralis canals in Agonostomus and perhaps in most mugilids are covered by membrane, not by bone, and hence are open as in Percopsis, osmerids, and paracanthopterygians in general. The open con- dition in these fishes is apparently the generalized and not the advanced state. The other similarity is that the lacri- mal bone is serrated with strong teeth and looks much like the lacrimal in Aphredoderus and the fossil percopsiform Sphenocephalus . 14 CALIFORNIA ACADEMY OF SCIENCES [Occ. Papers There are, therefore, a number of specific morphological features that support the hypothesis that mugilids may have evolved from a percopsiform type of fish. Another aspect of mugilids is that they are thought to have reached what is broadly called the percoid level of or- ganization. Supporting this view is the fact that mugilids have a pattern of the recurrent facial nerve (RLA) like that of Serranus which has one of the main percoid patterns of RLA. According to Rosen (1964) mugilids, sphyraenids, and polynemids have their jaw skeleton, jaw musculature, and caudal fin, as well as almost all other structural complexes usually studied, constructed much like they are in percoids. This may be correct, but what the new facts reported here for mugilids indicate is that the percoid developments are asso- ciated with other structural states that are prepercoidean and some that specifically are percopsiform. Sphyraenids and polynemids still need more study. The polynemid pattern of the recurrent facial nerve is not a Serranus pattern as reported earlier from dissection of a preserved specimen (Freihofer, 1960) . New work (unpublished) on polynemids shows these fishes to have a pattern of the re- current facial nerve much like that in scorpaenids. The scorpaenid pattern is most similar to the Serranus pattern as is also the pattern in polynemids. The polynemids also have a percoid or acanthopterygian pattern of trunk lateral line nerves. On this evidence polynemids would be further removed from the old percesocine affinities they have often been given. There is at present insufficient evidence of similarity of polynemids to other percoid families to place them in close relationship to any family of these fishes. The pattern of RLA if it is any yardstick of affinity would indicate that polynemids stand in the basal complex of early perciform evolution somewhere near the scorpaenoid line. Reflecting this possibility, the polynemids may be placed as a suborder before the Percoidei but with the notation that they apparently did not give rise to other percoids. A similar problem exists with sphyraenids. They differ from atheriniform fishes in many structural complexes (Rosen, 1964) but they also have two important indications of a pre- percoidean ancestry. One is the trunk lateral line pattern and the other is the pattern of the recurrent facial nerve (RLA) . In Sphyraena barracuda the "collector" nerve is rudi- mentary. It lies very close to the horizontal septum lateral line nerve, and it is short, so short that it barely reaches beyond the tip of the pectoral fin before it returns to the horizontal septum. There are only two dorsal rami forming it. In percoids the collector nerve returns to the horizon- tal septum on the caudal peduncle. It also lies much more dorsally in percoids and has numerous dorsal rami composing it. The short and low condition of the longitudinal collec- tor nerve may well not be the result of the elongate body form of barracudas. The trunk lateral line nerves in Sphyraena look suspiciously like they are partially in a pre- percoid pattern . In pattern of RLA, Sphyraena resembles Aphredoderus and Atherinops af finis . In these fishes RLA emerges from the No. 95] FREIHOFER: ATHERINIFORM NERVE PATTERNS 15 cranium through the parietal bone or from under its posterior edge and courses back under the skin to the dorsal fin. In these fishes there is a dipping down of RLA towards the ver- tebral column in the area of the pectoral fin. It is at this point or at the posterior end of the cranium that there are apparent nerve fiber connections of RLA with crossing dorsal rami of the occipitospinal or first spinal nerves. Fibers from RLA presumably could be distributed to the pectoral area via the ventral rami of the spinal nerves in these fishes. The nerve fiber connections of RLA with the spinal rami need confirmation by special stain preparations showing the fiber tracts. The pattern of RLA appears to be basically the same in Aphredoderus, atherinids, and sphyraenids and different from patterns in other fishes so far examined. If the details of the fiber relations of RLA with the dorsal rami are correctly interpreted here, then this similarity in RLA and in trunk lateral line pattern would be of basic systematic importance in setting up a relationship between these fishes. The condition of the first ray of the second dorsal fin in the Sphyraenidae may also indicate that sphyraenids are descendants of fishes which were in the early stages of de- veloping spines in their fins. In sphyraenids the first ray of the second dorsal fin has fused to the point of having lost its articulations but not to the point of having lost its bi- laterally double structure (Hubbs, 1944). The two dorsal fins are also widely separate in the sphyraenids as they are in some other seemingly generalized early spiny-rayed fishes (e.g., percopsiforms, atherinids, phallosthethoids, mugilids, acropomatids, and apogonids) . The several features discussed for sphyraenids give cause for questioning the present systematic placement of barracudas (Greenwood et al., 1966) as a suborder of perciform fishes presumably derived- From a generalized percoid ancestor of a hypothetical sea bass type as implied in Regan's (1913) classification of percoid fishes. An alternative view is that sphyraenids, similarly to mugilids, have branched off early from a percopsiform-like ancestor and attained some percoid-like developments but did not reach a generalized percoid level and are not percoid derivatives. A reasonable position for the barracudas for the present is as a suborder placed before the suborder Percoidei and separate from the Mugiloidei. The possibility exists that similarities between some of the atheriniform groups is the result of convergence and does not indicate phylogenetic relationships. It seems far more likely that similarities between these fishes are due to natural relationships. There has not been any good demonstra- tion of convergence in them. Unless there is good anatomi- cal evidence of convergence, the safer procedure is to accept the similarities until convincing evidence of convergence is available. Otherwise one can be wrong for the right reasons. If the superorder Atherinomorpha (Greenwood et al_ . , 1966) was proposed mainly on the basis of spines In the ather- iniforms having been evolved independently from the spines in other acanthopterygians, then the rank of superorder for the 16 CALIFORNIA ACADEMY OF SCIENCES [Occ. Papers atheriniforms should be abandoned. No reasons were given for its creation. There is little or no evidence that the nonspinous atheriniform fishes are the most generalized. There is good evidence for the fact that spines in fins were the generalized atheriniform condition and that spines were present in the fins of their forebears. The atheriniform fishes can be depicted as follows for the three systems of nerves described and discussed in the preceding sections. The ventral segmental trunk lateralis nerves are well developed in the generalized forms such as the Atherinidae, but are reduced in number of ventral rami in more specialized forms. There is no evidence of a dorsal longitudinal collector-like lateral line nerve. The recur- rent facial nerves are present in atherinoids and exocoetoids and are absent, presumably through loss, in cyprinodontoids, the pattern of RLA present being most similar to that in Aphredoderus . The ramus canalis lateralis facialis system is apparently absent or not recognizable in atheriniforms. The following changes in atheriniform and mugilid clas- sification are recommended: that the atheriniforms be placed in the superorder Acanthopterygii; that the basal family in the order be the Atherinidae of the suborder Atherinoidei; that the suborders Exocoetoidei and Cyprinodontoidei be con- sidered as derivatives of an atherinid-like ancestral stock; that the atheriniforms be shown in a phylogenetic classifi- cation as descended from percopsiform ancestors; that the mugilids be placed first as a suborder in the perciforms with the notations that they did not give rise to other percoids and that they are apparently atheriniform relatives but per- copsiform derivatives as are possibly also the perciforms and beryciforms. Figure 11 shows the proposed systematic scheme for the hypothetical origins and relationships of percopsiforms, atheriniforms, perciforms, beryciforms, and paracanthop- terygians. Lastly, by way of summary, the broader systematic inter- pretations discussed in the preceding pages are compared with the interpretations of the most recent authors working on atheriniform classification (see Gosline, 1962, 1963, 1968, and 1971; Rosen, 1962, 1964; Greenwood et al., 1966; Rosen and Patterson, 1969) . A point of ma jor- cTirTerence between the views of these workers and the views given in the present paper concerns the origin of the atheriniform fishes. Rosen (1964) postulated that the precursor of the atheriniforms may have been basically of a hemiramphid form with an incipient tendency to develop spinous rays but without a spinous or an- terior dorsal fin. The view in the present paper is that the precursor was basically of an atherinid form with a spinous anterior dorsal fin. Rosen (1964) specifically argued against any connection of atheriniforms with percopsiforms. Gosline most recently (1971) thought that the origin of the cyprino- donts was at about the percopsiform level of organization. Such an origin may be correct, but the view expressed in the present paper is that the cyprinodontoid fishes have evolved from an atherinoid-like ancestor which in turn was derived from a percopsiform-like ancestor. Anatomical studies by No. 95] FREIHOFER: ATHERINIFORM NERVE PATTERNS 17 Rosen (1964) , and by the author in the present paper and hy- bridization experiments by Clark Hubbs (1970) support the hypothesis that cyprinodontoids have closer relationships to atherinids than they do to percopsiforms . Gosline (1962) pointed out similarities between cyprinodontoids and ather- inids that would seem to indicate relationship, but then he wrote that the similarities between these two groups could be the result of adaptations to a common mode of life. He con- cluded that the two groups were convergently alike and not phylogenetically related. Little is known about the "mode of life" of any of these fishes. Besides, the two groups could have similar "modes of life" and be phylogenetically related. Rosen (1964) concluded that cyprinodontoids and atherinoids were part of a phylogenetically natural group in- cluding also the phallostethoids, exocoetoids, scomberesocoids, and adrianichthyoids . Evidence from particular groups of nerves presented in the present paper supports Rosen's inter- pretations. The nerve evidence does not support the placing of the Atherinidae together with the Mugilidae, Sphyraenidae, and Polynemidae as a suborder in the Perciformes as advocated by Gosline (1971) . The views presented in the present paper agree with Gosline 's consideration of the Mugilidae as an early percoid-like group with a relationship to the Atherini- dae and the placing of the Mugilidae as a suborder at the be- ginning of the Perciformes. The further views advocated in the present paper are that the Mugilidae is the lone family in that suborder and that the relationships of the Mugilidae to the Atherinidae are as two separate lines stemming from a percopsiform-like ancestor. These points bear on Rosen's conclusion (Rosen, 1964, p. 260) that mugilids, sphyraenids, and polynemids are not separable from the Perciformes. The nerve evidence supports the view that mugilids have an an- cestral, lineal relationship to percopsiforms and a collat- eral relationship with atherinif orms . Mugilids, therefore, in a sense would be partially separable from perciforms. The nerve evidence also supports the view that sphyraenids may have a percopsiform or atheriniform relationship and that they never were full percoids and were not derived from percoids. Sphyraenids also would be in a sense partially separable from perciforms. It has been suggested in the present study that, although mugilids are percoids of a type, traces of their ancestry are still apparent, an ancestry that is percopsi- form and not beryciform. If a percopsiform ancestry is cor- rect for mugilids, the most interesting new possibility is that the same ancestry may also hold for the most generalized percoids. The success of hybridization experiments (Clark Hubbs, 1970) between the pirateperch, Aphredoderus sayanus, and darters (family Percidae) seems to be significant sup- port of such an ancestry. 18 CALIFORNIA ACADEMY OF SCIENCES [Occ .Papers LITERATURE CITED ALLIS, EDWARD PHELPS, JR. 1897 FRASER, T 1971 FREIHOFER, 1960a. 1960b. 1963 1966 1970 GOSLINE, W. 1962. 1963 1968 1971 GREENWOOD, 1966. HANDRICK, 1901 The cranial muscles and cranial and first spinal nerves of Amia calva. Journal of Morphology, vol. 12, no. 3, pp. 487-808, pis. 20-37. H., AND W. C. FREIHOFER Trypsin modification for Sihler technique of staining nerves for systematic studies of fishes. Copeia, no. 3, pp. 574-576, 1 fig. W. C. Neurological evidence for the relationships of some percomorph fishes. 196 pp. Doctoral dis- sertation, Stanford University, Stanford, California . Dissertation Abstracts, vol. 31, Oct. 1960, no. 4, University Microfilms, Inc., Ann Arbor, Michigan. Patterns of the ramus lateralis accessorius and their systematic significance in teleostean fishes. Stanford Ichthyoloqical Bulletin, vol. 8, no. 2, pp. 79-189, 29 figs., 2 tables. The Sihler technique of staining nerves for sys- tematic study especially of fishes. Copeia, no. 3, pp. 470-475, 2 figs. Some nerve patterns and their systematic signifi- cance in paracanthopterygian, salmoniform, gobioid, and apogonid fishes. Proceedings of the California Academy of Sciences, ser. 4, vol. 38, Festschrift for George Sprague Myers, no. 10, pp. 215-263, 20 figs. A. Systematic position and relationships of the percesocine fishes. Pacific Science, vol. 16, no. 2, pp. 207-217, 3 figs. Considerations regarding the relationships of the percopsiform, cyprinodontiform, and gadiform fishes. Occasional Papers of the Museum of Zoology, University of Michigan, no. 629, 38 pp., 11 figs., 2 tables. The suborders of perciform fishes. Proceedings of the United States National Museum, vol. 124, no. 3647, 78 pp., 12 figs., 3 tables. Functional morphology and the classification of teleostean fishes. Honolulu. The University Press of Hawaii, IX+ 208 pp., 29 figs. P. H., D. E. ROSEN, S. H. WEITZMAN, AND G. S. MYERS Phyletic studies of teleostean fishes with a pro- visional classification of living forms. Bul- letin of the American Museum of Natural History, vol. 131, article 4, pp. 339-456, text-figs. 1-9, pis. 21-23, charts 1-32. Zur Kenntnis des Nervensystems und der Leucht- organe von Argyropelecus hemigymnus . Zoologica, Band 13, Heft 32, 68 pp., 6 pis., Stuttgart. K No. 95] FREIHOFER: ATHERINIFORM NERVE PATTERNS 19 HOLL, A. 1967 HUBBS, CARL 1944. HUBBS, CLARK 1970 JACOT, H. 1920 RAY, D. L, 1950, REGAN, C. 1913 ROSEN, D. 1962 1964 ROSEN, D. 1969 STANNIUS, 1849 SUCKLING, 1967 WEITZMAN, 1967 Zur Position der Bulbi olfactorie bei Knochen- fischen unter besonderer Beriicksichtigung von Mugil auratus (Actinopterygii , Mugiliformes) . Helgolander wissenschaf tliche Meeresuntersuch- ungen. vol. 16, pp. 255-261, 3 figs. L. Fin structure and relationship of the phallostethid fishes. Copeia, no. 2, pp. 69-79. Teleost hybridization studies. Proceedings of the California Academy of Sciences, ser. 4, vol. 38, Festschrift for George Sprague Myers, no. 15, pp. 289-298. Age, growth and scale characters of the mullets, Mugil cephalus and Mugil curema . Transactions of the American Microscopical Society, vol. 39, pp. 199-299, pis. 20-26, figs. 1-7. The peripheral nervous system of Lampanyctus leucopsarus. Journal of Morphology, vol. 87, no. 1, pp. 61-178, 6 text-figs., 1 table, 17 pis. The classification of the percoid fishes, and Magazine of Natural History, series 12, pp. 111-145. Annals 8, vol. H. Comments on the relationships of the North Ameri- can cave fishes of the family Amblyopsidae. American Museum Novitates, no. 2109, 35 pp., 24 figs. The relationships and taxonomic position of the halfbeaks, killifishes, silversides, and their relatives. Bulletin of the American Museum of Natural History, vol. 127, article 5, pp. 217- 268, 1-23 figs., pis. 14-15. , , AND C. PATTERSON The structure and relationships of the paracan- thopterygian fishes. Bulletin of the American Museum of Natural History, vol. 141, article 3, pp. 357-474, figs. 1-74, pis. 52-78, tables 1-8 Das peripherische Nervensystem der Fische. Rostock: 156 pp., 5 pis. Trunk lateral pects. In: P. H. Cahn. pp. 45-52. H. The origin of line nerves: Some anatomical as- Lateral line detectors, edited by Indiana University Press (496 pp.) the stomiatoid fishes with comments on the classification of salmoniform fishes Copeia, no. 3, pp. 507-540, 18 figs. 20 CALIFORNIA ACADEMY OF SCIENCES [Occ. Papers ADDENDUM Partial list of material examined. The letters SU or CAS before a catalogue number indicate the specimens are from the fish collection of Stanford University or of the California Academy of Sciences. The Stanford collection was recently (1971) removed to the California Academy of Sciences. Agonostomus monticola, SU 37434. Gorgona Island, Colombia. Atherinops af finis. Not catalogued. San Francisco Bay, California. Melanotaenia nigrans, CAS 14411. Aquarium specimens, no data. Myxus elongatus, CAS 144 06. Dee Why Lagoon, near Sydney, Australia . Oryzias latipes, CAS 14408. Aquarium specimen, no data. Percopsis transmontana , not catalogued. Marys River, near Wren, Oregon. Rivulus milesi, CAS 14412. Sphyraena barracuda, not catalogued. Miami, Florida, 1967. Xenomugil thoburni, CAS 14407. Galapagos Islands. No. 95] FREIHOFER: ATHERINIFORM NERVE PATTERNS 21 •H E O c X c •H en U-l O >H O W d) <4-i OlTl O 0) > u u o 0) M-i •P o, C 0) -i >-l -H U ■P tn 0) S 0) x; a. Eh >iO a u c H >-l Cm 3 .Q O iw -H — 4-> (0 (0 W M 0) CPfH tO tO Q W 22 CALIFORNIA ACADEMY OF SCIENCES [Occ. Papers w Cu o c •H U Cu •H M O 4-1 W O d) T3 (/I > d) C -P X 0) CD c -p •H .H ^-. (0 CD i-H i +J-HXI (d rH i td CD (d rH ^ ^i O •H -P O -H g G G 73 n) (DM C i+-l > -H — 03 m o> w w O rH 0) •H (d u c coo -H 0 W (^ 14-1 ■H m -P 03 0 td P-H -P iH rH 03 O rd >i (h > MH 0 0) 0) c C -P O •H C fd M •H rH Q) CD g Xi CD 73 O -P tp CD 03 < (d X P 3 - in WEh-O 0 0) irH Q) M-i ,G H o x: o (d g c 0 tr> w >i -H C 73 10 P C •H rH -H ■- Q iH >i >i •H iH iH iW rH • «• (d ro ^ mh j-i 0) 0 P H (d C os tj o C -H H -H rH M Em >n a> • Q) rH xi w 43 (d g -P -P M 3 0 < ID CO 24 CALIFORNIA ACADEMY OF SCIENCES [Occ. Papers C 0 •H W -P ■H Cu W •H Cu P. 0 O O to ^ H CO -P 0) X > 0) p. -p d) C d) 0) 0) w c •H rH • ,— s -H 0) tO (0 U V Q) -H P W f0 a -H O o ^ p. C (1) 3 to u p >1 rH 1 a U ■H O ^1 o m w 0 0 U 4-1 •H rH (D (D rH W (0 >-< P (0 S-l -H +J N >i CD >H X! O En >i rH • -H • IT) g U Pi ~ fd D g O w e H +J (T3 tfl •H rH Q 26 CALIFORNIA ACADEMY OF SCIENCES [Occ. Papers x fO >H o u to CM -P CM QJO S-l O M-l «4-l o p w X Q) 0> > P 0) 0) C 1 u >1 •H g «£> M-l O Pi o ■H n •H Pn 4-1 P id n3 No. 95] FREIHOFER: ATHERINIFORM NERVE PATTERNS 27 w C 0) u w • •H (0 4-> ■H n5 iW Ej to (0 u M ^ tn •H u 4-> CD D, c •H U -i IH a) -p -P (0 X iH CD -P ^ c (1) D CD S-i CO +) (1) • XJ ^~. Eh -l W i H iH fa •H E= (0 28 CALIFORNIA ACADEMY OF SCIENCES [Occ .Papers OB OOR FIGURE 8. The position of the olfactory bulb in the mugilid Agonostomus monticola. OB, olfactory bulb; OL, olfac- tory lobe of forebrain; OOr, olfactory organ; OT, olfactory tract. FIGURE 9. The position of the olfactory bulb in Percopsis omiscomaycus. OB, olfactory bulb; OL, olfactory lobe of fore- brain; OT, olfactory tract. No. 95] FREIHOFER: ATHERINIFORM NERVE PATTERNS 29 p •H 5-1 O -P G O •H P o u Cu O •H ■H U >1-p •H rfl CD rt T3 .H C 0 rd u •H .h -P fd C >i 0 A e o -p S-i Q) O m o 5 tn (U < 0|* fd > ig TS •P 3 •H >i +J c