DEPARTMENT OF COMMERCE BULLETIN OF THE UNITED STATES BUREAU OF FISHERIES VOL. XXXVIII 1921-1922 HENRY O'MALLEY COMMISSIONER WASHINGTON GOVERNMENT PRINTING OFFICE 1923 CONTENTS. Page. Ecological study of aquatic midges and some related insects with special reference to feeding habits. By Adelbert L. Leathers. (Document No. 915, issued May 26, 1922.) i-°2 Experiments in the culture op fresh-water mussels. By Arthur Day Howard. (Docu- ment No. 916, issued May 12, 1922.) 63-90 Further notes on the natural history and artificial propagation of the diamond- back terrapin. By R. L. Barney. (Document No. 917, issued April 24, 1922.) 91-na Notes on habits and development of eggs and larvae of the silversides Menidia menidia and Menidia beryllina. By Samuel F. Hildebrand. (Document No. 918, issued April 25, 1922.) 113-120 Deductions concerning the air bladder and the specific gravity of fishes. By Harden F. Taylor. (Document No. 921, issued April 24, 1922.) 121-126 Biology and economic value of the sea missel Mytii.is edulis. By Irving A. Field. (Document No. 922, issued July 11, 1922.) 127-260 A new bacterial disease of fresh-water fishes. By H. S. Davis. (Document No. 924, issued August 4, 1922.) 261-280 The spiny lobster, Panulirus argus, of Southern Florida: Its natural history and utilization. By D. R. Crawford and W. J. J. De Smidt. (Document No. 925, issued August 4, 1922.) 281-310 Some embryonic and larval stages of the winter flounder. By C. M. Breder, Jr. (Doc- ument No. 927, issued August 4, 1922.) 3II_310 The salmon of the Yukon River. By Charles H. Gilbert. (Document No. 928, issued November 21, 1922.) 3J7_332 General index 333~34i in \ 1 5 % H ERRATA. Page 4, third paragraph: Meclriocnemus should read Metriocnemus. Page 43, twelfth line from bottom: Culicoidies should read Culicoides. Page 86, eleventh line under "Commercial Possibilities:" Margaratifera should read Margaritifera. Page 93, Table heading: Terrapin should read terrapins. Last column of table: Reference b should read a in all three instances. Page 105, Legend under Fig. 81: Dash line preceding hibernated should be solid line. Page 127, third line from bottom in right-hand column: 775 should read 2/5. Page 128, sixth line from bottom: 12S should read 727. Page 138, ninth line from bottom: Second Exs should read ExS. Page 18S, fifth line from bottom: ptonucleus should read pronucleus. Page 193, seventh line from bottom: anlarge should read anlage. Page 209, fifth line below table: peridineans should read peridinians. Page 221, Last line: Colon preceding iron should be deleted. Page 281, ninth line from bottom in left-hand column: 282 should read 292. IV ECOLOGICAL STUDY OF AQUATIC MIDGES AND SOME RELATED INSECTS WITH SPECIAL REFERENCE TO FEEDING HABITS. By ADELBERT L. LEATHERS. J* CONTENTS. Page. Chironomida? 2 Introduction 2 Technique 3 Structure and function of head of Chironomus brasenice with reference to feeding habits 6 Subfamily Chironomins 9 Group I. — Chironomus lobiferus Say , 9 Habitat 9 Uses made of silk n Method of spinning silk _ 13 Silk structures 15 Related forms 16 Group II. — Tanytarsus pusio Meigen 17 Construction of tube 17 Variations in tubes 18 The net 18 Net making 18 Silk spinning 19 Adaptability 19 Group III. — Chironomus cayugcB Johannsen 20 Habitat • 20 The burrow 21 Feeding habits 22 Group IV. — Chironomus brasenur, n. sp., a leaf-eating chironomid , 23 Introduction 23 General habits 24 Life history 24 Penetrating the epidermis 26 The burrow 27 Respiration. .. ..• 28 Feeding habits 28 Economic importance 29 Control 29 Description of Chironomus braseniee, n. sp 30 Larva 3° Pupa 3° Male 30 Female 31 Group V. — Trichocladius nitidellus Malloch 31 Feeding habits : 31 The burrow 32 2 BULLETIN OF THE BUREAU OF FISHERIES. Chironomidae — Continued. Page. Group VI .' 33 Group VI: Subgroup A. — Metriocnemus knabi Coquillett 33 Head structures 33 Feeding habits 34 Group VI : Subgroup B. — Orthocladius sp.( ?) 35 Larval characters 35 Feeding habits 33 Group VI : Subgroup C. — Prodiamesa sp 36 Body structures 36 Mouth parts 36 Feeding habits 37 Subfamily Tanypinse 37 Mouth parts 39 Feeding habits 4° Subfamily Ceratopogonins: 42 Body structures 43 Head structures 44 Feeding habits ■ ■ 45 Summary 4° Orphnephilid* 4^ Habitat 4§ Feeding habits 5° Explanation of figures 51 Bibliography °° CHIRONOMIDAE. INTRODUCTION. The insects belonging to the family Chironomidae, commonly known as midges, constitute an obscure group of Diptera which, on account of their small size and inoffen- sive habits, have very largely escaped notice except as they may have been mistaken for mosquitoes, which they resemble only in general appearence. They are, however, very- common in every community from the polar region to the Tropics. The adults are often seen on moist evenings flying in dense swarms near the ground, over sidewalks, or under trees by the roadside, and it is in this brief period of their existence, consisting of from 5 to 10 days, that they are most familiar to the general public. Closely related to the Chironomidae are the Orphnephilidae, a family of semiaquatic insects as scarce as the Chironomidae are common, the only known habitat in this country being in the environment of Ithaca, N. Y. The larval stages of the Chironomidae, which extend over a period varying from all winter to 25 or 30 days, according to food and weather conditions, are only infrequently observed, chiefly because of the small size and secluded habits of the larvae. They are aquatic, mainly fresh-water, insects living in burrows which they construct by fastening together the debris found at the bottom of ponds with silk secreted by their salivary glands. The great abundance of these larvae and their relation to other aquatic organisms were the fundamental consid- erations that gave impetus to this study. It was hoped that an investigation of their feeding habits would give a clue to the chief adaptations which have given rise to their numerical dominance and widespread distribution. AQUATIC MIDGES AND SOME RELATED INSECTS. 3 That they do subsist in great numbers has been called to the author's attention not alone by his own observations, but by various published and unpublished works of students who have recorded them as forming an important part of the food of trout, suckers, and various other fish, and of salamanders, dragonflies, mayflies, and a variety of other predacious aquatic organisms. Herein lies the chief interest of these obser- vations from the fish-cultural point of view, that a careful study is made of a particular group of animals which are engaged in converting vegetable detritus and other organic materials existing in fishponds into a form suitable for consumption by fish. How useful they are as a direct medium in transforming and conserving the food supply furnished by the microorganisms found in small quantities in all habitats will be shown in greater detail in the subsequent discussion. In beginning this work the larvae of many species were examined in order to deter- mine their stomach contents. The organisms found were so similar, both in number and variety, to those available in a given locality that there seemed to be little or no sorting in their method of feeding. Consequently, attention was directed more to their method of capturing food than to the substances eaten, and it was here that the funda- mental adaptations were found which enable the different genera and species to live in a similar environment with a minimum amount of competition. The small size of the larvae and their great power of tiding over periods of food shortage, together with their capacity to live in habitats containing a scanty supply of oxygen, readily enable them to subsist where a larger animal would find the food supply insufficient or the environ- ment unsuited to its manner of life. In this study of the feeding habits the author has endeavored to associate into groups those larvae which obtain their food in an essentially similar manner. An attempt has been made to cover the entire family. The subfamily Chironominae has been divided into six groups, while the subfamilies Tanypinae and Ceratopogoninse each constitute but a single group. The number of these divisions shows in a somewhat graphic way the relative size and amount of specialization of the three subfamilies. It is to be hoped that these groups will be found sufficient to accommodate all the various species of the family, although the two consecutive seasons devoted to this work, in the absence of ' any considerable literature on the feeding habits of the larvae, are all too short a time to exhaust a study involving such small and relatively obscure organisms. This work was done in the entomological laboratory of Cornell University, under the direction of Prof. J. G. Needham, to whom the author is greatly indebted for much counsel, assistance, and encouragement in the prosecution of the work. The author wishes to acknowledge his appreciation of the assistance rendered by Prof. O. A. Johann- sen in the identification of specimens, general suggestions, and sympathetic interest and encouragement in every phase of the work. He is also greatly indebted for many favors from the various members of the Department, to whom he wishes to express his appre- ciation for the thoughtfulness that prompted such generous cooperation. TECHNIQUE. In order to carry on the laboratory experiments with various chironomids it was found desirable to keep a number of living larvae always on hand. For most larvae very simple containers proved most satisfactory. Those that live in the manner described under Group III were brought home, together with "a small mass of the debris in which 4 BULLETIN OF THE BUREAU OF FISHERIES. they were living, and placed in shallow agateware trays. The debris containing the larvae was usually spread out, so as not to be more than one-fourth of an inch deep, and was then covered to a depth of half an inch with tap water. After a day or so, when the larvae were especially numerous, as a rule they used all the loose debris in constructing rather long U-shaped tubes, where they usually succeeded in maintaining themselves for weeks at a time. The shallow agateware trays were rather generally used for the various forms of larvae that could be collected in numbers. They are especially to be recom- mended on account of the large amount of water surface exposed to the air, thus facili- tating aeration. In breeding the various species collected a considerable number of individual receptacles were required. In the early experiments square watch glasses with covers were used successfully. They were later discarded in favor of medium-sized test tubes. Such a tube with a cotton wool plug has numerous points of advantage over a watch glass. First, the cotton plug permits a free exchange of gases. This circulation pre- vents the accumulation of moisture on the inside of the test tube, so that the newly immerged fly is not so liable to be caught in a water film and drowned. Second, the cotton plug makes a very satisfactory surface to which a freshly immerged fly may cling. Third, a number of test tubes may be placed together in a slanting position, so that the water which they contain will expose a proportionally large surface to the air, thus insuring perfect aeration. Fourth, a considerable number of tubes may be placed in a tray and a uniform temperature maintained either by flowing water or by evapora- tion from the surface of standing water. Fifth, the data concerning the larva may be written on a small piece of paper and inserted with the cotton in the mouth of the test tube. Mectriocnemus knabi larvae were kept for several weeks by bringing in the leaves of the pitcher plant and placing them so that they would remain in an upright position. They were kept full by the occasional addition of small amounts of water. The larvae were also kept for weeks at a time in petri dishes containing the water and insect remains obtained by emptying the leaves of the pitcher plant. They do not appear to be so exacting in their environmental requirements as most chironomid larvae and can doubt- less be reared in most any sort of a container. Chironomus lobijerus were brought into the laboratory in water-soaked Sparganium stems, which were allowed to float freely in trays filled with water. In this condition the larvae maintained themselves for considerable periods at a time. Upon removing them from their burrow they were found to adjust themselves to various artificial receptacles. The most satisfactory glass preparations for the observation of the habits of Chironomus lobijerus larvae were constructed so as to give flat horizontal surfaces. This was accomplished by cementing two rectangular strips of glass cut from cover slips to either side of parallel capillary glass tubes. The size of the capillary tubes used was slightly larger than the full-grown larva. It was found that King's "Microscopical Cement" is more satisfactory than a cement made by dissolving asphaltum in turpentine or xylol, especially when it is desired to make permanent mounts of the silken tubes. It is necessary to dehydrate rapidly in order not to dissolve this cement, but even with this defect it is more satis- factory than other cements soluble in xylol. Because of the uniform thickness and the AQUATIC MIDGES AND SOME RELATED INSECTS. 5 flatness of the surfaces (fig. 26) this type of glass preparation is more satisfactory than any flattened tube the author was able to manufacture. Ehrlich's acid hematoxlin was found to be the most satisfactory of any stain used for bringing out the silk structures. "Licht green" and "eosin" were also used but were not found satisfactory. The licht green, while staining the silk glands and the silk within the silk duct, did not stain the silk outside of the body. The eosin, while staining the silk slightly, was found unsatisfactory because of the ease with which it was removed in dehydration. There are few animals that lend themselves more readily to a laboratory or lecture demonstration than do these stem-dwelling larva?. They seem to differ decidedly from other chironomids, especially those that characteristically live in a mud burrow, in their reactions to strong light. Their greater tolerance of light enables one to demon- strate the silk-spinning movements and- the general behavior of the larva by means of a stereopticon. The only requirements are that the larva? shall have recently built a fresh, clean silk tube in a glass preparation, and that the temperature of the water be kept down to normal room temperature. The reason for requiring that the silken tube be a fresh one is that after a week or two the silk becomes much discolored by the lodgment of fine particles as well as by the deliberate attachment of masses of castings to the ends of the burrow. It is only necessary to remove the larva by a jet of water and then the tube can be removed by a needle. A few hours will usually suffice to enable them to again replace the silken tube. The temperature is easily controlled by having a specially constructed lantern slide through which a current of water can be made to pass. A type used with considerable success was constructed as follows: Two pieces of sheet brass the size of a lantern slide were cut so as to give a sym- metrically placed rectangular opening \x,i by 2 inches near their center. These two brass strips were drilled and fitted with screw bolts. Two sheets of transparent celluloid and a single sheet of rubber packing material about an eighth of an inch in thickness were punched so that the holes coincided with those in the brass plates. The rubber packing was cut so as to give a rectangular opening which coincided with that in the brass plates, and the parts were assembled in the following order: Sheet brass, celluloid, rubber packing, celluloid, and brass. Two one-eighth-inch rubber tubes were connected with the inclosed chamber by openings at diagonally opposite corners. It was found easier to make this connection through one of the sheets of celluloid than through the rubber packing material. These tubes were fused in with beeswax and their ends weighted and put into separate jars. The tube opening at the bottom of the lantern slide was used as the intake tube and the jar to which it was connected was filled with cold water. Then by gravity the water was made to flow through the chamber. Two adjust- able pinchcocks were provided and the flow of water stopped while the glass prepara- tions containing the larva? were being placed in the chamber. Then by regulating the flow of water by means of the pinchcocks the preparation was used as long as desired. When the water had all been passed through, the jars were changed and it was sent through a second time. It was found desirable to have the water removed from the top of such a chamber on account of air bubbles which tend to accumulate when well- aerated water is heated slightly. 6 BULLETIN OF THE BUREAU OF FISHERIES. STRUCTURE AND FUNCTION OF HEAD OF Chironomus brasenise WITH REFERENCE TO FEEDING HABITS. The head of the chironomid larva, while so constructed as to be wonderfully adapted for feeding upon a large variety of foods in diverse environmental conditions, nevertheless shows a wide range of variations. These variations, while especially well marked in the subf amilies, are also to be found among the different genera and to a lesser extent within the genus. They have been taken advantage of by the systematists, who have figured the structures that best lend themselves to their purposes. Miall and Hammond (1900) and more recently Goetghebuer (191 1) have made more careful studies of these structures, with special consideration of their morphology. The object in this discussion is therefore to consider the special adaptation of the mouth parts, with particular reference to their function in the feeding habits. As certain of these structures have already been treated more fully than others, only those parts whose function appears to the author to be either poorly or inadequately discussed elsewhere will be considered in great detail here. In this study the head parts of Chironomus brasenice n. sp. are figured, and an at- tempt is made to point out the more conspicuous differences between this species and the larger and better known species upon which Miall and Hammond (1900) worked. The most noticeable feature about the head of the larva of C. brasenice is its great width rela- tive to the length. The labrum is also unusually narrow and the head has a roughly triangular outline (fig. 10). The labrum undergoes a remarkable amount of variation in minor details, such as the presence or absence of a thin triangular labral comb, variously arranged pectinate hairs, and paired lobular bodies. The structures located on the ventral surface of the labrum are commonly assigned to the epipharynx and consist of a three to many toothed epipharyngeal comb located on the anterior border, a thickened chitinized horseshoe- shaped area just posterior to it, within which are attached a variable number of claws or spines, and just outside of these spines a pair of peculiarly mandiblelike structures, known as premandibles (Goetghebuer) or lateral arms (Johannsen) . The corresponding struc- tures of C. brasenice are peculiar in being reduced in number, larger in size, and more strongly chitinized as an adaptation to its leaf-eating habits. The function of the labrum is that of a very complexed scraping organ, and the degree of specialization of its various parts is usually found to be correlated with the nature of the food and its method of collection. The labrum is rather less specialized in this species than in Chironomus cayugce and the others included by Goetghebuer (191 1) in his Group I. The pectinate hairs are fewer in number and simpler (figs. 1 and 2, s). The epipharyngeal comb (figs. 1 and 2, co) con- sists of three large rounded teeth with smooth inner surfaces. The horseshoe-shaped chitinous area (figs. 1 and 2, h) on the ventral surface of the epipharynx is in this species much less horseshoe shaped than usual. It is here represented by two chitinous bars which articulate in front with the thickened anterior border of the labrum and poste- riorly with a median caudad projecting process (fig. 1) within this horseshoe area. There are four pairs of chitinous hooks (figs. 1 and 2 , e) , rather blunt in outline in Chironomus brasenice, but often very much specialized and developed as minutely serrate plates. Just lateral to the posterior end of the horseshoe-shaped chitinous area are the " lateral arms" of Johannsen or "premandibles" of Goetghebuer (figs. 1 and 2, a). These are provided AQUATIC MIDGES AND SOME RELATED INSECTS. 7 with a mesad projecting process, which loosely articulates which the central chitinous structure (figs, i and 2). These arms are provided with muscles and are capable of a wide variety of movements. The lateral arms, while figured for a considerable number of species, do not seem to have been treated at all from a functional standpoint. From the author's experience it seems possible that the small size of the head and the constant activity of the larva have served to vitiate many attempts in this direction. It is easy to see from the study of a large number of dead larvse that the arms are to be found in a variety of positions, the most frequently observed position being that found when the labrum is drawn in between the maxillae. When the labrum is in this position, the arms project posteriorly down into the pharynx, just above the surface of the hypopharynx. When the labrum is raised somewhat, they are seen to lie just above the labium. When the labium is elevated, as in the normal feeding, the ends of the arms are farther forward. Several times while examining the labium of living larvae the author has observed what he considers the normal movements of these appendages. They are moved forward and toward each other when the labrum is elevated, so that their setigerous anterior mar- gins (fig. 2, a) scrape the chitinous claws (figs. 1 and 2, e) attached within the horseshoe area, removing any food material that they may have collected. They are then swung backward in close proximity to each other as the labrum is pressed down. From these occasional observations, together with the structure of associated parts of the pharynx, it seems reasonable to conclude that the lateral arms have an important function, as they convey the food down the alimentary tract to such a level that the circular muscles of the esophagus can act upon it in the swallowing process. They would, therefore, appear to supplement the mandibles and maxilla, which may have lost something of their primitive functions as an adaptation to their present manner of life. The mandibles have been figured by a large number of authors, especially from the systematic standpoint. The author has tried to show in detail the method of articula- tion of the mandibles with the head because of the restricted movements of these append- ages resulting from their method of attachment. The anterior median margins of the epicranial plate (fig. 5) carry on their inner surfaces special internal chitinous processes (figs. 4 and 5, i) upon which the mandibles articulate. These processes alone would give the mandibles a considerable freedom of movement. This movement, however, is some- what restricted by the process q (figs. 3 and 6) and the plate st (figs. 4, 1 1, and 12). Their chief movements are consequently confined to plains approximately a.t right angles to each other. In this motion they oppose the labium rather than each other. The complexity of the adductor muscles, however, enables the mandibles to oppose each other when elevated. The external process of the mandible (fig. 3, q), which projects beyond the point of articulation out over the thickened margin of the epicranial plate (fig. 12, />;), adds considerable firmness and rigidity. The function of the mandibles is of especial interest, because they, next to the labium, limit the range of adaptability of the chironomids. This is especially emphasized in the discussion of the adaptability of Chironomus brasenice and C. lobijerus. The maxilla (fig. 11) has been the object of considerable speculation especially as regards its homologies. Mundy (1909) figures vibrissa?, which he considers as replacing the striated structures shown in figures 11 and 12, c. This structure Goetghebuer (191 1) considers a part of the labium. The attachment of the movable parts of the 80285°— 22 2 8 BULLETIN OP THE BUREAU OF FISHERIES. maxilla is especially interesting in this connection. The structures below p (fig. n) are attached to the chitinous plate st (fig. n) along its outer margin. The parts marked / and g (fig. n) are attached to the part beneath the letter p (fig. u) and are capable of being folded over it. They articulate at w (fig. n) and swing inward. In attempting to homologize these structures the plate si (figs, n and 12) is considered as representing the stipes, which is fused to the anterior margin of the epicranial plate p. Evidently p (fig. 11) represents the palpus and the structures below it (/>) the palpifer; / and g represent the combined lacinia and galea and c the two cardos, one beneath the other. The movements of the maxillae are restricted by their attachment to a fixed plate outside the plane of movement of the mandibles. It seems probable that their function has been largely taken over by the labrum and especially the lateral arms. The anterior part (fig. 4, g and /) is capable of a considerable movement in a lateromedial direction and although rather thin is doubtless an important factor in the concentration of the food particles. This part of the maxilla, as well as the palpifer, carries a number of sense papillae which doubtless have more or less well-developed taste cells, as it is easy to see that the larvae have very acute taste organs in this part of the head. It therefore seems probable that as the function of the maxilla has decreased the maxilla itself has become very much modified. The labium in the family Chironomida? is very important in the determination of the larva and consequently is a familiar structure in the systematic literature. This structure is developed as a thickened plate with an anterior toothed margin. It is so closely fused with the lower surface of the epicranial plate that in many species it seems to be only a modification of the anterior border of this part of the head. This is especially noticeable in those species which show a suture between these plates in the labial region. This structufe is, however, capable of being removed as a separate plate, and more complete study will doubtless show a similar arrangement throughout the subfamily. Its function is that of a scraping and cutting edge, and it is next to the mandibles in its importance in governing the range of adaptability of the species. The hypopharynx (figs. 7 and 8) is furnished with chitinous plates th and a variety of spines and setae. This anterior portion is separated from the posterior by a cavity 2 (fig. 7), which is continuous with the salivary ducts d (fig. 8). It is supported by a chitinous ring / (fig. 8) . The posterior part is furnished with a large number of backward pointing setae on- its dorsal surface and is supported by a chitinous skeleton shown in figure 8, k and /. The arms (fig. 7, k) of the hypopharynx form a point of attachment for the upper end of the pharynx and hold this part extended. The function of the hypopharynx is doubtless sensory to a large extent, as its r61e of guarding the entrance to the alimentary tract and the exit of the salivary ducts would naturally demand. It seems probable that the backward projecting setae (fig. 7, e) at the entrance of the pharynx may also serve to disentangle the food material brought in by the lateral arms. The structure of the anterior and posterior borders of the cavity in the hypopharynx through which the silk escapes is of special interest in connection with the study of the silk structures spun both by this and other species of Chironomidae, although the part which it plays is still uncertain. AQUATIC MIDGES AND SOME RELATED INSECTS. 9 SUBFAMILY CHIRONOMINyE. Croup I. — Chironomus lobiferus Say. In this first group the author wishes to consider as a type one of the chironomid larva that seems to have departed most wide'ly from the more familiar examples. This species, however, is capable of living in a loose mud burrow and of collecting and eating its food directly from the surface of the accumulated debris about it, but this is not its most characteristic method of feeding when living in competition with other species. HABITAT. The burrows of Chironomus lobiferus may be found on floating logs, at the bottoms of ponds, or attached to stems, stumps, and other perpendicular surfaces. In these habitats the larvae live by straining the fine particles from the water which passes through their burrows. A still more unique mode of life is shown by C. lobiferus in the readiness and frequency with which it penetrates the stems of aquatic plants. A list of the plants attacked includes so nearly all the submerged aquatics that it is concluded that the structure of the epidermis is the important limiting factor. The presence of larvae within a stem is easily recognized by two small round open- ings through the epidermis which they make at either end of that portion of the tissue occupied by their burrows. These openings enable the larvae to set up a current through their burrows by throwing their bodies into an undulatory motion. In this way the larvae are able to obtain food and carry on their respiratory processes at the same time. The general behavior of Chironomus sparganii Kieffer [lobijerus(?)] larvae has been observed and well described by Willem (1908). The general facts are as follows : The larvae are found in both dead and living stems of Sparganium, in the softer tissue where the chlorophyll is lacking. They are commonly located some 8 or 10 inches below the surface of the water. In the dead and especially the well water-soaked stems of Sparganium they are to be found in abundance. Their burrows communicate with the exterior by two small openings from one-quarter to one- half millimeter in diameter. The openings are at varying distances from each other, but usually measure in a rough way the relative lengths of the larvae, the average distances being about 15 millimeters. The method by which the larvae penetrate these stems seems not to have been ob- served nor questioned so far as the literature is concerned. In Group IV is discussed the adaptation of the head of Chironomus brascnicc for burrowing, and evidence is given that the penetration of the uninjured epidermis is a matter of very considerable difficulty. The larvae of C. braseniec, however, show a unique adaptation to this procedure by spinning a special silken arch by which they are able to apply pressure more advanta- geously to their mouth parts. This phenomenon was not seen in a considerable series of C. lobiferus larvae that were kept under observation for this purpose, and it is concluded that this species has not yet developed such an adaptation. The experiments set up for the purpose of testing out the ability of a larva to enter an uninjured stem were of two kinds: First, outdoor experiments with uninjured stems fastened together as rafts and placed among the infested stems; and, second, small sections of infested stems taken into the laboratory and placed in watch glasses, in which IO BULLETIN OF THE BUREAU OF FISHERIES. several larvae removed from similar stems were placed. In the outdoor experiments the rafts were made of freshly cut stems about 2 feet long and were left to float freely in an infested portion of a pool where similar larvae could be taken at any time during the year. These stems were observed at intervals for two months, and none showed any signs of the presence of the larvae. In this relatively short period they showed but slight signs of decay and practically no accumulation of diatoms. In the laboratory experiments with sections of similar stems in a more advanced stage of decay only such stems as had already been infested were used. It was soon found that the larvae would readily accept these stems, which they usually entered by creeping into the openings at the ends. The sections cut to fit into a Syracuse watch glass could ordinarily be entered from the ends and were usually short enough to enable the larvae to maintain their water current without penetrating the epidermis. In order to make it necessary for the larvae to penetrate the epidermis, the cut ends were coated with melted paraffin. The result in many cases was that they simply crawled under the stems and spun their silken tubes, fastening them to the stem above and the glass below. In only one instance did a larva penetrate the stem from the side. In this case the opening was rough and jagged in outline and was located near one of the lower corners of the stem, where it seems fair to assume that the larva might have gained some advantage (by catching its posterior end under the edge of the stem) from the sharpness of the angle that would in a way compensate for its light weight in bringing pressure to bear on the mouth parts. This seems especially possible when it is observed that the claws of the pos- terior prolegs point forward and are capable of holding the posterior end of the body in place while the muscles of the body are used in flexing the body and holding the mouth parts of the larvae in contact with the epidermis. Observation on a series of stems selected at random from among a considerable number dipped up from the bottom of a pool where the larvae were abundant showed a greater number of larvae near the ends of the stems. In some cases a larva was so located that one end of its tube opened at the end of the stem and the other by an opening bored through the epidermis. Several stems were found to have openings along their entire length, but all were confined to what had been the inner or upper surface of the leaf where the epidermis was thinnest. In other cases the larvae had an opening on one side of the leaf with a long vertical tube leading to its gallery which was on the opposite side, where it opened to the surface through a thickened epidermis. Old Typha stems were occasionally found with larvae located near the broken ends, but in no case was there noticed a larval-made opening penetrating the epidermis. When the Typha stems were tested with a sharp point, the epidermis was found to be very much tougher than that of Sparganium which is most frequently inhabited by the larvae. The thickness and texture of the epidermis of a stem is an especially important factor, as the above observations indicate. This, however, is not the only source of evidence, but when considered in connection with the fact that two larval molts out of six examined had one of the lateral teeth of the labial plate broken (fig. 2S) it becomes evident that the larvae exert themselves to the limit in penetrating the various plants in which they construct their galleries. That the larvae more frequently penetrate the epidermis from the inside than from the outside seems to be shown from the greater AQUATIC MIDGES AND SOME RELATED INSECTS. II abundance of openings near the broken end of stems, and that the penetration is more easily accomplished in a small gallery, where the larva is able to brace its body against a somewhat resistant parenchymous tissue, is obvious when the nature of the larval mouth parts is understood. These structures have been fully discussed in connection with Chironomus brasenuB, and it is only necessary to consider them very briefly here. The labium is used as a cutting edge and is applied at an angle of about 45 ° to the surface. Pressure is brought to bear upon it by the mandibles, which on a flat surface have to be widely extended in order to bring their pointed tips into use. This pressure is therefore applied very largely as a sidewise pull and has the effect of using the labium more or less like a scraper. Hence, the strength of the larva and the toughness of its labial plate are important factors limiting its attack on plant tissue. USES MADE OF SILK. The fact that the larva of Chironomus lobifcrus lives as it does in a burrow which communicates with the exterior by two small openings, too small to allow the larva within to extend its body, naturally makes one curious to know how it is able to obtain food. The natural conclusion, of course, would be that it ate the plant tissue, but this is not found to be the case when the stomach content of the larva is examined. Willem (1908) observed this and stated that the stomach content was composed of organic debris analogous to that which floats in the water — "desmids, diatoms, Pedi- astrum, Clathrocystis, spicules of Spongilla, carapace of hydrachnids, rotifers, together with grains of sand and sometimes the fragments of plant diaphragms." The author's study of stomach contents fully corroborated the above observations, although at the time the author was not acquainted with Willem's work. In the author's study of the behavior of the larvae a number of burrows were cut from the stems with just enough tissue to prevent disturbing the silk lining. These preparations were placed in Syracuse watch glasses and observed under a binocular microscope. Considerable difficulty was encountered in seeing through the epidermis, so it was cut away and replaced with a cover glass. The larva readily readjusted themselves by making their burrows open at the ends of the section of tissue instead of up through the epidermis. In this way the behavior could be watched much more exactly, but it was not until one of the most characteristic performances, over an area where the underlying tissue had been entirely removed, was observed that a clew to the method by which the larvae obtain their food was discovered. Willem (1908) dis- misses this subject by stating that the food is removed by adhering to the walls of the burrow near the end at which it enters. The following statement, translated from the same source, seems to refer to the movement that gave this clue: Sometimes the larva is fixed posteriorly retracting and elongating in the act of going and coming rhythmically, its body playing the role of a piston for renewing the water in the tube. This movement, so well described by Willem, the author has been able to demon- strate is concerned in the spinning of a thin conical net across the end of the burrow. This net is used to strain the floating organisms out of the water which the larva forces through it by the rhythmic undulatory motion of its body. In this process the larva 12 BULLETIN OF THE BUREAU OF FISHERIES. clings to the silk with which its burrow is lined by means of the hooked claws on the anterior and posterior prolegs. The current of water which is driven through the burrow by the undulating motion of the body of the larva serves the double function of bathing the branchial gills, thus renewing the oxygen supply, and of bringing in whatever particles may be floating in the adjacent water that are of use to the larva as food. The normal undulations move from the head backward, and the larva always turns about after spinning its net, so that the current is driven into the open end of the conical net. The position assumed by the larva places the caudal filaments in such proximity to the net that they are able to serve a more or less important tactile function. When the larva has maintained this current for about 10 minutes (the time element appearing more uniform than the amount of food actually present in the net at any one time), it turns about in its burrow quickly and gathers in and swallows the catch, net and all. The net is "hauled" in a very characteristic way. The larva seizes that portion of the rim with which it first comes in contact. The mandibles, the labrum, and prob- ably the lateral arm of the epipharynx are brought into use, and the flimsy net is torn away from the silk of the burrow and crowded down the throat of the larva by the labrum. Then the larva rotates its body and seizes the other side, which is swallowed at once. Then the remainder of the net is swallowed while the larva rotates its body first to one side and then to the other as if to wring out or twist up the net, so that it can be more easily swallowed. The conical tip of the net usually contains a consider- able variety of plankton organisms ranging from bacteria, which are either stuck to the net or caught in its meshes, to crustaceans and various rotifers, which sometimes succeed in escaping but are nevertheless often captured. The entire process of "haul- ing" the net and eating it takes only about six seconds. The most striking and fundamental use made of silk by Chirotxotrms lobiferus is in the construction of a net by means of which the larva obtains its entire food supply. Silk has, however, other uses of very great adaptive importance even in this unusual habitat. Many burrows are found where old openings have been entirely sealed up by its use. The regular openings through the epidermis are usually made round and smaller in size by the addition of a silk margin, and the burrow itself is lined with silk which is uniformly made of such a diameter that the movements of the larva are espe- cially effective. This ability to spin a thin, flexible, and at the same time practically water-tight lining enables the larva to adapt itself to cavities of varying sizes. The small size of the openings at the ends of the burrow seems to be a special adaptation, for when the larvae live under the very different conditions afforded by glass tubes they retain this same habit. It seems probable that the narrow openings increase the speed of the current and so prevent Protozoa, Crustacea, and other small organisms from swimming against it. Large particles are also prevented from entering the burrow. In case these small openings are plugged by an accumulation of particles the larva stops its rhythmic undulatory movements and suddenly throws its body into several much shorter waves which move in the reverse direction. This sets up a strong countercurrent which usually dislodges the obstruction, although the contents of the net are usually lost. In case an obstruction is not readily dislodged the larva creeps forward and brings its mandibles and labrum into play. AQUATIC MIDGES AND SOME RELATED INSECTS. 13 METHOD OF SPINNING SILK. The method by which Chironomus lobiferus larvae spin or spread out the silk used in the construction of their burrows and in the formation of the little conical nets mentioned above is very simple. The anterior pair of prolegs is the chief implement employed and so far as can be observed the only part of the body used for this purpose. The structure of these appendages takes on a new significance when function is suggested, and we at once notice the difference in structure between the anterior and posterior prolegs. The chitinous claws of the posterior pair are widened at their base (fig. 33), are few in number, and are arranged around the front and lateral margins of the prolegs (fig. 15). The muscles of the prolegs are so arranged as to set these hooks into the silken lining of the burrow, and thus hold the larva firmly in place. The hooks point outward and are so attached that by the contraction of the muscles of the proleg they are all brought close together in the center. When extended, the hooks all move outward in different directions, with the result that the prolegs are hooked fast to the silk lining of the burrow. Their function is preeminently that of an attachment, and it is to this specialization of the posterior appendages that the anterior prolegs owe their greater freedom of movement. The anterior prolegs are often mistaken for a part of the head because of their position just posterior to the chitinized portion of the head proper. They commonly appear as a mass of bristles radiating in all directions. From the side they appear as one, because they are always moved together and are so completely covered by rela- tively long spines that it is hard to see how they are attached. A sagittal view shows them to be made up of two rounded lobes separated by a narrow depression. The spines are graded in length from mere tubercles in front to long narrow hooked and barbed spines in the centre and again decreasing in size on the posterior surface. Here the short spines have rather wide bases and the tips are deeply serrate and somewhat hooked (fig. 34). The spines are obviously arranged in rows which diverge somewhat from the mid line laterally (fig. 35). The spines located near the centre of the prolegs are the best developed and are probably the most used in silk spinning. They are curved backward and hooked at their tips. Near the end there are a number of barbs on both the anterior and posterior edges. They are flattened laterally and are capable of being condensed into a very compact mass by the contraction of the muscles of the proleg. The hooks at the end of the spines point backward, and all the long spines are hooked except a few of the very outer spines, which seem to be slimmer and more hairlike. The actual process of silk spinning is much more easily studied by observing the construction of the conical net mentioned above than in any other way. It is constructed out free from other substances and is consumed and replaced every 10 minutes night and day until the activities of the larva are slowed down by the approach of the pupal stage. The larva begins the spinning by extending its body well forward and making several fairly rapid passes with its anterior prolegs in various radial directions. These movements place the silk strands that form the attachment for the apex of the net. Then, withdrawing its body somewhat and attaching the silk to the place where these radiating strands fuse with each other, the larva retracts its body, drawing out a ribbon 14 BULLETIN OF THE BUREAU OF FISHERIES. of silk spread by the prolegs. During this retraction the prolegs are held pointing forward at an angle of about 45 degrees with the body, and their exact use can only be surmised, but from their position and the speed of the movements it seems possible that the semifluid silk is spread either by the short spines in front or what seems more probable by the carding effect of variously hooked and serrate spines located farther back on the prolegs. When the larva reaches the end of its backward movement, the prolegs are spread and rapidly touched to the silk lining of the burrow at two nearby points. Then the forward movement is carried out. In this movement the prolegs are extended slightly forward and are more or less spread out. When the end of this movement is reached, the thread is attached either by the contact of the head or the prolegs to this central point of attachment and the process repeated. It is impossible to tell whether the head takes part in the process of attachment or not, because both the head and the prolegs are so close together at this point. It would seem probable from the small size of the apex of the finished structure that at some point in its construction the head occupying such an advanced position would be the only possible part of the body that could accomplish the attachment of the fibers. It is obvious, however, that the head does not touch the wall in the process of attaching the silk at the rim of the net, for the head is held projecting straight out and the move- ments of the prolegs are unmistakable. The forward and backward movements of the body are accomplished largely through the instrumentality of the posterior prolegs. These are held attached to the silk, and the last three or four segments of the body are flexed on them as axes. On the forward movement the body is straightened and the prolegs extended forward ; on the backward stroke the prolegs point backward according to the degree with which the body is flexed. The silk net (figs. 26 and 27) is too long to be spun from one place by the simple flexing of the body. This means that it has to be spun in two sections. The over- lapping of the sections gives the appearance of a continuous sheet of silk extending from the apex to the base of the net, and the original posterior attachments of the first section appear as radiating strands from the sides of the net. The entire process of constructing the net requires less than half a minute and in- volves the spinning of 42 to 44 ribbons or sheets, as determined by counting the move- ments. When this process is completed and the larva turns about and begins forcing the water into the net, it can readily be made visible by adding a few drops of water con- taining powdered carmine. The method by which the silk lining of a burrow is spun is not so easy to observe as the process of spinning a net. It takes longer, and the number of movements is so great that it is almost impossible to correlate them with any definite structure later observed. But even here, if proof were lacking that the prolegs are the one necessary factor to explain the entire process, there are structures that bear unmistakable evi- dence of their use. The lining, as the silk net, is spun in sections which, while not of uniform length all the way around the tube, are nevertheless approximately so. The exact way in which the first section is constructed is not so easily understood, but from this on the process involves a considerable repetition of the method employed in the construction of the net. The body is extended and retracted in the process of attaching the sheets of silk to the first section, to each other, and to whatever support AQUATIC MIDGES AND SOME RELATED INSECTS. 1 5 there may be available. These silken sheets are held extended by the thin branching threads of silk (figs. 36 and 37). The whole aggregation of silk sheets and threads is held under tension by silk layers attached in a spiral position. The supporting threads are then originally used as attachment fibers to hold a section of the tube extended and are later pulled into a position nearly at right angles to the lining by the tension exerted by the addition of another section. In this way the lining or tube appears slung in the centre of a cavity with numerous threads radiating in various directions (fig. 26). SILK STRUCTURES. The completed silk lining shows relatively little structure as far as the tube itself is concerned, but the supporting lines thrown out when the larva fastens this lining between two parallel glass surfaces are quite interesting. In studying the structure the tube is seen to be of a fairly uniform diameter and to be composed of a thick layer of silk, which shows no definite layers or strands. At intervals the silk is pulled out into conical en- largements. At these points the tube is seen to be made up of more than one layer, for the lining continues straight on leaving a space. The lining is held extended in the form of a cylinder by very interesting branched threads. These threads are often more or less sheetlike next to the tube, but divide and subdivide toward their point of attach- ment where they are much more widely spread out than at their origin (fig. 37). These structures show unquestionably the use of the prolegs, for it is inconceivable that such fine threads often ending in more than one plane could have been attached in any other way. The structure of the conical net is not easily made out even under high powers of the microscope, but the addition of powdered carmine to the water passing through the net gives it such a uniform coat that the author is inclined to think the entire structure porous. At times Protozoa and other relatively large organisms are seen to be forced into one of these nets and to escape by a circuitous route, which would suggest a breach between ribbons or sheets of silk. In other cases relatively large gaps, opening directly through one side of the net, are indicated by the escape of particles. When the net is collapsed, as it always is when the larva is not forcing water through it (the condition always existing in stained material), none but the grosser structures are visible (fig. 27). The net, as explained above, is spun in sections, but the position of the threads attached to its sides, as well as the observed behavior of the larva, shows these sections to be less regular than one might infer from the previous description. The arrange- ment of the net in sections in a manner similar to that of the lining of the gallery suggests the possibility of narrow slits in its surface of the same nature as those in the attach- ments of the tube (fig. 36). It is probable, however, that the impact of the current is necessary to open them wide enough to allow water to pass through. The conical net is spun exceedingly thin, as one would expect from the frequence with which it is consumed and replaced. This is doubtless correlated with the speed of the movements involved in its construction. In fact, it seems reasonable to conclude that the nature of the silk rather than the psychology of the larva dictates the speed of its movements. The spinning of one part of the net upon another in such rapid 80285°— 22 3 l6 BULLETIN OF THE BUREAU OF FISHERIES. succession indicates that the silk hardens very quickly on contact with the water. The quickness with which the silk hardens determines the speed at which the larva must work in order to spin silk of a given thickness. Hence, the thinner the structure the greater the speed required, because of the greater surface relative to the volume exposed. Thus it appears that the very rapid movements of the larvae are dictated by considerations of economy in the silk used. RELATED FORMS. Goetghebuer (1911) in a special examination of the external structures of the larvae of the genus Chironomus established three groups, as follows: Group I. — Containing those species possessing two pairs of branching filaments on the eleventh segment; a thickened oval area on the labrum; an epipharyngeal comb composed of a row of regular teeth; the antennae with five segments; and the abdominal segments of the pupa without spinose protuberances. Group II. — Branchial filaments of the eleventh segment lacking; the median anterior piece of the labrum simple without the thickened oval area; the comb of the epipharynx not composed of a regular row of teeth; the antennae with five segments without Lauterborn's organs; and the pupa without spinose protuberances. • Group III. — Agrees with Group II except in the presence of small granulations on the labrum of the larva and the presence of spinose protuberances on the posterior abdominal segments of the pupa. The specimens upon which the last two groups were founded all live in the paren- chyma of submerged leaves of numerous aquatic plants and are as follows : Chironomus sparganii Kieffer, C. viridis Macquert, C. niverpennis Fabricus, C. tendens Fabricus, and C. dispar Meigen. The list of plants in which these larvae were found as given by Goetghebuer is as follows: Straiioies aloides, Sparganium ramosum, Butomus umbel- latus, and Alisma plantago. In addition to this list the author has bred Chironomus lobijerus Say, C. pedellus Deger and Tanytarsus obediens Johannsen, from Sparganium stems, and Needham (1908) reports Chironomus albistria Walker from Nymphaea stems. While, of course, only the bred specimens have actually been observed to build conical feeding nets, yet the similarity of their external structures and the nature of their habitat give a considerable justification for including them in this group, especially when it is observed that Tanytarsus obediens, a member of another genus, possesses this habit. A bit of information regarding the similarity of the larvae of Chironomus sparganii Kieffer is contained in a paper by Willem (1908). He finds the uniform punctations of the abdominal tergites, the posterior teeth of the lateral plate of the eighth segment, and especially the peculiar process carried by certain abdominal segments would suggest C. lobiferus Say, the description of which was found in Johannsen's monograph. He finds his most striking difference in the fact that Johannsen says that these processes occur on all the segments, while he finds them on segments two to six only. Upon examining his own material the author finds this to be also true for C. lobiferus Say, as well as C. sparganii. Dr. O. A. Johannsen has also observed the author's material and agrees with him in the identification of this species. While it is not known how great weight this observation had with Kieffer in establishing the species C. sparganii, AQUATIC MIDGES AND SOME RELATED INSECTS. 1 7 yet there is no doubt whatever that the two species will be found to resemble each other very closely, as it is difficult to find any satisfactory distinctions between them from their descriptions. Group II.— Tanytarsus pusio Meigen. For this group Tanytarsus pusio Meigen has been selected as a type, because Mundy (1909) has already studied it so completely that there is relatively little new material to be added. The only larva whose feeding habits he describes is the species given above, but he designates "Larva No. 1 " and apparently "Larva No. 18" as also feeding in a similar manner. "Larva No. 18," he says, "builds a still more elaborate case, composed of long stalks to which is attached a short tube with three long arms given off at the free end. The case is not quite so opaque as that of T. pusio and is of a light biown color." The author has bred T. exiguus from similar tubes and observed its habits, which resemble very closely those of T. pusio, as described by Mundy. CONSTRUCTION OF TUBE. The following description is taken from Mundy's work (1909). The latter part is condensed from a more complete description. The first thing the larva does is to gather a number of particles of mud together and form them into a short strap or band passing across the body and fixed to the dish on each side. Using this band as a starting point the larva sets about building a simple straight tube closely applied to the dish and open at both ends. At first the band is merely broadened so as to cover more of the body, but soon it is shortened as well until length and breadth change places and a real tube is formed. * *. Anchored, as it were, to the strap by its anal feet it rapidly sweeps through an angle of about 6o°, touching the surface here and there with its mouth as it passes. Then, firmly grasping a particle by means of the labial armament and the anterior appendages, it powerfully contracts its body, thus drawing the particles toward the centre of operations; but not only do the above mentioned particles move, but all those touched during the sweeping movement follow in its wake, having been united together by silk threads or mucus during the first action. In this way abundance of material is col- lected and the building of the case proceeds rapidly. According to Mundy (1909) Tanytarsus pusio and "Larva No. 1," which builds a stalk case, begin their tubes and construct them to a large extent exactly alike. When the tube of " No. 1" is 3 millimeters long, it begins to build it up horizontally, removing material from the opposite end of the old tube for this purpose. This is carried on until there is only a narrow stalk projecting up from one side of the original burrow supporting on its top end a short tube. This tube is later strengthened by the addition of saliva especially at the attachment. Then three arms are provided and the web attached. In strengthening an arm the larva twists its head right around it, describing thereby a complete circle, completing the forward and return movements with the greatest rapidity. [Mundy, 1909.] Tanytarsus pusio makes a dark-brown mud tube fastened together with saliva but not lined with a distinct silk lining. The tube is attached for a variable length to rock or moss stem in the bed of a river, but it grad- ually curves away from its support, so that the anterior end projects freely in the water. This end is the widest, from which it gradually tapers toward the base. 1 8 BULLETIN OF THE BUREAU OF FISHERIES. VARIATIONS IN TUBES. The tubes, as explained above, are composed of debris fastened securely together with silk. Taylor (1905) and Lauterborn (1905) have described other closely related larva?, living in similar situations, that spin tubes of nearly pure silk. In texture the tubes of Tanytarsus pusio Meigen and T. exiguus Johannsen are intermediate between those of pure silk, such as are spun by Chironomus lobiferus when living within a stem, and those composed of a mass of debris only loosely fastened together with silk, such as are characteristic of the group that is represented by C. cayugce. The tubes figured by Mundy for Tanytarsus pusio are rather different in structure and proportions from those of T. exiguus. The substance of the tubes gives them a gray, slatey appearance that closely resembles the general color of the bottom. The arms are proportionally stouter and are represented on the sides of the burrow by elevated ridges. The tubes are as often fastened flat down to the surface upon which they rest as elevated at the end, apparently depending upon the convexity of such surfaces. The arrangement of the tubes is not to any great extent dependent upon the direction of the current. Small stones having from 8 to 10 tubes on their undersides usually showed such a variety in the arrangement of those tubes that it would seem that free space was of more importance than the direction of the. current. Johannsen (1905) says of the tubes of Tanytarsus exiguus: During the early summer most of the cases will be found attached by the stems alone, but later in the season most of them lie flat on the rocks and are attached on one side like Simulium pupal cases. It seems evident that this species varies considerably in the type of tube which it builds. The author's observations on this species in nature are confined to small streams which were not very rapid, and in these localities the food supply has been fairly abundant, as shown by the number and variety of the population. In such habitats the predominance of the attached type of tube would seem to indicate that the strength of the current and perhaps the food supply are the governing factors. Since the writer's observations were made both in the fall and in the spring, the effect of seasonal changes should be eliminated unless these tubes were able to persist throughout the winter, which seems improbable in most cases considering the erosion to which such small streams are subjected. THE NET. The arms, as before mentioned, are connected by webs so as to form a net to retain all passing objects; but even with a high-power lens I have been unable to detect single threads. The network seems only to be made up of irregular bands of slime or mucus passing between the neighboring arms, so probably it issues from the creature's mouth in this form. [Mundy, 1909.] NET MAKING. To build its net the larva proceeds as follows: Running up one of the arms for some distance it swings across to the next arm, carrying with it a thread of silk, then quickly back again, at the same time retreating somewhat into its case. This zigzag movement is repeated two or three times until the base of the arms is reached, when the whole process may be repeated over again until a sufficient num- ber of threads have been stretched across to make a rude network which, whatever its workmanship compared with that of a spider, is at any rate good enough for its purpose and effectually stops all objects AQUATIC MIDGES AND SOME RELATED INSECTS 19 floating by. In the case of larva No. 1 this has only to be done twice, but in Tanytarsus pusio from four to seven times, according to the number of arms present. From time to time the larva pulls down the net between two arms, using the labrum and thoracic feet to collect the particles together into a compact mass, which may then be used for further building operations or may be pressed into the mouth to be consumed. [Mundy, 1909.] SILK SPINNING. This process has not been treated in any considerable detail by Mundy except that he rightly inferred that the silk was made up of bands of slime or mucus instead of threads. The author has followed the activities of Tanytarsus exiguus in its silk-spinning movements and finds it a very difficult species to observe in this particular. Its chief silk-spinning activities consist of the rapid movements of its head and anterior prolegs in such close proximity to the surface that in spite of the numerous repetitions of the same movement, while applying layer after layer of silk to the rim of its burrow, the author was unable to determine that the head did not play an equally important part in this process. It was more nearly possible to distinguish the use of the prolegs in the work of reinforcing the arms. Here the most characteristic movements were upward, in which movements the head was held somewhat away from the arm as the body encircled it. The most satisfactory movements in this process were those concerned in the con- struction of the web mentioned above. The specimen studied in this particular had a tube fastened to the bottom of a glass vessel. This tube had two radiating arms on which the larva spun a single thread. This web was attached to the glass as far out as the larva could reach, then to the nearest arm, and from this arm to the second and down to the glass again on the opposite side. In this process the larva in swinging from one arm to the other repeatedly struck the end with its prolegs while its head projected well beyond. The silk is especially viscous, and the particles swept against it by the current readily stick fast. By this means the single thread spun by Tanytarsus exiguus was very eflective in catching particles. At intervals this thread was pulled down and consumed, the labrum and maxillae playing the important part in the process. The prolegs were not seen to be employed in the pulling down or rather pulling in of this single thread, as stated by Mundy. ADAPTABILITY. The Tanytarsus pusio larvae were taken from flowing water and placed in dishes containing only about a quarter of an inch of water with relatively few fatalities, con- sidering the crude methods employed in removing their burrows. The dead larvae were removed and a small amount of organic debris added. This the larvae raked together in masses near the ends of their burrows and consumed in what seemed tremendous amounts for such small larvae. They simply placed their heads against one side of a mass, and by the motion of the appendages of the head alone the food was passed down their throats in a steady stream. It was apparently fastened together by the silk spun during the process of collecting it together. In this connection Mundy's observation, 20 BULLETIN OF THE BUREAU OF FISHERIES. quoted above, on the method of collecting particles in the construction of their tubes seemed to be related phenomena. Perhaps the most remarkable change in the behavior of these larvae was that exhib- ited by a specimen which, after living a week in quiet water, suddenly found its food swept out of reach by a current. This larva in less than 1 5 minutes raked away a part of the rim that it had spun between the radiating arms and after reinforcing these out- rakers spun a web upon them. The current was set up by a pipette operated by hand, and gave a very satisfactory means of testing the reactions of the larva, for the strength and direction of the current could be changed at will. A complete reversal in the direc- tion of the current seemed to alter the behavior of the larva not at all in regard to its web or any other observed activity. It would seem feasible to demonstrate the behavior of this larva in such a special lantern slide as recommended for this purpose with Chironomus lobijerus. The larvae, while sensitive to a jar, do not seem to notice the light particularly, and the adhesive nature of the silk makes it possible to use powdered carmine or India ink to make the strands visible. The current recommended to keep the temperature down could be adjusted to answer for the natural flow of a stream. Group III. — Chironomus cayugae Johannsen. This group is based upon a recently described species which, so far as can be judged from direct observation as well as indirect references, will prove to be one of the most widely distributed species of the family. It is on this account, as well as upon its unique habit of living in watering troughs where it is easily accessible, that Chironomus cayugcB Johannsen has been selected for the purposes of this study. It is a type of a very large group which is included in Goetghebuer's first group and characterized by the presence of two pairs of branchial filaments on the eleventh segment, a thickened oval area on the labrum, an epipharyngeal comb composed of a row of regular teeth, antennae with five segments, and the pupa without spinose protuberances on the abdom- inal segments. The division includes most of the bigger red chironomid larva? and is probably of greater economic importance than any other group in the family. At this point it may be of interest to recall that most, if not all, of the species of Group I under stress of circumstances adopt the habitat and behavior of this group. HABITAT. These larvae are fitted by their extra branchial filaments and red blood for life in the debris at the bottom of lakes, ponds, and stagnant pools. The larva? of the species selected as the type, while living in various other habitats, are especially common in horse troughs, having been taken by the author from troughs in Orrington, Me. ; Woods Hole, Mass.; Ithaca, N. Y. ; Dayton, Ohio; Greencastle, Ind. ; Evanston, 111.; and Mil- waukee, Wis. The troughs most carefully studied are those in Woods Hole, Mass., Ithaca, N. Y., and Greencastle, Ind. In none of these troughs was the author able to find any other species belonging to this group represented, and it seems that by some special adaptation this species has succeeded in adjusting itself to conditions different from those common to the group. It is also found associated with Chironomus decorus and others in the debris at the bottoms of larger bodies of water, and it is obvious that it AQUATIC MIDGES AND SOME RELATED INSECTS. 21 is not only capable of living in the same conditions as they, but, as can be shown by a simple experiment, both are able to live on the debris found in a watering trough. It seems possible, then, that the difference may consist in such a simple adaptation as in the manner of depositing eggs. Needham (1906) has referred to the habit of Chironomus annularis Degeer of extruding its eggs while in flight and depositing them free in the water. Mundy (1909) has referred to the fact that T any tarsus pusio eggs were found attached to leaves several centimeters below the surface. J. T. Lloyd informs the author that he has observed masses of chironomid eggs of considerable extent blown upon the shore of Cayuga Lake. The author has observed Chironomus hyperboreus Staeger depositing its eggs upon the surface. Some of these females were caught in flight and found to have a considerable mass of eggs ready to be deposited. On the other hand, Chironomus cayugce females were observed just at dusk to light upon small stones which projected slightly above the water level and to thrust the tips of their abdomens beneath the surface of the water and there deposit egg masses attached to the stones. All the eggs taken from troughs have been found attached, and it seems possible that by this habit alone C. cayugce may be especially adapted to such a singular habitat. It is interesting to note that troughs fed from flowing streams where a considerable amount of silt is constantly present have in every case been found to contain but few or no larvae, and it seems probable that the choking out or covering up of the food supply is the controlling factor. Other members of this group are found in streams, ponds, reservoirs, and lakes, even at very great depths, as in Lac Leman. Here Mile. Zebrowska (19 14) found specimens designated as Chironomus "B" abundant to a depth of 20 meters and rare to the extreme depth of 100. THE BURROW. The process of building a burrow has probably been observed in this group more frequently than in any other, because the larvae when out of their burrows are very restless and at once begin to rake particles together. The pectinate hairs and comb of the labrum and the epipharyngeal comb are used in this work. The anterior prolegs usually form the limit of the backward stroke of the head, and it is difficult to say for certain that they remove the accumulated debris; but it is clear that this debris is fastened together with silk, and it seems possible that they may be instrumental in spreading it. When a certain amount of debris is accumulated, it is raked back by a looping of the body, so that the posterior prolegs hook into the silk that holds the particles together. When a sufficient amount has been so accumulated, the larva seizes the mass adhering to its posterior prolegs by means of its head and anterior prolegs and fastens it over the posterior end of the body in such a manner as to form a narrow band or strap, which is referred to in Group II of Mundy's description. This narrow band has the dimensions of a cross section of a burrow, and with this as a beginning the construction work con- sists of a direct application of building material to either side of the strap. From this stage on the behavior is the same as that observed in the ordinary lengthening of the burrow. The larva now reaches out and grasps by means of its labrum, mandibles, and prolegs a mass of debris and draws it in and puts it in place at the edge of the burrow. 22 BULLETIN OF THE BUREAU OF FISHERIES. Then siik is spun by the obvious use of the prolegs, as in the case of Chironomus lobiferus. Each addition of debris is fastened in place by silk which is attached to the older parts of the burrow and spun out and part way around this material. In this way the larva? construct long tubes that give them protection from enemies and at the same time help support them on the surface of the soft debris where they are usually found. The tubes are often U-shaped, and thus serve to bring in fresh water from which the larvae are able to carry on their respiration while living among decaying organisms at the bottom. FEEDING HABITS. The author has found it exceedingly difficult to satisfy himself that the members of this group are not really similar in habit to those included in Group I, but repeated experiment has convinced him that their habits are distinct. When a large number of these larvae are scraped up together with a mass of the surrounding debris and then spread out in a shallow dish, they literally spin every bit of the debris into loose inter- woven U-shaped burrows. When one tests the current in these burrows, the water is found to be flowing through them in a definite direction. Methods have been repeatedly tried to get these larvae to adapt themselves to glass tubes of the sort used so successfully with Chironomus lobiferus, but in no case have these experiments succeeded except when sufficient debris was present to make it possible for the larva to completely conceal itself. Several larvae were put upon pure sand with the hope that it would furnish pro- tection, if that was what was desired, and at the same time fail to serve as food. The larvae were obviously not well satisfied with their surroundings and moved about over the surface apparently in search of more suitable conditions. In removing the larvae from their burrows a small piece of the organic debris of which their tubes are character- istically composed was left adhering to one of them. This the larva kept clinging to and trying to roll up into a burrow. The other larvae as soon as they encountered this debris attempted to get possession of it; after a few hours they all made burrows out of sand. The current was tried by means of powdered carmine but without satisfactory results. In another experiment several larvae which had well-constructed tubes were removed, tubes and all, to a flat dish. These larvae were keeping a strong current of water flowing through their burrows. After being removed they were placed in shallow water. The tubes were well separated from each other, and the bottom was lightly sprinkled with loose debris similar to that from which the tubes were constructed. After a few hours the larvae ceased to maintain so strong a current and in most cases maintained it only spasmodically. The debris sprinkled over the surface was not disturbed even after being left over night. The current was repeatedly tested and found to be insufficient to furnish any considerable amount of food. The tubes were then dissected under a microscope and their inner surfaces were found to be eaten full of rounded holes and enlarged in places. From this it was de- termined that the larva ate away the substance of its burrow from within. While the larva is in such a tube it would probably not be possible to maintain a sufficient current through the burrow to bring in much food on account of the number of openings through its wall. It seems possible, nevertheless, that the current would at times bring in and AQUATIC MIDGES AND SOME RELATED INSECTS. 23 deposit substances that could be used as food. This seems especially possible, because the larvae of Group I are known to eat and replace certain parts of the silk composing the wall of their burrows at irregular intervals. The larvae, on the other hand, are known to have the habit of scraping up substances to be eaten directly as food. The author has observed this behavior in the case of well- soaked pieces of cracked corn. These the larvae seemed to have eaten exclusively, for their stomachs were full of the starch grains. The larvae frequently reached out for some distance and unless the fragments were easily moved did not seem to attempt to drag them in. Once a piece of corn was found it was usually eaten out until nothing but the hull remained. An examination of the debris at the bottom of the Greencastle (Ind.) troughs showed the greatest number of larvae per square foot yet found, which by count of a smaller area was estimated to be 500 to the foot. Here an abundance of diatoms, in- terspersed with corn and oats brought in the mouths of the horses from a near-by livery stable, formed a layer about an inch and a half in depth. The flowing water and the undulating motion of the larvae kept the conditions suitable to favor the development of diatoms, as was indicated by the great abundance of a relatively few species. The presence of a considerable amount of horse champings did not seem to upset the balance, as a too liberal addition of corn has been found to do in laboratory cultures. Miss Tilbury (19 13) found it possible to rear the larvae of this species from egg to adult on Potamogeion crispus alone. This she grated up and fed to them in small amounts. It will be seen from the above observations that Chironomus cayugcB is well suited to experimental culture methods, and it seems probable that the group as a whole is equally hardy. Their large size and overlapping broods offer considerable encourage- ment to the hope that they may sometime be an important factor in fish culture. Group IV. — Chironomus braseniae. n. sp., a leaf-eating chironomid. INTRODUCTION. While on a lymnological trip to North Spencer, N. Y., the author's attention was called by Dr. Needham to the work of an insect larva that was cutting burrows in the floating leaves of the water shield, Brasenia schreberi, and to a lesser extent in the leaves of the sweet-scented water lily, Castalia odorata. The larvae were found to be those of a midge of the genus Chironomus and apparently an undescribed species, al- though this or a species with similar habits seems to have been observed by workers in several different parts of the country. Mr. Isley, of the U. S. Bureau of Entomology, informs the writer that he has seen larvae of this genus with similar habits in the vicinity of Washington, D. C. Dr. R. H. Pettit has referred to a species which he bred from the leaves of both Nuphar advena and Nymphea odorata in the Wild Gardens, Forest Hill, Mass. He also observed this same species at Pine Lake, Ingham County, Mich. The author has seen specimens from Fair Haven and North Spencer, N. Y. Dr. Pettit's note on an undescribed species published in the first report of the Michigan Academy of Science (1900) is the only reference the writer has found in the literature, however, to a species of Chironomus with similar habits, although the closely related genus of Cricatopus, according to a brjef note by C. W. Johnson published in the Entomological News (vol. 80285°— 22 1 24 BULLETIN OF THE BUREAU OF FISHERIES. 12, p. 30) apparently contains a leaf-eating species, as Mr. Johnson states that Crica- iopus sylvestris was bred by Prof. Smith from the leaves of Victoria regia. These two species, so far as the author knows, are the only ones that feed directly on living plant tissue. Other larvse, however, are found in the large air spaces of dead and living aquatic plants, where they maintain themselves in the same way as while living in a burrow made of trash. Dr. Pettit's note gives only a brief explanation of the nature of the damage done and a general description of the larva and pupa based on the color characters. He bred the adults and states that they belong to the genus Chironomus and are probably a new species. Since the above quoted studies are substantially in agreement with the writer's observations it would seem probable that all the above chironomid records refer to this species. The unusual food habits and other unique adaptations seem to justify a rather more comprehensive study of this species. GENERAL HABITS. The larva? of Chironomus brasenia; from a superficial examination would appear to be true leaf miners. The straight or winding galleries are covered with a green ridge, which closely resembles the epidermis in color, and the lower epidermis is left intact. Closer examination, however, shows that the cover of the larval burrow is not the upper epidermis but rather an artificial cover made up of plant fragments fastened together with silk and moistened with a film of water which floods the entire burrow and spreads out in a thin film a little way on each side of the burrow. The larvae can also be seen when at work to actually project their body out onto the surface of the leaf at times. They are completely immersed in the water which constantly floods their burrows, and they breathe by blood gills. Their burrows are lined with silk, which is also used in constructing the coverings of the burrows. The larvae of the typical rep- resentatives of this genus do not carry their cases about with them as was erroneously stated in his discussion of this species by Dr. Pettit. On the contrary, this larva resembles the other members of the genus except in the method of obtaining and the nature of its food. LIFE HISTORY. The life history differs in several important particulars from that of the typical species of the genus Chironomus. The eggs are laid on the surface of partially submerged . leaves of both the water shield and the sweet-scented pond lily (fig. 19). They are laid in strings which tend to show a double arrangement of the eggs, due doubtless to an egg coming from each ovary simultaneously. These egg strings are wound about and crisscrossed in such a way that they form a somewhat disklike mass which tends to be only one layer deep, the gelatinous coating fusing to unite the whole into a single mass. In the limited area where the eggs were found most abundantly the leaves of Castalia were selected rather more frequently than those of Brasenia, although the latter is clearly preferred by the larva;. It seems probable that the chief factor governing the selection of these leaves is their partial submergence, as the eggs are laid on the top surface of the leaf and are unable to endure desiccation. The young larvae obtained from these eggs were placed on sections of Brasenia leaves and confined there in drops of water. These preparations were placed in watch glasses AQUATIC MIDGES AND SOME RELATED INSECTS. 25 and the normal conditions maintained for several days. Careful observations showed no signs of the larvae having begun burrows. They were then allowed to enter the water beneath the leaf where they lived for several weeks, but failed to develop, even though a miscellaneous supply of aquatic organisms was furnished. Several of the larvae died and the experiment was abandoned after two months' observation. Just how the very young larvae maintain themselves is still undetermined. A thorough search in the early spring before the leaves of Castalia or Brasenia had reached the surface failed to reveal the presence of any of the larvae on the roots, stems, or leaves of these, their characteristic host plants, or of any of the other near-by aquatic vegetation. From the facts that none of the small larvae succeeded in penetrating the leaves under laboratory conditions, that only large larvae were found in all the burrows opened, and that the burrows are of a uniform width in practically all cases, it seems probable that the larvae do not enter the burrows before mid-larval life (figs. 20, 21, 22). The author has been unable to make any direct observations on the length of time spent in feeding on the leaves, as the larvae are not to be found in any of the apparently similar aquatic situations about Ithaca and when brought into the laboratory on leaves frequently leave their old burrows and start new ones. This confusion, together with the writer's inability to rear the young under laboratory conditions, forced him to use indirect means in determining the length of time spent in feeding on the leaf tissue. Late in the season a dozen leaves containing active larvae were each labeled by pinning a square piece of paper so that a marked corner came opposite the end of the burrow. The label carried the number of the larva, and the rate of progress was measured daily. These results showed that not more than 10 days on the average would be required to construct a burrow of average length, while the larva that made the greatest progress would not have required more than seven days. Subsequent examination of these larvae showed 100 per cent infested with a Gordian worm, so the results are doubtless inaccurate, although all larvae that ceased burrowing after a day or two were omitted in making up the average. The larva transforms to a pupa on the leaf where it has been feeding. The pupal chamber can often be seen at the end of a burrow 2 inches in length on the leaves of Brasenia and somewhat shorter on the leaves of Castalia. A burrow of this length represents the work as a rule of a single individual. The pupal chamber can be easily picked out because of its club-shaped appearance (fig. 24), the big portion being at the end of the burrow. The pupal chamber is, however, often completely separated from the ordinary burrow, suggesting a vagrant tendency on the part of the larva just before pupation. When separated from the larval burrow it has the same general shape and appearance as when attached. It is but little longer than the pupa and is in rough agreement with it in general outline. The pupa lies in this burrow with the head next to the large open end. When ready to transform it wriggles through this opening and the imago escapes. The pupal molt is usually ft with the thoracic part projecting from the pupal chamber (fig. 24) . The length of lme spent as a pupa is about five days, varying considerably with the temperature and the condition of the individual pupa and the larva from which it transforms. The adults, both males and females, were found among the bushes and underbrush along the banks, and several were shaken from the tops of trees 6 to 8 feet high. The 26 BULLETIN OF THE BUREAU OP FISHERIES. leaves found with egg masses on them were near these trees and were shaded during a part of the day. Mating and egg laying probably takes place in the early evening, as is the habit of the family. When infested leaves are brought into the laboratory, the adults begin to emerge in a day or so and continue to transform a few at a time until they have all completed their development. This lack of uniformity in the time of transforming is character- istic of the Chironomidae and is an important factor in their adaptation to the dominant position that they hold in the life of the fresh waters. In the case of Chironomus bra- senice this adaptation makes it a worse pest. Dr. Pettit bred his first specimens about the middle of May while at Forest Hill, Mass., and in his note before the Academy of Science he states that a second brood was seen on August i at Pine Lake, Ingham County, Mich. It seems probable that the term "brood" is here rather loosely applied. PENETRATING THE EPIDERMIS. The author has given evidence above to show that the leaf-mining method of feed- ing seems not to be adopted before mid-larval life, and hence is doubtless a less primi- tive habit than that of the young larvae. That this method of feeding is probably impossible for the young larvae seems borne out by observations made upon the behavior of the half-grown larvae in penetrat- ing the epidermis of aquatic leaves. In attempting to induce the larvae to start new burrows the writer removed them from their old ones and placed them on leaves where there were no unoccupied burrows. This work was for the most part rewarded only by observing the larvae searching vainly for their original burrows. If by chance one encountered the burrow of another larva it crept boldly in, only to be met by the owner, who usually administered a sharp nip by means of its mandibles. Since the members of the genus are chiefly herbivorous and therefore for the most part peaceably inclined, the intruder usually retires quickly. It often repeats its attempt to enter the same burrow several times in succession, each time more cautiously, until it finally gives up or in some cases enters an unused part of the burrow and begins feeding. In this case it extends its burrow as a continuation of the original burrow or as a side branch, show- ing that it finds a decided advantage in using a burrow already started. In case the larva gives up the attempt to enter an already formed burrow it begins a new one. The first requisite is the spinning of an arch formed of many thicknesses of silk about the size of an ordinary burrow. This, from the author's laboratory observations, is preferably located near another burrow, perhaps because of the water film that always accompanies the burrow. The larva next enters this silken arch, which is about as long as wide, turns itself on its back, and bends its head backward. This position enables the larva to brace its anterior prolegs against the underside of the silken arch and so bring pressure upon its head, which is at such an angle to the surface that the mandibles and the labrum are in contact with the surface of the leaf. In this position, with the head bent backwards a little more than at right angles, it extends its mandibles and rasps its way slowly through the epidermis. The spinning of the arch and the penetrating of the epidermis take over an hour and are doubtless severe tests on the strength of so small a larva. AQUATIC MIDGES AND SOME RELATED INSECTS. 27 Chironomids labor under greater handicaps than other gnawing larvse in not having strong legs provided with claws for holding them in position and in not having mandibles that are opposable for cutting the tissue. Instead they have to depend upon indirect methods, of applying pressure to their mouth parts, and the utilization of their labial plates in conjunction with their mandibles for cutting the tissue. THE BURROW. The method of penetrating the epidermis in the beginning of the burrow is the same for the leaves of Brasenia schreberi as for Casialia odorata, but from this point on it differs markedly, due to differences in the texture of the plant. That chironomids show adaptability in their feeding habits is well shown by the differences in their bur- rows in B. schreberi and C. odorata, resulting from a difference in the thickness and texture of the leaves of these two plants. We will take up first the nature of the burrow made on the leaves of Brasenia, since they are thinner, more easily penetrated, and where equally available more seri- ously attacked. This shows an evident selective power on the part of the larva. The writer is, however, aware that the softness of texture may be the deciding factor. When the larva has penetrated the epidermis of a leaf, it is able to bring pressure to bear more directly on the labial plate. The mandibles are hooked under the edge, and they, together with the pressure derived from the anterior prolegs, readily force the labial plate down through the epidermis. Then the larva moves a little to one side and repeats the operation. In this way the epidermis and parenchymatous tissue are removed from under the silken arch. Then the larva commences in the mid line and makes a cut as explained above from the center of the burrow to the outside edge. The larva during this operation is inverted, with its head turned backward. It next assumes an upright position, grasps the strip near its free end between its mandibles and labial plate, and pulls the strip backward, raising it upward at the same time. Then by bending its body to one side in its burrow it gets under the loose end and scrapes it clean of the green parenchymatous cells that adhere to it and fastens it in place against the silken arch (fig. 23). It next rakes this exposed area free of all the parenchymatous cells down to the lower epidermis. This removal of tissue usually results in the admission of water, probably through the mucous gland, as no openings are visible; at any rate the burrow becomes flooded and capillarity keeps it wet both inside and out. The larva continues to cut slabs and to extend its burrow as long as it cares for food. These slabs are twisted backward and fastened in an upright position and their tips bound together with silk secreted by the salivary glands. The bottom as well as the sides and top are lined with silk. In the leaves of Brasenia the bottom and sides of the burrow have a very thin layer of silk which is closely applied to the surrounding tissue. When castings are to be extruded, the larva turns about in its burrow and the partially digested material is fastened to the arched top of the burrow by silk in such a way as to serve as sort of a porch. It is held extended by silk threads which are fas- tened out on the surface of the leaf. This porch or canopy serves a threefold purpose, being a shelter from the sun, a means of retaining a film of water over the area that is being excavated, and an entanglement in which the free end of the slab of epidermis becomes lodged so as to be held up while the tissue is being removed from its underside. 28 BULLETIN OF THE BUREAU OF FISHERIES. Later it becomes a part of the roof. Figure 23 is a diagram representing the nature of the burrow and an area with the top removed to show the general appearance of the lower epidermis, which forms the floor of the burrow. In the case of the leaves of Castalia odorata the larva is obliged to bite the epidermis to pieces and remove it by sections because of its thickness. The head is applied at different angles and pieces of varying sizes are removed. Those that are small enough are swallowed and the rest are used in the construction of the sides and top of the burrow. The large spines are also removed and woven into the burrow with silk. On account of their relatively large size they greatly tax the strength of the larva. The thickness of these leaves is considerably greater than that of Brasenia leaves and the work of excavating greater. This necessitates a much greater thickness in the layer of silk making up the bottom of the burrow, especially as its bottom is at a higher level than the lower epidermis of the leaf. The castings are utilized in the same way as explained above for Brasenia, and the parenchymatous cells are eaten as food. The above changes in the method of procedure show a marked contrast to the habits of many other insect larvae. This ability to adapt themselves to a variety of conditions has doubtless been an important factor in the adoption of their present unusual feeding habits. The pupal burrow is essentially similar to the larval burrow in structure, but is made up more largely of silk, is larger in diameter, and persists longer than the larval burrow (fig. 24). RESPIRATION. The larva breathes by means of four blood gills located on the posterior part of the last segment. These gills are longer and more pointed than in the species having red blood (fig. 15). The water in the burrow while small in amount is kept in circulation by an undulating motion of the body during the intervals while the larva is not feeding. The current flows from the head backward over the gills and out through the chinks in the sides of the burrow, passes forward over the surface of the leaf in thin films on either side, and again enters the open end of the burrow. The water should be well aerated, since it is exposed in thin films both to the air and to the surface of the leaf while flowing forward outside of the burrow. It is also exposed to favorable conditions for the desired exchange of gases while within the burrow, as it comes in contact with the air in the air-containing spaces of the parenchyma which is rich in oxygen and poor in carbon dioxide. That the oxygen supply is rich seems to be demonstrated by repeated accidental experiments where leaves were submerged overnight with the result that the larvae died in the submerged leaves but lived in those on the surface. The pupa is active and continues to aerate its burrow by occasional undulations of its abdomen. The repiratory filaments consist of several much-branched tufts located on each side of the thorax. FEEDING HABITS. The larvae feed intermittently. They find an abundant food supply at hand, and the only limit set them is the rate at which digestion can be carried on. They are yel- lowish white in color, their blood lacking the haemoglobin which gives the characteristic color to the other chironomid larvae which are known as bloodworms. The green food material can be readily seen through the translucent body. It is seldom that the AQUATIC MIDGES AND SOME RELATED INSECTS. 29 stomach is allowed to become more than half empty, and often the larva resumes its feeding operations when the stomach is practically full. An attempt was made to determine the length of time that the food remained in the alimentary canal, but the small size of the larvse and the acute discrimination in their feeding habits prevented the use of any coloring substances to mark any portion of their food. Direct observation was resorted to but proved too tedious to afford accurate data. The larvae often withdraw from the end of their burrows and remain almost motionless for an hour or so at a time. Then they will begin feeding again, working for a half or three quarters of an hour at a time with only occasional short intermissions for the purpose of renewing their air supply by setting the water in circulation. By noting the intervals between feeding and resting it seems doubtful if the food remains in the body for more than two hours. There is no special masticating apparatus present, and the result is that a very large per cent of the parenchymatous cells swallowed pass through the body entirely unaltered. The use of the castings for roofing material in connection with the burrow places these cells in a position which, while artificial, nevertheless offers conditions under which the carrying on of their life processes should be partially possible. These cells are held suspended in a silken mesh, bathed in water rich in the mineral salts, resulting from the digestion of similar cells, and favorably placed for the obtaining of carbon dioxide. That the covering of a larval burrow remains green for a considerable time is readily observed, and it seems possible that the larvae have in this matter hit upon a favorable adaptation. ECONOMIC IMPORTANCE. The aquatic conditions of life required by Chironomus brasenia; larva? confine their attacks to leaves at or beneath the surface where their burrows may be flooded with water. This requirement limits their attacks to a restricted variety of plants. The writer's observations on the injury done aquatic plants by C. brasenice are confined to the one place where they occur within a reasonable distance of Ithaca, N. Y., which is Spencer Lake. Here the conditions seem to be excellently adapted to the growth of aquatic plants. The lake is shallow and Brasenia schrebcri is the dominant plant with floating leaves, while Castalia ordorata is present in various parts of the lake and is next in abundance. Observations made on July 22 show a very considerable proportion of the leaves of Brasenia infested, while only one or two doubtful cases of the infestation of Castalia were observed. On October 7 the entire pond was examined, and a leaf of Brasenia which had not been injured by this larva was so rare as to make it difficult to explain how it escaped. The leaves of Castalia showed a greater percentage of infestation later in the season than at the time of the author's earlier visit, but they were not badly injured. Dr. Pettit says of the damage to water lilies: The pads of both Nuphar advena and of Nymphea odorata were furrowed by some miner. The pads had been badly eaten in some places and many contained living larvae and pupae. CONTROL. The injury done by these larvae in parks and private gardens may some time become so great that methods of control will be necessary. At first thought it would not seem feasible to spray for an aquatic larva, but, as shown above, the water is kept circulating 3° BULLETIN OF THE BUREAU OF FISHERIES. through the burrow and out on the surface of the leaf again. This use of the same water over and over, except as it is removed by evaporation and replaced by a fresh supply drawn in by capillarity through minute openings, prevents the dilution of any poison that may be added. Any arsenical spray should be effective. There is, however, an important difficulty encountered by the lack of uniformity in the rate of development of the larvae. They are present in increasing numbers from the first of the season to late in September and infestation is taking place constantly. Where feasible the larvae may be destroyed in the early part of the season by draining the pond and allowing the bottom to become dry for a few days. The larvae are unable to breathe unless immersed in water and are, therefore, easily destroyed by a relatively short period of drying. In small pools the mechanical removal and destruction of eggs and larvae should be effective. DESCRIPTION OF CHIRONOMUS BRASENLE, N. SP. Larva. — Light green in color, the chitinized areas such as the head and claws reddish brown; antennae slender, about three-quarters as long as the mandibles, the basal joint four-ninths of the whole length; a small spine on the apex of the basal joint and another at the apex of the second joint probably represent Lauterborn's organ. Each eye consists of two black spots in such close contact as to appear as one on superficial examination. The labrum much narrowed anteriorally, with a few setae and four pec- tinate hairs. The epipharynx with three blunt teeth on its anterior border, the usual chitinized horseshoe area laterally compressed with the usual pectinate setae, a pos- teriorly projecting median process and the two lateral arms articulate with the posterior margin of this area. The lateral arms also have dorsally projecting portions for the attachment of muscles. They are furnished with a median projecting membraneous flap. Maxilla with short palpus, several setae, and two mesad projecting lobes. Man- dibles with blackened teeth, the two median and outermost teeth not much blackened. Labrum with blunt-pointed margin, the teeth with rounded outline. Posterior prolegs with bilobed claws. Anal blood gills long and somewhat pointed. The posterior dor- sal tufts of setae are each placed upon a papilla, which is about as broad as long (fig. 15). There is also a pair of setae just dorsal to the anal gills. Length, 7 millimeters. Pupa. — Light green in color, the chitinized parts somewhat infuscated. Respiratory organs consist of a pair of tufts of white filaments. Dorsal surface of the second to sixth abdominal segments with a well-developed anterior band of brown setae, the second and third segments with a posterior row of coarse spines, the entire surface covered with minute setae, which are slightly smaller on a few irregularly placed areas, thus giving the surface a slightly mottled appearance. The lateral fin of the eighth segment with the usual set of four filaments and a brownish slightly toothed chitinized portion seen best in the pupal molt (fig. 16). The caudal fin has the usual fringe of filaments. Length, 5 millimeters. Male. — Head, proboscis, palpi, and basal joint of antennae yellow, tubercle slightly developed, eyes black. Antennal shaft and verticils brown. Antennae with 14 joints, the terminal two-thirds as long as the rest of the antennae. Pronotum projecting lat- erally, but not reaching the level of the mesonotum dorsally. Mesonotum greenish yellow, translucent, somewhat pruinose; vittae of a light buff color; scutellum and AQUATIC MIDGES AND SOME RELATED INSECTS. 31 halteres yellow; metanotum and sternopleura buff colored. Wings white, longitudinal veins and cross veins not infuscated. Cubitus forking distinctly beyond the cross vein; third and fourth veins ending about equally distant from the apex of the wing. Abdomen light green, densely clothed with long yellow hairs. Segments without distinct fascia. Hypopygium as in figure 17. Legs whitish, fore tarsus not bearded, second and third joints densely bearded for their entire length. Tibial comb darkened on all legs; basal segments of fore tarsus more than one-half longer than the tibia, proportions as 47 : 30. Pulvilli well developed, empodium, narrow. Length, 4 millimeters. Female. — Antennae yellow, apical joint slightly infuscated, seven jointed; posterior margins of the abdominal segments with a narrow whitish fascia. Otherwise like the male. Length, 3.5 to 4 millimeters. Group V. — Trichocladius nitidellus Malloch. To this group belong those larvae which feed directly and apparently by preference on filaments of Spirogyra. Trichocladius nitidellus Malloch, a species described in 1915, is the only species that the author has thus far found which properly belongs here, although Lyonet (1832) described the habits of a species which obviously belongs to this group. Miall and Hammond (1900) state that this species has been rediscovered and studied by T. H. Taylor. They give in considerable detail the habits and behavior of this larva based on Taylor's observations. But it seems to the writer that there is a considerable difference between the two species and that the larva studied by Taylor agrees more nearly with that of Trichocladius nitidellus, which the author has studied, than with that ob- served by Lyonet. This difference will be more readily understood after the feeding habits of Trichocladius have been outlined. FEEDING HABITS. The description of the habits of this group has been well given by Miall and Ham- mond (1900, p. 11-17) based on Taylor's studies. Trichocladius nitidellus differs only in a few details. Filaments of Spirogyra are eaten exclusively by the older larvae. There seems to be some selective ability exercised in the choice of filaments when more than one species of Spirogyra is present. This selection favors the smaller filaments. The larva often seizes a filament near the middle and forces the loop down its throat two fibers abreast. The same thing is often done with the larger filaments, and only occa- sionally are they bitten off completely. In this respect the author's observations differ from those of Taylor, who says: "A filament of Spirogyra is seized by the mandibles and bitten in two." Taylor also states: "The labram beginning at one end of the filament draws it into the gullet by a stroking action." The labruin is not so well adapted to meet the require- ments of this method of feeding as the lateral arms which are located on the epipharynx. This will become more obvious when it is understood that the filaments are only crushed a little between the mandibles and possibly also between the labrum and the labium. This leaves the filaments in so natural a condition that when evacuated they immediately straighten out into their original shape. The stroking action of the labrum on a smooth filament, it would seem, is a far less effective method of forcing such a filament down into the stomach than the contact on either side by a well-developed pair of lateral arms 32 BULLETIN OF THE BUREAU OF FISHERIES. which can be swung through an especially wide angle due to their position on the under- side of the labrum. The movements of the lateral arms are so correlated with those of the labrum that the apparent stroking action observed by Taylor might well have been misinterpreted. The correctness of this line of reasoning has been confirmed by the author by direct observation with the low power of the compound microscope on a larva which was feeding with the ventral side up. Digestion in Trichocladius nitidellus as in Chironomus brasenice is incomplete, many cells appearing to be unaltered in the course of their passage through the body. THE BURROW. The burrow, as far as the author has been able to observe, is quite variable. In the older species it is made up of filaments of Spirogyra which have already passed through the body, as Taylor has also observed. In the case of the very young larva the burrow seems to be made up of dark-colored debris due to a different type of food eaten by the young larva. It is found to be made up of organic debris in which diatoms figure very largely. The writer is not sure at what stage the larvae begin to feed upon Spirogyra, but it is certain that specimens not more than one-fourth the size of the mature larva? do feed upon Spirogyra. These larvae have been observed to drag their burrow after them in a manner similar to that described by Lyonet. His observations when translated are approximately as follows : Its activity in transporting itself from one region to another is very great and its behavior is peculiar. It extends its head for this purpose, seizes in its teeth all objects which it encounters, retiring quickly without relaxing its hold. The claws of the anterior prolegs hook themselves into the object seized by the teeth, loosening them it elongates itself again in order to seize some more distant object and draw itself forward. Taylor does not seem to have noticed this habit, for he states under the heading of "locomotion": As the case is not fixed the larva can travel without leaving it. It does not creep like a caddis larva, but jerks itself forward by a few powerful undulations, in which the flexible case participates. The older larvae, however, according to repeated observations made on Trichocla- dius nitidellus both in the laboratory and out of doors, show so wide a range of behavior that it is impossible to confirm or disprove any of the above statements. Larvae of this species found living on the algae near the surface in a watering trough were placed in watch glasses and fed diatomaceous debris, which they ate and from which they con- structed tubes in no way different from those characteristic of Chironomus cayugce, as described above. Others, fed on a scant amount of Spirogyra, built no tube at all. About 30 larvae were found at the bottom of a small pool clinging to an old and partly decayed table leg. These had no tubes. It is not difficult to find all intermediate stages between the attached, the free tube, and the larvae without any tubes. The food supply seems to be a controlling factor in the nature of the tube built. Since the larva lives on Spirogyra by preference, it eats away all the filaments in its immediate vicinity unless they are very abundant and closely matted. In that case it selects out the filaments which it prefers and simply extends the case. When the food becomes scarce, the larva is able to feed upon whatever debris it finds available, and when feeding in this manner it constructs the tube characteristic of larvae feeding in this manner. AQUATIC MIDGES AND SOME RELATED INSECTS. 33 Taylor's observation upon the use of evacuated Spirogyra filaments in the construc- tion of the tube is quite correct for Trichocladius nitidellus, as is also his observation of the use of silk to fasten the fragments together. "Fibrous structure," which Taylor saw only "faintly," was not noticed. The larva carries on its respiration in the usual way, but since it habitually lives in a well-aerated environment it lacks haemoglobin in its blood and simply takes the color of the food contained in its stomach. Such larvae characteristically have much better- developed tracheal systems than those provided with haemoglobin. This fact, together with Taylor's drawing (Miall and Hammond, 1900, p. 15, fig. 8), which shows a well- developed tracheal system, tends to corroborate his identification to the genus Ortho- cladius to which all the species of Trichocladius were formerly referred. The available literature on the feeding habits of all the Orthocladius larvae is so limited that it is impossible to tell whether any considerable number feed upon different filamentous algae or not. It seems probable that the specimen observed by Taylor and assigned to this genus is either identical or closely related to Trichocladius nitidellus. Group VI. This group or subdivision of the Chironominae is erected to include several species known to live throughout the larval stage without building any semblance of a tube. Since there are several genera represented by these forms which have relatively little more in common than their free living manner of life, the author has divided them into subgroups based upon the structure of their mouth parts and the nature of their habitats. Group VI: Subgroup A. — Metriocnemus knabi Coquillett. These larvae are quite unique in their habitat, food, and manner of life. They were apparently first studied in the larval condition by Rnab (1905), who made a number of observations on the larvae and pupae. He found them in the pitchers of the pitcher plant (Sarracenia purpurea), and, so far as known, the larvae are not found in any other habitat. Although confined apparently to a single limited habitat, they are evidently widely distributed, for Knab found them at Westfield, Springfield, and Wilbraham, Mass., and at Cedar Lake, 111. The writer has found them at McLean, N. Y., and they are doubtless to be found wherever Sarracenia purpurea occurs. The larvae of the present species live at the bottom of the water-filled leafcups of Sarracenia pur. purea, burrowing in the closely packed debris composed of the fragments of decomposing insects. Evi- dently their food is from this source. [Knab, 1905.] HEAD STRUCTURES. The head structures for this subfamily have already been discussed, but this par- ticular species and doubtless the entire genus (for the larval stage of very few are known) show decided modifications in their mouth parts. These modifications are well adapted to the present mode of life of the larvae. The labrum has the usual epipharyngeal comb well, though not strongly developed. It also has the labral comb well developed, very wide, and finely toothed. The usual hooks, pectinate hairs, and spines are present, but somewhat modified. The pectinate hairs are here quite broad at their tip and furnished with a straight and uniformly 34 BULLETIN OF THE BUREAU OF FISHERIES. toothed margin. There are two pairs of these pectinate hairs, which, together with the other spines and processes, make quite a formidable and closely set array of scraping implements. The labium is, on the whole, not very different from the one figured for Chironomus brasenice except in the number and relative lengths of the teeth. In Metrioc- nemus knabi the teeth of the labium are of nearly uniform length. The maxillae are quite remarkably different. The inner mesad projecting portion has a number of long close set spines which lie in the same plane as the labium and doubtless are of great assistance in supplementing it and the hypopharynx in then- scraping action. The basal portion labeled for Chironomus brascnice is in Metriocnemus knabi narrower and less firmly united to the epicranial plate. The fan-shaped structure marked c (figs. 1 1 and 12) is entirely lacking in this species. The hypopharynx is long and well supplied with short, blunt processes. Its margin lacks the chitinized plates shown for Chironomus brascnice and is obviously specialized as a delicately sensitive scraping structure. Its general appearance is that of a soft and somewhat flexible tonguelike structure covered with processes that are doubtless tactile in function. The epipharynx is provided with the usual pair of lateral arms which are here similar but less strongly developed than is the case with Chironomus brasenus. FEEDING HABITS. Knab's remark that the larvae burrow among the fragments of decayed insects and evidently obtain their food from this source is true as far as it goes. It leaves one in doubt, however, as to the actual food of the larvae. Several times the author saw larvae with broad chitinous bands around their bodies, evidently segments of insects' legs. Other observations have shown the larvae with considerable parts of their bodies extend- ing into these narrow insect appendages where the larvae were apparently feeding. This would lead one to think that the larvae feed upon the decaying tissues found there. This is doubtless true, for they are quite adaptive in their habits; but from other observa- tions it would seem that they were, perhaps, even here feeding indirectlv on the insect structures by devouring the large numbers of bacteria that in turn break down the insect tissues. This conclusion has been reached after considerable experience with these larvae under artificial conditions. Larvae were removed from the pitcher plants and placed in petri dishes, together with the insect debris in which they were living. They were found to be perfectly well fitted to live in this manner. It was also found that they could live on beef broth, smoked beef, and decaying plant material. It is obvious, therefore, that insects are not the only source of food for these larvae. The question at once arises, Why is this species found so universally in the pitcher plant and nowhere else if it can live upon so wide a range of food ? The answer is obvi- ously given in the adaptation of this larva to a particular kind of food. In this adapta- tion the mouth parts are doubtless most fundamental. As explained above, they are fitted with a number of combs, spines, and fingerlike processes. The structure and length of the hypopharynx also indicate that it, too, is used as a delicate scraping organ, which is of prime importance in assembling the scattered bacteria. In this connection the presence of well-developed silk glands in close association with the hypopharynx suggests the possible function of their secretion in assembling the bacteria upon which the larva largely subsists. AQUATIC MIDGES AND SOME RELATED INSECTS. 35 That the larvae prefer bacteria to the more solid tissues can be observed by their behavior in a petri dish, where they move about in a very characteristic manner. The head is carried at an angle of about 450 to the bottom, and the anterior prolegs are the chief organs of locomotion. They alternate with the head in supporting the body and are provided with about three rows of strong coarse spines. The larva moves along by a rapid alternate depression of the head and backward stroke of the anterior prolegs. The posterior prolegs are little used, and the larva curves its body up first on one side and then on the other, thus aiding the head and anterior prolegs in their forward move- ment. The larvae always move forward, and the spines on the inner border of the max- illae and hypopharynx are doubtless of prime importance in collecting the fine organisms, in correlation with this progressive method of feeding. The pitcher plant seems to be the chief natural environment where such food sub- stances are available. While its inner surface is covered with closely placed spines which all point inward, it nevertheless offers a favorable environment to such small larvae, for they are able to move about among these spines and collect the bacteria and mold spores which accumulate there. That these larvae may find conditions at least imper- fectly suited to their method of feeding in other environments would seem to simplify the explanation of their distribution. The swamps are widely separated, and the pitcher plants are not numerous, and if a few larvae could live in other environments the distribution would be more readily accounted for. Group VI : Subgroup B. — Orthocladius sp. (?) These larvae have not been bred, but are abundant and will doubtless be found to be one of the common species. They are found in flowing streams about Ithaca and were collected the first part of June among the debris resulting from the disintegration of Cladophora. They were at that time very numerous, but the writer has been unable to find them on several occasions during the last of July and the first part of August. As forms found in flowing water are hard to rear, none of those taken early in the season were reared. LARVAL CHARACTERS. • The larvae are bluish green in color and have several rows of coarse black claws both on the anterior and posterior prolegs. The caudal filaments are placed on very small short papillae, and the filaments themselves are very short, scarcely extending beyond the anal gills. The head parts are rather stout (figs. 29 and 30), and the fan-shape mem- branes are entirely absent. The larvae apparently do not make use of their well-developed silk glands for the purpose of building tubes. It seems probable that they would not be so well developed unless they had some important function. The author has concluded therefore that their development is correlated with the nature of the food eaten. FEEDING HABITS. The larvae creep about on the surface of submerged stones and even out of the water where the rocks are only moist. A very noticeable feature in the behavior of these larvae is the frequency with which they turn over. This habit of rolling over every few 36 BULLETIN OF THE BUREAU OF FISHERIES. minutes while creeping over moist stones seems to be an adaptive measure. The upper portion of the body would become dry and the surface film would break away from the larva if it were not for the frequent moistening of the entire surface of the body. This revolving of the body about on its long axis the larva accomplishes by bending its head and the first two or three segments of its body off to one side, then, by relaxing the underneath muscles and stressing the upper muscles, this angle is made to revolve about the body until the larva again reaches an upright position. These larvae feed upon the organic debris to be found on the ledges in flowing water. Those whose stomach contents were studied had eaten Cladophora almost exclusively, and both the cell contents and the cell walls could be recognized. There were also a few diatoms, several filaments of Oscillatoria, and a few spherical cells, possibly of Aphano- capsa sp. The well-developed silk glands are doubtless used by the larvae as digestive glands, and their development may be due to the coarse organic nature of their food. Group VI: Subgroup C. — Prodiamesa sp. The genus Prodiamesa was established by Kieffer in 1906. In the "Genera Insectorum" he refers six European species to this genus, and the writer has found only one South American species which has subsequently been referred to it. There is therefore no record of the occurrence of the species of this genus in this country. The species upon which these observations are based has been bred, and, from the larval mouth parts figured by Kieffer and Thienemann (1908), it is found to resemble Prodiamesa prcecox very closely. The adult description, however, does not agree with that of P. prcecox, and it seems probable that it will be found to be a new species. The larvae are yellowish white with reddish-brown heads and brownish claws on the prolegs. They, as well as the larvae of the genus Diamesa, are characteristically found in flowing streams. The larvae of Prodiamesa were found burrowing through the coarse debris that had accumulated in a roadside watering trough which was fed by a rapidly flowing stream from a near-by hillside. The trough was nearly full of sticks, grass, and leaf mold in various stages of decay, and it was through this rather loosely aggregated material that the larvae were seeking their food. BODY STRUCTURES. This genus as represented by the writer's material (which is essentially in agree- ment with the larva of Prodiamesa prcecox) is structurally quite similar to the burrowing forms. This is especially true of the anterior prolegs which are made up of several rows of fine spines. The branchial gills of the eleventh segment are absent in this genus. The caudal filaments are well developed. MOUTH PARTS. The most characteristic modifications in the mouth parts are to be observed in connection with the maxilla, the hypopharynx, and another structure the homology of which has not yet been satisfactorily established. The maxilla is capable of a great deal of free movement, as the basal portion (as in the other species in this group) is not fused to the adjacent epicranial plate (fig. 31). Its inner mesad projecting process (fig. 31, t) is provided with numerous fingerlike AQUATIC MIDGES AND SOME RELATED INSECTS. 37 processes resembling those observed in Metriocnemus knabi. The hypopharynx also suggests that of M. knabi and doubtless shows a close generic relationship. The structures referred to above as not having their homologies well established are those to be found on the ventral side of the head (fig. 32, c), and suggest at once the fan-shaped membrane which the writer has labeled c in figure 11. Mundy (1909) has observed and mentioned these structures. He also gives a figure (PI. V, fig. 18) and states that it is his opinion that the long filaments which the author has shown in figure 32 doubtless fuse and form the fanlike membrane which is so characteristic of the genus Chironomus. The author has been unable to establish the connection between these structures and the articulation of the maxilla shown in figure 1 1 and is therefore doubtful about their identity. The other mouth parts are very similar to those given for Chironomus brasenice and therefore need no special mention. FEEDING HABITS. The feeding habits, judging from the food found in the stomach, involve a process of selection. The larvae, as indicated above, creep along through the trash at the bot- tom of streams and consume whatever they encounter that seems most edible. Those examined had a considerable quantity of plant fragments and some soft, brown unrecog- nizable substance in their alimentary canal, but no diatoms were found. It was suggested by Mundy that the vibrissas on the ventral side of the head of this larva might serve as the vibrissa? on the sides of a cat's head. It seems probable to the writer that their function is a tactile one. They doubtless enable the larva to distinguish the different substances which are more or less edible and in this way supplement the more delicate sense structures within the head of the larva. The group as a whole contains larvae with mouth parts that allow greater freedom of movement, especially in connection with the maxillae, than in any of the other groups. Associated with this modification of the maxillae is the entire lack or slight development of the fanlike membrane on each side of the labrum. The hypopharynx seems to be developed as a more efficient organ of sense and doubtless serves an important function in the selection of the food. The other structures of the head are in general similar to those characteristic of the previously discussed groups. It seems probable from the structure of the mouth parts that this is the more primitive group, as Goetghebuer has also suggested. The free-living habits of the larvae seem also to supplement this conclusion. SUBFAMILY TANYPINjE. This subfamily contains seven genera, and at least 45 species are known to occur in North America; but in spite of the number of known adults the author has failed to find any considerable literature bearing on the larval habits. Fr. Meinert (1886) made a number of observations upon a Tanypus species which he figures. These observations have been followed quite closely by Miall and Hammond (1900) and Johannsen (1903). This literature while good as a general treatise fails to give any very adequate idea of the feeding habits beyond the statement that the larvae are predacious. Miall and Hammond, however, state : Bloodworms are preyed upon by many aquatic insects as well as by fishes. Caddis worms, Perla larvae, Sialis larvae, and Tanypus larvae devour them greedily. A number of empty heads of the blood- worms may often be seen in the stomachs of a single Perla or Tanypus larva. 38 BULLETIN OF THE BUREAU OF FISHERIES. The behavior of the larvae belonging to this subfamily is quite similar for the members of the group but very different from that of the larvae of the other subfamilies. The Tanypinae differ from the Chironominae in the length and structure of the head and the function and arrangement of the mouth parts. The anterior prolegs are less strongly developed in Tanypus and are capable of being entirely retracted, so as not to give even a protuberance on that part of the body. The posterior prolegs are more strongly developed, are longer, and are furnished with claws of greater length than those of the Chironominae. In other respects they differ less from the Chironominae in appearance than in structure. They are not all nearly colorless, as Miall (1895) states, but there are a few which are blood red in color. These differ in their behavior as will be explained below. They also differ from all the aquatic members of the subfamily Ceratopogonince in the possession of both anterior and posterior prolegs. In his study of this group the author has observed as many different species as he could find, but his chief attention has been given to Tanypus carneus, T. hirtipennis, T. monilis, and T. dyari, which have also been bred in the course of this work. Tanypus hirtipennis differs from the other species mentioned in having red blood with which is correlated a burrowing habit. The presence of haemoglobin in the blood seems to enable this species to live in a less well-aerated environment in the same way that it does in the case of certain members of the subfamily Chironominae. They do not build tubes as so many of the true bloodworms, however, but simply prowl around pushing their inquisitive heads here and there among the organic debris at the bottom. On this particular point Meinert (1886) states that the Tanypus larvae construct tubes where they remain concealed. Dr. Johannsen tells the writer that in no case has he observed them to behave in this manner. Since the salivary glands are much smaller in proportion to the rest of the body, since the larvae live upon tube-dwelling larvae, and since the pupae are active like those of Culex, it seems probable that they are only found in tubes where they have gone in pursuit of their normal food. This also seems most likely from the fact that the Tanypus larvae when disturbed flap themselves out of the tube or debris where they are, as readily as otherwise, which is not the case with the tube-building larvae. This species, except for its adaptation to a lower level where it is more protected from bottom-crawling enemies as well as impeded in its locomotion, differs but little from the surface-dwelling forms. Tanypus carneus is perhaps the best representative of a surface-dwelling member of this subfamily. It is slim, has a head about three times as long as wide, and its anterior proleg is long and slim and shows its double nature only toward the tip, where it is divided into two rounded branches provided with a few rather delicate claws. This proleg is capable of being completely withdrawn and thus adapts the larva to life among filamentous algae where it seems most at home. The long posterior prolegs enable the larvae to glide along snakelike through the filaments. When an enemy approaches, they are able to withdraw by a backward flexing of these prolegs and the posterior end of the body. Their behavior when sufficiently stimulated resembles that of the crayfish. So rapid is their movement that whether their prolegs catch on any solid particles or not they shoot backward far out of danger. When at rest on the debris beneath the surface, the stimulation of their caudal setae causes them to give a little flip to their bodies which brings their heads almost exactly at the point where their posterior ends had been. This power to rapidly right about face, while especially characteristic of all AQUATIC MIDGES AND SOME RELATED INSECTS. 39 the Tanypinae, is not confined to this subfamily, but is a common reaction in all free- living species of the Chironominae. The author's chief justification for considering the feeding habits of this entire subfamily together is the similarity in the structures of the head and the mouth parts. The figures and the discussion of these parts found chiefly in systematic works have misinterpreted the homology of these parts. On this account, as well as the fact that the mouth parts in this group are of prime importance in any discussion of their feeding habits, they will be discussed in considerable detail. MOUTH PARTS. The hypopharynx of the larva of the subfamily Tanypinae has commonly been called the labium. But it seems probable from figures 42 and 43, tk, that what has hitherto been called the labial plate is really an especially well chitinized anterior border of the hypopharynx. Its strong development is here associated with its very much greater functional importance in this species. The strong muscles attached to this part of the head swing the plate upward and backward, the entire chitinized framework of the hypopharynx taking part in this movement. The toothed border of the upper chiti- nized bar also serves a similar scraping and cutting function (figs. 42 and 43, hy). The labium proper is double and has been labeled hypopharynx. From its position it seems more properly called the labium, and its double nature finds a partial counter- part in the labium of Chironomus digitatus Malloch (Malloch, 1915, pi. 30, fig. 13). In the latter species the central part of the labium is a large rounded light-colored process, while the two sides are black and toothed in a manner very similar to the two- toothed areas in the Tanypinae. The central area appears to be homologous to the "labial papillae" of Malloch, 1915 (pi. 25, figs. 4 and 9; see also fig. 41 in this work). This centrally arranged flap is soft and muscular and has a band of roughened scales on its dorsal surface. The mandibles are opposable and very pointed. They are also able to be used in opposition to the labium. In Tanypus dyari they are furnished with a row of seven lateral teeth (fig. 41, md.), the first of which is especially well developed in practically all species, as is the case in T. dyari. It seems possible that this long-pointed tip, together with the first well-developed tooth, are structures homologous to the double tip so common in the Chironominae, which is frequently mentioned as of specific value. The maxilla in the Tanypinae (fig. 40) is very different from the homologous struc- ture in the Chironominae. Here instead of being attached to a flattened plate (figs 11 and 12) it is capable of a considerable movement. It consists of a flattened append- age with a roughly circular ehitinous supporting structure made up of several partially fused sclerites (fig. 40). This freedom of movement and increased functional impor- tance of the maxilla have an important bearing on the freedom of the movement of the mandible, as one can readily appreciate who is familiar with the restricted move- ment of the latter in the genus Chironomus due to the fusion of the maxilla to the sides of the head. Correlated with this freedom of movement of the maxilla one is able to note that the mandibles may be employed either above or below the maxilla. The anterior portion of the maxilla (fig. 40) is furnished with a large number of thin plate- like processes which doubtless have a tactile function. 4-0 BULLETIN OF THE BUREAU OF FISHERIES. The labrum (fig. 39) is thin and but slightly developed compared to the labrum in the Chironominse. It is furnished on its anterior border with several processes apparently possessing sensory functions. These processes seem to be of more or less specific value and may sometime be of use, as this group is lacking in really good larval characters of systematic value. The chitinous processes (cp) may be homologous to the lateral arms of the hypopharynx of the Chironominae. The head structure of Tanypus is quite unique, being developed for a special manner of life, and the constant recurrence of this structure throughout the group, together with the great similarity in the habits of these species, seems to abundantly justify the placing of all the species in one group. From the ventral side of the head one is able to distinguish longitudinally arranged muscles which are attached at one end to the chitinous framework of the hypopharynx and at the other end to the posterior border of the head. From the dorsal surface, however, one sees a very different mus- cular arrangement. Here the muscles in the anterior part of the head radiate anterio- laterad from the mid-dorsal line, and in the posterior part of the head they radiate posterior-laterad from this same mid-dorsal line. Here in the center of the head is an area which doubtless serves as a sort of pump and to which are attached long muscles which radiate anteriorly and posteriorly. This pump is a structure present throughout the subfamily so far as the author has observed. Another unique feature of the head is the presence of retractile antenna;, which Meinert (1882) has figured and described in detail. He does not seem to have observed their functional significance, however, but considers them only from the standpoint of their anatomy and homology. It will, therefore, be sufficient to state that the antenna; are withdrawn into the head capsule itself where there are special chitinous sheaths to receive them. They are withdrawn by special well-developed muscles and are said to be extended by blood pressure. They are of great functional importance in that they enable the larvae to actually measure the distance from their prey. A few easily made observations enable one to see how constantly they are used. The larvae prowl about with their antennae partially extended, and upon encountering an active object they withdraw them as they approach, thus keeping in touch until near enough to seize the object. FEEDING HABITS. The larvae are all predacious as far as the author has been able to determine, although it is very difficult to actually observe them feeding. Numerous studies of the stomach contents of the larvae have shown such an array of diatoms and desmids as to entirely mislead one looking for proof of their predacious habits. Meinert (1886) states that he has observed a living Simocephalus in the intestine of a Tanypus larva, while Miall and Hammond (1900) have apparently observed indubitable evidence that they were predacious on bloodworms from the presence of the heads of these larvae in their stomachs. It was not, however, until the author had starved a Tanypus carneus larva for a week that he was able to observe the actual feeding habits. This larva when put in a dish containing a number of large Cyprididae would apparently strike at them when they came in contact with its head. The striking seemed to be a more or less involuntary reaction, for when the nature and size of these crustaceans were discovered they were AQUATIC MIDGES AND SOME RELATED INSECTS. 41 allowed to go uninjured. One of these crustaceans was killed and placed near the head of the hungry larva, but it was left undisturbed until movement was imparted to it by the aid of a needle. From repeated observations it seems apparent thac the larvae of this subfamily will not touch anything which is not moving. When movement was imparted artificially to dead pyschodid larvae they were attacked, but before the skin was broken the larva abandoned them. When tried a day or so later on this larva it would not touch it, thus showing that decaying or dead material is not eaten even when the larva is very hungry, a fact in decided contrast with the behavior of a number of Chironomus larvae, especially Chironomus lobiferus. The Tanypus larva to which was offered a freshly killed crustacean, however, ate it readily. It nevertheless showed a preference for small recently hatched bloodworms. These were swallowed whole and were apparently uninjured, as they were capable of moving for a time after being swallowed. The method of attack and the defensive attitude of the larvae of this subfamily were well shown in an encounter which occurred between two larvae which the author was keeping on short rations preparatory to making observations on their feeding habits. The larvae were of different species, one having a smaller and longer head than the other. The encounter was a head-on collision, each apparently striving to defend itself. They were taken under the compound microscope and their behavior observed. The head of the smaller larva was apparently not much within that of the larger one, but it was easy to see that the muscles within its head were being sucked toward the anterior tip of the head. The result was that the smaller larva was killed, although it was not consumed nor were any of the muscles of the head actually sucked out. The survivor, although unprovided with food from any other source, left its victim undisturbed as a result, doubtless, of its lack of movement. An observation on Tanypus carneus well illustrates the function of the head as a sucking organ. A specimen that had been without food for four or five days was placed in a watch glass with a very active bloodworm (Chironomus sp. ?) which was about the same size as the Tanypus larva. The Tanypus larva attacked the bloodworm just back of the head, employing its sharp mandibles to hold the larva. Very soon a reddish color could be observed in the head of the attacking larva, showing that it was beginning to suck the blood of the other. Then the alimentary canal was cut off, probably by the mandibles, and with its contents (diatoms, etc.) sucked into the body of the Tanypus. This left only the collapsed body wall of the larva to be consumed. This was accom- plished by the use of the same powerful sucking apparatus. The body wall was drawn into the mouth while the hypopharynx rasped a hole through it, then the continued squeezing and sucking action of the head removed the muscles of the body wall. This method of treatment was repeated on different parts of the body until finally all the muscles of the body wall were removed. In this case the head was not swallowed and the muscle fragments of the bloodworm were in such a broken state that they would almost defy identification. Miall and Hammond have remarked on the presence of red coloring matter in the body of Tanypus larvae which they consider due to the bloodworms they have eaten. It is easy to confuse a natural red color with the color due to the food eaten, but a little experience will enable one to see a difference in the intensity of the color that is unmis- takable. 42 BULLETIN OF THE BUREAU OF FISHERIES. The silk glands (fig. 44) are small and egg-shaped in general outline. Their ducts fuse some little distance posterior to their opening, which is situated just dorsal to the anterior border of the hypopharynx. The shape, relative size, and transparency of these salivary glands, together with the very different functions of the head and mouth parts in the Tanypinae, at least suggest that their function is more exclusively that of a digestive gland than it is in the Chironominae. The alimentary canal is developed rather differently in the Tanypinae than in the Chironominae. Miall (1895) figures the alimentary tract for Tanypus maculatus but does not label the parts. In comparing these structures with the drawings of the ali- mentary canal of Chironomus sp. ? given by Miall and Hammond the croplike enlarge- ment so easily distinguished in the Tanypinae (fig. 44, cr) is represented only by a narrow esophagus. The cardial chamber (fig. 44, c) is narrow and sharply marked off, and its surface is covered with longer coeca than in the Chironomus, but, on the whole, not so very different from it. The stomach proper (fig. 44, st) is proportionally shorter and of less functional importance. The remainder of the alimentary canal is quite similar in both subfamilies. The food is retained in the crop (fig. 44, cr) part of the alimentary canal when first consumed and is constantly being stirred about by a peristaltic motion. When speci- mens are starved for a considerable time, the food is retained in this part of the ali- mentary tract often for the greater part of a week, which would seem to indicate its relative importance. The peritrophic membrane, if present, is very thin and inconspicuous. The author has been unable to discover its presence by gross dissection and has consequently con- cluded that in this respect the Tanypinae are decidedly modified as a result of then- carnivorous habits. From the length of time that the diatoms are retained in the ali- mentary canal it would seem probable that they also are as well digested as they are in the stomach of the Chironominae. As stated above, the digestion in this latter group is quite incomplete and any comparative statement must be relative in its nature. It will be clear from the above considerations that the fundamental structures of this subfamily are closely correlated with its peculiar manner of life. It seems probable that this subfamily represents a more primitive type of insect than those included in the Chironominae. This conclusion is based not alone on the free-living active behavior of the larvae, but also upon the pupae, which resemble the pupae of Culex in their manner of life, as well as upon the primitive venation of the wings of the adults. SUBFAMILY CERATOPOGON I Nj£. This is a widely distributed group. Many of the adults are known as blood-sucking insects, some attacking other insects exclusively and some turning their attention to the higher animals including man, while others appear not to take any food in the adult condition. In the larval condition their habits are also variable. Guerin (1833) found the larvae of Ceratopogon geniculatus Guerin and C. flavifrons Guerin under the bark of dead trees in a humid environment. Dufour (1845) found larvae of a species which he identified as Ceratopogon geniculatus Guerin in decomposing onions. Perris (1847) found the larvae of Ceratopogon brunnipes Perris in decomposing mushrooms at the AQUATIC MIDGES AND SOME RELATED INSECTS. 43 base of a poplar tree. He also found Ceratopogon lucorum Mirgen in a heap of decom- posing elm leaves and succeeded in rearing them indoors in this same material Laboul- bene (1869) found the larva? of an unidentified species of Ceratopogon in the ulcers or injured places in elm trees where they were living in what Dufour (1845) calls "la marmelade de l'Orme." These larvae were reared and the species named Ceratopogon dujouri in honor of Leon Dufour by Laboulbene. Long (1902) found Ceratopogon brumalis Long in great numbers on the underside of nearly dry cow dung. He also found several hundred larvae of all ages on the undersurface of a piece of moist rotting elm wood. He found similar larvae and pupae in the nests of the common foraging ants (Eciton coecum). The larvae of Ceratopogon specularis Coquillett were found by Long to live gregariously in cow dung. Larvae of Ceratopogon stenomaiis Long were found by Dr. W. M. Wheeler in an ant nest, where they were moving about in the refuse heaped up by the ants in certain portions of their nests. The larvae of Ceratopogon taxanus Long (Long, 1902) were found beneath the bark of old dead trees in moist places or on the underside of very damp rotting wood. The only other habitat so far as known where the larvae are commonly found is a strictly aquatic one. This latter environment according to Johannsen (1905) is occupied by the species having smooth wings. An examination of Malloch's (1915) keys, which cover only the Illinois species of this subfamily, shows 4 genera and 22 species with hairy wings to 9 genera and 72 species with smooth wings. It would appear, therefore (granting the supposition that smooth wings and aquatic habitat for the larvae are correlated characters), that the greater number of the species are aquatic, but so few species are known in the immature stages that it is impossible to say whether the greater number undergoes development in water or in some more distinctly terrestrial environment. BODY STRUCTURES. The bodies of the aquatic larvae are long and tapering, and their heads are propor- tionally longer and slimmer than those of the semiaquatic and terrestrial forms. The aquatic larvae are entirely devoid of walking appendages, and the only external body structures that link them up with their near allies are the caudal filaments. These have either been considered homologous with claws of the posterior prolegs or left with- out any attempt at a comparison. In a permanent preparation of the larva of a Culi- coidies sp. ? the author has discovered that these filaments are arranged in two groups (figs. 45 and 48), which clearly suggest that they are homologous with the caudal filaments of the Chironominae. Several authors have suggested that these structures, since they can be made to point either forward or backward, function as locomotor appendages. This observation is apparently correct. The great relative size and length of these caudal filaments seem to be functional modifications, for they are sense organs in other genera of the family. The semiaquatic species (fig. 49) found in the sap flows of injured elms here at Ithaca, N. Y., resembles the one described by Laboulbene (1869), which he named Ceratopogon dufouri. These larvae differ from the aquatic forms in having only poorly developed caudal setae and in the presence of very short and contractile posterior pro- legs, which are fused together and provided with a circle of hooked claws (fig. 50) . This 44 BULLETIN OF THE BUREAU OF FISHERIES. form was not bred, but from the mouth parts it seems possible that this is a myceto- philid and not a chironomid at all. The typically terrestrial forms have well-developed anterior and posterior prolegs. Perris's (1847) observations on Ceratopogon brunnipes and C. lucorum, together with a study of the aquatic species already mentioned, would indicate a series of intermediate stages between those with both anterior and posterior prolegs and those without either. The author has indicated one stage of this series above. The first stage has caudal filaments which replace the posterior prolegs. The other stages are represented by the two species described by Perris. The first, Ceratopogon brunnipes, he found in decaying mushrooms at the foot of a poplar tree. Of this species he says the anterior prolegs are deeply dilobed and each lobe is furnished with a few claws. These are completely retractile, but those of the posterior prolegs are not. The other species, Ceratopogon lucorum, he says, appears to have a proleg formed of two pieces united by a suture, each of which is feebly bilobed. The exterior lobe is bare, and the inner lobe is furnished with fine spines. The last species, found in decaying elm leaves, resembles the typically terrestrial forms in general ; besides possessing both anterior and posterior prolegs it has a spiny body. In considering the terrestrial nature of these larvae Laboulbene (1869) says that stigmates certainly exist but that he has not been able to count the openings. His opinion that they really do exist seems to be due to his observations on the arrange- ment of the tracheae. The authors quoted in regard to the variety of habitats occupied by the larvae of the Ceratopogoninae all emphasize the humid condition of the habitat, and in the absence of any direct observations on the presence of spiracles it seems probable that in this respect at least the group is a unit. HEAD STRUCTURES. To understand the feeding habits of any insect, the structure of the mouth parts lends an important clue. This is equally true of the mouth parts of the larvae of the Ceratopogoninae. The frequency with which the early students complain of the diffi- culties of such a study is a sufficient justification for a somewhat incomplete consideration of these structures here. In the aquatic larvae Culicoides sp. ?, probably C. guttipennis, the head is long and slim, about four times as long as wide. The antennae, so conspicuous and useful in the other subfamilies, are here very slightly developed, scarcely reaching to the anterior border of the head. Their location (fig. 45) on the dorso-lateral border of the head, together with their slight development, fits them to serve as a sense organ with only a very limited function. This slight development of the antennae is characteristic of the entire subfamily and is doubtless associated with the nature of the food consumed. The mandibles, in Culicoides sp. ?, possibly guttipennis, are quite characteristic structures (figs. 46, 47, md). They are so hinged as to be capable of being extended beyond the head and are opposable. They are also often observed within the head with their tips pointing backward, showing that they have a wide range in their move- ments. The fact that the mandibles are capable of being swung through such a wide angle shows that they are doubtless very essential to the feeding habits of the larvae. The larvae resembling Laboulbene's larvae which the author found in the sap flows on elm trees about Ithaca have mandibles with teeth resembling those on the typical AQUATIC MIDGES AND SOME RELATED INSECTS. 45 Chironomus mandibles (fig. 51). These were used by the larvae for the purpose of locomotion. They move alternately, and the head is tipped down slightly, so that their motion in a dorso-ventral direction enables them to function as feet. In this species the anterior prolegs are entirely lacking, and the larva moves by a gliding eel-like motion aided by the mandibles, which also help to clear the way. A similar function of the mandibles in Culicoides sp. is suggested (figs. 46 and 47). The labium has several times been figured showing a strongly chitinized central tooth and in some species a single pair of lateral points on the otherwise smooth, some- what thickened lower margin of the head. This central tooth appears to be the hypo- pharynx. Its probable function as a piercing organ would doubtless be suggested to everyone by its shape. The epipharynx in Culicoides sp. seems to be located near the middle of the head and, so far as it is possible to tell from drawings, what the author considers as the epi- pharynx (figs. 46 and 47, ep) is what has been called the hypopharynx by Malloch (191 5). Its function seems to be that of a strainer or comb, coupled doubtless with a tactile function. The ventral half of the head seems to be fitted with long muscles which doubtless operate the mouth parts. The dorsal half of the head posterior to the epipharynx seems to be filled with radiating muscles, as described in the case of the Tanypina, which doubtless serve a similar function, namely, that of a pump or sucking organ. FEEDING HABITS. Culicoides (guttipcnnis?) larvae, which were under observation for some time, were extremely hard to observe while feeding. The only case actually seen was that of a larva feeding within the dead body of a pupa. This pupa was of the same species as the larva. When observed, the body of the larva was thrust deep into the nearly empty shell. The larva was revolving its head and first two segments about, and of course could not bring pressure upon its mouth parts because of its lack of prolegs. It seems probable that the mandibles came into use at this time, as they are the only mouth parts adequate to the purpose. It was impossible to observe the activity of the mouth parts on account of the thickness of the chitinous wall. The author's experience in trying to study the stomach contents of these species is exactly parallel to that of Miall and Hammond (1900), who were unable to identify the small particles occasionally found in the stomach. According to these same authors ' 'The digestive system is straight and simple and apparently adapted to the wants of a carnivorous animal." From the obvious specialization of these aquatic larvae, as shown by their relatively great length and slight breadth, it seems fair to assume that they are adapted to an environment where they are able to reach food inaccessible to thicker and more chubby larvae. During the winter larvae of an aquatic species were found deep in various decaying stems, especially those of Typha and Sparganium. The following summer the writer was unable to find them in connection with these stems, but by dipping up masses of floating green algae in the same pool he found them in considerable numbers. What then is the nature of the food upon which the larvae of the Ceratopogoninae live? It seems probable that larvae living under decaying bark, in rotting onions, among decaying elm leaves, or under cow dung would have but little choice in the food 46 BULLETIN OF THE BUREAU OF FISHERIES. which they obtain, especially when we learn that these larva? do not burrow but simply wiggle and creep along through the moist and semifluid portions of their environment. That they in all probability live on decaying organic matter, together with the bacteria and mold which are always present in such substances, seems obviously a case of neces- sity. That an aquatic environment offers a greater opportunity for larvae adapted to such a life to select their food is readily seen from the study of even a limited habitat. The fact that aquatic larva? kept in confinement will eat animal material is well shown by the observations mentioned above of a larva found eating the tissue of a dead pupa of its own kind. Their presence among filamentous algae suggests that they may also eat out the protoplasmic contents of the larger filaments. It seems probable that various other organisms might become entangled in these filaments and be used as food by these larvae. In conclusion, it may be said that the larvae of this group as a whole show a speciali- zation of the mouth parts which fit them to live on soft substances. The various habi- tats in which the larvae are found seem to bear testimony to the organic nature of the food consumed. The uniform failure of all attempts at microscopic analysis of the stomach contents of these larvae suggests the structureless nature of the food taken. It seems apparent, therefore, that, since all decomposing organic matter offers a very similar food supply, the larvae of this subfamily are capable of adapting themselves to a wide range of food substances. The humidity more than any one other factor doubt- less limits this adaptability, since moisture not only aids decay, thus making hard inedi- ble substances available as food, but also serves as a factor of prime importance from the standpoint of respiration to a larva lacking spiracles. SUMMARY. In the following groups the author has tried to show the more striking differences in the feeding habits of the Chironomidae. In each group the mouth parts and general behavior have been made use of in determining the feeding habits. The stomach con- tents have been depended upon only as a -confirmation of activities actually observed, thus avoiding several errors in connection with the predacious forms. The family as a whole shows a wide range of structural variations and a wider range, if possible, of special adaptations. In Group I the larvae, although somewhat variable in habits, can and do live to a very considerable extent upon bacteria, Protozoa, diatoms, small Crustacea and other free-floating aquatic organisms, which they strain from the current driven through their burrows by means of delicate silken nets. In Group II the larvae utilize the natural flow of the stream and subsist on the plankton organisms found there. The individuals of this group are usually very numer- ous, but as Group I consists of forms characteristic of quiet water, Group II does not compete with it. Group III contains the greater number of the typical Chironomus larvae, known as bloodworms. They are found wherever any considerable accumulation of diatoms and plant debris occurs. In laboratory experiments they were found to be able to sub- sist for considerable periods upon a very scanty supply of food material. Thus, we find them to be a group capable of utilizing and conserving whatever amount of diatoms, algae, and plant debris may chance to fall upon the bottoms of fresh-water ponds. Their AQUATIC MIDGES AND SOME RELATED INSECTS. 47 large size and overlapping broods indicate their possible importance to fish life and in fish culture. Group IV contains an aberrant species that feeds directly upon floating aquatic leaves and is noteworthy chiefly for its direct injury to those plants. Group V contains at least one species (Trichocladius nitidellus) that promises to be of considerable importance, as it is able to subsist entirely upon filamentous algae, chiefly Spirogyra. Group VI includes a number of free-living forms that occupy somewhat unique habitats but constitute a group of minor importance. The entire subfamily Tanypinae consists of predacious forms, which as a group apparently do not contribute anything of economic importance to the Chironomidae as a whole. They, however, do occasionally feed upon small Crustacea and the more rapidly moving diatoms and in this way help to counteract their otherwise well-merited position as an economically undesirable group, from the standpoint of the fish-culturist. The subfamily Ceratopogoninae are scavengers as a group and as such fulfill a useful function. From the consideration of these rather arbitrary divisions, as well as the natural subdivisions of this large family, it becomes evident that there is a wide range in the adaptations of its different members. Some of these adaptations are of generic value, while others seem to vary within the genus, as in Tanytarsus obediens, which is in- cluded in Group I, although the other members of the group belong to the genus Chironomus. In a similar manner the red color of the larvae seems to occur with little or no relation to the genus or subfamily but is rather more closely associated with the nature of the environment. It is obvious from the above that the family has become specialized for different habitats. While the author has tried to point out what seems to be the behavior nor- mally characterizing each group, it is easily apparent from a few observations that the great adaptability of all these species when under the stress of adverse conditions reduces them to what is probably the primitive habit of the group, namely, that of direct feeding on the debris about them. The degree of departure from the primitive method of feeding, however, varies considerably. In the Tanypinae we have a form that is strictly predacious, while in the Ceratopogoninae we have a form that is adapted to live on dead and even decaying organic matter. The latter group seems to be about as abundantly represented in semiterrestrial environments as in those that are strictly aquatic, and it is this group that doubtless contains the most or perhaps the least primitive representatives of the family. The Tanypinae doubtless come next and then the Chironominse. In the Chironominse it would seem that from direct feeding by the use of silk to attach together the particles fed upon the use of silk in entangling particles in a stream would be but a simple step. Then, from this beginning an artificial current, made necessary by the poor supply of air, might readily lead up through a series of stages, from the entanglement of particles in the lining of their burrows to the present highly specialized silk net, which characterizes Group I. The small size of these larvae and their adaptability to such a wide range of habitats enable them to take possession of an environment where the food supply would be 48 BULLETIN OP THE BUREAU OP FISHERIES. insufficient for a larger form with similar food requirements. It is this factor that seems most readily to explain the wide range and great numerical dominance of the family. It is this fact, too, which seems best to account for their numerous enemies among the aquatic animals. ORPHNEPHILIDjE. This family is included here because of its close kinship with the Chironomidae, as shown by the structure of the adult. It is also of considerable interest on account of the unique and little known habits of the larva, which lives on the surface of ledges covered by only a thin film of water (fig. 38) and breathes by means of a trachea, ren- dering it entirely unable to live submerged for any considerable time. As might be inferred from these two conditioning factors this family is not likely to occur in many parts of the country. That it is really scarce is well illustrated by the fact that the record for the family in this country prior to 1916, so far as known, was based on three specimens found by Dr. O. A. Johannsen at Ithaca, N. Y. Dr. V. L. Kellogg (1902) states that he has examined specimens of every family of the Nematocera except Orphnephila. References in the literature to this family are so infrequent as to make it almost unknown except to a specialist in Diptera. Thienemann (1909), however, found and described the larva which he obtained from mountain streams in Europe. His paper considers the nature of the habitat, the distribution, and the method of locomotion of the larvae of Orphnephila testacea. According to Kellogg (1905, p. 327) this species is the only one representing the family in this country and as far as the author is aware the only species known, if the American and European forms are actu- ally identical. The three adult flies found by Dr. Johannsen referred to above were taken in sweep- ing for insects, and none were taken in a manner to reveal the whereabouts of their immature stages. It was therefore a very pleasant surprise to the author to accidentally run across the habitat of this most unique semiaquatic insect in the environment of Ithaca, the only place in this country where this species is known to occur. As the interests of the author centered about the ecology of the species, especially as it concerns the feeding habits of the larvae, he several times attempted to take speci- mens into the laboratory that the necessary conditions of their environment might be more readily studied. Several of these attempts were failures because the larvae were drowned while en route, but by lining test tubes with moist cheesecloth it was found very easy to carry any number of the larvae considerable distances under perfectly normal conditions. HABITAT. While Thienemann's description is in substantial agreement with the writer's own observations, it seems best to summarize the conditions under which the larvae were found. The horizontal strata of the rock, so characteristic of all the gorges and "hanging valleys" in the environment of Ithaca, together with the usually rather irregular ver- tical cleavage, frequently gives rise to a stair-stepped bottom to the streams that enter the deeper valleys. The only habitat in which these larvae were found was on a series of "giant steps" (fig. 38), where a small stream spreads out over these broad and nearly horizontal stones in its precipitous descent to the valley of Six Mile Creek. Here the larvae were found rather more frequently on the vertical than the horizontal surfaces of AQUATIC MIDGES AND SOME RELATED INSECTS. 49 the ledges. They also seemed to select those parts of the rocks which were free from any vegetable growth, a selection probably closely correlated with their method of locomotion. » That the larvae are unable to live on the surface of any other than perfectly quiet water was discovered by repeated attempts, as above stated, to carry living larvae home in bottles half filled with water. The result was nearly ioo per cent fatalities. This does not mean that they can not move over quiet water, for they are very much at home in such conditions ; but as soon as the surface film rises above the ventral third of the body, which is distinguished by being somewhat flattened and white in color, the result is total submergence. Total submergence is, of course, only fatal in larvae which breathe by means of tracheae and have no special means of escaping from the water. The larvae of Orphnephila do breathe by tracheae, and, while they are very well adapted to move rapidly on the surface of moist ledges, their very peculiar sidewise move- ment of the body is not at all suited to locomotion beneath the surface film. In fact, they depend very largely upon the surface film which holds them so closely in contact with the rock's surface that with the claws of their anterior and posterior prolegs they are able to anchor one end while the other is being swung around in a horizontal plane. This zigzag sidewise movement is sufficiently rapid to enable them to move four or five times as fast as a chironomid can crawl and gives them the appearance of being very sprightly. This poor adaption to an aquatic environment is one of the factors that doubtless makes for their infrequent occurrence, next to the nature of their habitat, which is of itself rather unique. In several laboratory experiments in which the author attempted to duplicate natural conditions the larvae were observed leaving the moist stones upon which they had been placed and voluntarily subjecting themselves to the current which swept them over the edge of the dish. So far as it is possible to judge, the same thing is liable to occur in nature, and the results are doubtless fatal, for the bigger streams are constantly agitated by swirls and cross currents which would submerge and drown the larvae. In order to try to eliminate the nonessentials in the environment of the larvae, the author began searching for suitable methods of rearing them in captivity. At first an experiment, referred to above, was set up in which rocks taken from the natural habitat were placed in a tray through which a stream of gently flowing water was maintained. The result was that the larvae allowed themselves to be carried over the edge and were lost down the sink spout. At Dr. Johannsen's suggestion a cheesecloth pocket was used and resulted in the successful completion of the transformation of some five or six adults. This pocket was made by placing a double thickness of cheesecloth over the top of a wide lamp chimney and pressing it down so that it would hang in a conical sort of a pocket. This was covered with two thicknesses of cheesecloth after the larvae and a fair supply of food had been placed within. Then both the cover and the pocket were made fast by successive coils of white thread, which were wound about so tightly that the larvae were unable to creep out between the layers of the cheesecloth. The pocket was moistened by water which was kept dripping slowly over its surface. To insure the more uniform distribution of the water to all parts of the receptacle, a mass of cotton wool was placed on the cheesecloth cover. That this sort of an artificial environment seemed to meet their every need was demonstrated by the fact that the larvae lived under 50 BULLETIN OF THE BUREAU OF FISHERIES. these conditions for several weeks with very few fatalities, even though the water ceased dripping several times for a number of hours, thus causing a considerable drying out. The pupa, which seem never to have been found in nature, were observed to be located in a fold in the pocket where the water supply was more uniform and where greater security of position was doubtless possible. The author was unable to find a pupa out of doors even after he had bred the pupae in an artificial environment. The larvae, while quite abundant in the one habitat in which the author succeeded in finding them, either do not live to transform to pupae in any considerable numbers or else they possess some unusual habits which entirely escaped the writer's notice, for repeated search for pupae in the most likely places and at such widely separated intervals was made that it does not seem possible that they could have been abundant in the environment occupied by the larvae. That the pupa can live under the same conditions in which the larvae are found is amply demonstrated by the author's laboratory experiments, where upwards of 50 per cent of the nearly mature larvae trans- formed to adults. Another source of information which seems to corroborate the notion that the pupae are not abundant was the fact that repeated sweeping over these rocks and in the adjoining region failed to give even a single adult specimen. The eggs so far as is known have never been found, and nothing is known of the mating habits of the adults. FEEDING HABITS. The feeding habits of Orphnephila are no less unique than its other environmental adaptations. Let us first take up the structure and arrangement of the mouth parts, as a knowledge of their nature and position is fundamental to all ecological considerations. Thienemann (1909), as mentioned above, has figured the more commonly observed mouth parts of Orphnephila, but the separate drawings give no adequate notion of the relative position of each part. The author has found it necessary to draw the mouth parts as they appear in position and then for the sake of comparison several of them separately. The assembled mouth parts (figs. 53 and 54) show that the mandibles, instead of moving from the outside inward toward the mid line, as described in the case of the chironomids, are so hinged as to move outward from the mid line when in use for the purpose of scraping food from the rocks. This arrangement of the mandibles in Orphnephila, so far as the author is aware, is unique among Arthropoda. Correlated with the mandibles are the maxillae which are furnished with a border of spoon-shaped plates which are opposable to the mandibles. This arrangement makes their function as collecting baskets, for gathering in the particles scraped free from the stones by the mandibles, quite obvious. The rods shown in figure 54, rd are supporting structures which fuse with the clypeal plate and extending beneath the mandibles form a partial support for the articulation of the maxillae. The very marked development of the labrum suggests at least its probable function, and while the writer has not been able to observe this particular mouth part in use it is probably brought into play in connection with the mandibles in such a way as to scrape an intermediate area not touched by them. The rather narrow labrum is provided with a considerable number and variety of spines at its terminal end and, together with the somewhat similarly clothed hypopharynx, is doubtless instrumental in collecting the food scraped loose by the labrum, as well as in the removal of the food particles assembled by the maxillae {fig. 5j, lb). AQUATIC MIDGES AND SOME RELATED INSECTS. 5 1 The food itself consists almost exclusively of diatoms, and as the number of kinds of diatoms in such streams is few and as those found on exposed ledges where the larvae feed are even less varied the variety is not great. Figure 57 shows a typical selection of the food from the stomach of one of these larvae. The many unique features in the habits of this insect seem to limit its life and activities to a very restricted environment to which it seems but poorly adapted, if the infrequent capture of adults and the evident scarcity of pupae can be taken as criterions. From the above observations it might readily be considered as a species which had only recently acquired the aquatic habit. EXPLANATION OF FIGURES. ChIRONOMCS BRASENI.E. Fig. i. — Ventral aspect of the epipharynx; a, lateral arm; co, epipharyngeal comb; e. chitinous claws; h, horseshoe-shaped chitinized area; p, lateral process, s, pectinate hairs; t, thickened border of the labrum; x, chitinous process. Fig. 2. — Lateral view of the strongly chitinized structures somewhat diagrammatic; p, lateral process; s, setae and pectinate hairs; u, clypeus; (other structures as above). Fig. 3. — Lateral aspect of the mandible (md); j, pectinate setae; q, external process; m, portion of the adductor muscle and its thickened exteria; v, extensor muscle. Fig. 4. — Median aspect of the left side of the head; c, cardo (position indicated); g, galea; i, in- ternal chitinous process; j, pectinate setae; 1, lacinia; la, labium; m, adductor muscle; md, mandible; p, maxillary palpus; st, stipes; u, clypeus; v, extensor muscle; w, center of articulation of the galea and lacinae. Fig. 5. — Ventral aspect of a portion of an epicranial plate; i, internal chitinous process; r, antenna; u, clypeus. Fig. 6. — Dorsal aspect of mandible; j, pectinate setae; o, articular surface of mandible; q, external process. » Fig. 7. — Dorsal aspect of hypopharynx; b, posterior lobe; e, backward pointing setae; k, arm or chitinous process; th, chitinous plate; z, exit of the salivary ducts. Fig. 8. — Dorso-ventral aspect of the hypopharynx; d, salivary duct; f, chitinous ring; k, arm or chitinous process; th, chitinous plate. Fig. 9. — Antenna, lateral aspect; 1, Lauterborn 's organ ; n, sensory spine. Fig. 10. — Ventral view of the head; c, cordo; d, salivary duct; c, epicranial plate; la, labium; md. mandible; mx, maxilla; r, antenna; u, clypeus. Fig. 11. — Lateral aspect of the maxilla; c, cardo or striated membrane; g, galea; 1, lacinia; p. palpus; pf, palpifer; st, stipes; w, center of articulation of the galea and lacinia. Fig. 12. —Lateral aspect of a portion of the epicranial plate; c, cardo or striated membrane, la, labium; pr, chitinous process limiting the movement of the mandibles; st, stipes. Fig. 13. — Ventral aspect of the anterior margin of the labium. Fig. 14. — Labium of young larva. Fig. 15. — Posterior segments of the larva; b, branchial gills; ca, caudal filaments; c, claws. Fig. 16. — Lateral fin on the 8th segment of the pupa; ca, chitinized area. Fig. 17. — Hypopygium of the imago, dorsal aspect. Fig. iS.— Wing. Fig. 19. — Portion of an egg mass. Fig. 20. — Tracings of the larval burrows in the leaves of Brasenia. Fig. 21. — The same as fig. 20. Fig. 22. — The same as fig. 20. Fig. 23. — Diagrammatic drawing of the larval burrow greatly enlarged; a, silk supporting threads; b, section of the epidermis cut out by the larva; c, canopy composed of larval castings; d, epidermal slab used as the sides of the larval burrow; e, edge of epidermis and the underlying parenchyma; f , vein of the leaf; g, mucus gland. Fig. 24. — Pupal chamber on the leaf of Casta Ha odorata;c, idioblast; p, anterior portion of pupa molt. Fig. 25. — Cross section of a Castalia odorala leaf showing its general structure; a, stomata; b, upper epidermis; c, idioblast; d, lower epidermis; e, air space. 52 BULLETIN OF THE BUREAU OF FISHERIES. Chironomus lobiferus. Fig. 26. — Diagram of a glass preparation showing the position occupied by the larva in its burrow; lc, lower glass; nt, conical net; t, glass tube; u, upper glass; v, larva. Fig. 27. — Diagram of the contracted conical net. Fig. 28 Fig. 29 Fig. 30 Fig. 31 Fig. 32 — Broken labial plate. Orthocladius sp. — Labium. — Mandible. Prodi ames a sp. — Maxilla; ar, articular process; c, vibrissa?; 1, lacinia; la, labium; p, palpus; st, stipes. — Ventral aspect of a portion of the head; ar, articular process; c, vibrissa?; la, labium. Chironomus lobiferus. Fig. ji- — Claw of posterior proleg. Fig. 34. — .Spines of anterior proleg; a, serrate spine; b, c, hooked spines; d, hairlike spine. Fig. 35. — Left anterior proleg; a, hairlike spines; b, hooked spines; c, serrate spines. Fig. 36. — An enlarged silken holdfast attached to a glass surface; g, a process on the side of a larval burrow. Fig. 37. — Silken network between two larval burrows ( g, h); i, interlacing silk fibers. ORPHNEPHILA AMERICANA. Fig. 38. — View of habitat where larvje were found, west bank of Six Mile Creek, Ithaca, N. Y. ; F, water level of the main stream; 1, ledges kept moist by water flowing in a small stream shown just above; w, water flowing over ledges. Tanypus DYARI. Fig. 39. — Labrum; ch, chitinous plate; cp, chitinized process; se, sensory process. Fig. 40. — Chitinous structures of the maxilla; 1, lacinia; p, palpus. Fig. 41. — Ventral aspect pf the head; hy, hypopharynx; la, labial papilla; md, mandible; mx, maxilla; p, maxillary palpus; r, antenna. Fig. 42. — Dorsal view of the labium and hypopharynx; ar, articular plate; cb, chitinous band; hy, hypopharynx; la, labial process; lp, lateral hypopharyngeal process; pr, sensory process; th, chitinous plate of the hypopharynx. Fig. 43. — Lateral view of the same. Fig. 44. — Dorsal view of the entire larva; br, branchial gill; c, cardiac chamber; cr, crop or pro- ventriculus; g, salivary gland; m, malpighian tubule; r, retractile antenna; st, stomach or ventriculus. Culicoides SP. Fig. 45. — Dorsal view of entire larva; f, caudal filaments; g, branchial gills. Fig. 46. Lateral view of the head; d, salivary duct; ep, epipharynx; ey, eye spot; hy, hypo- pharynx; la, labium; lb, labrum; md, mandible; mx, maxilla. Fig. 47. — Ventral view of the head; structures the same as shown in fig. 46. Fig. 48. — Dorsal view of the posterior end of the larva considerably enlarged; f, caudal filaments; ff, accessory caudal filaments; g, branchial gills. Larva from sap flow in elms resembling Culicoides hieroglyfthicus. Fig. 49. — The entire larva, dorsal view; ch, chitinous rod. Fig. 50. — Posterior end of the same greatly enlarged; c, caudal processes; g, branchial gills; s, sets resembling caudal filaments. Fig. 51. — Mandible. Orphnephila testacea. Fig. 52. — Side view of entire larva showing characteristic color pattern. Fig. 53. — Side view of the head and proleg; ey, eye spot; la, labium; lb, labrum; md, mandible; mx, maxilla; pr, proleg; r, antenna; sp, spines. Fig. 54. — Frontal view of the head; cl, clypial plate; la, labium; lb, labrum; md, mandible; mx, maxilla; rd, rod attached to the mandible. Fig. 55. — Ventral aspect of the mandible. Fig. 56. — Dorsal-lateral aspect of the hypopharynx. Fig. 57. — Miscellaneous diatoms from the stomach of Orphnephila. AQUATIC MIDGES AND SOME RELATED INSECTS. 53 54 BULLETIN OF THE BUREAU OF FISHERIES. 11 13 14 16 15 17 AQUATIC MIDGES AND SOME RELATED INSECTS. 55 " ■ ' ■ 23 56 BULLETIN OF THE BUREAU OF FISHERIES. Bull. U. S. B. F., 19 n t — 9 -> J^P ■'.• • *« ': 36 37 38 AQUATIC MIDGES AND SOME RELATED INSECTS. 57 43 41 58 BULLETIN OF THE BUREAU OF FISHERIES. 46 u 45 51 AQUATIC MIDGES AND SOME RELATED INSECTS. 59 52 53 54 55 56 0 *n$ 60 BULLETIN OF THE BUREAU OP FISHERIES. BIBLIOGRAPHY. DuFOUR, LEON. 1845. Observations sur les metamorphoses du Ceratopogon geniculatus Guerin. Annales de la Societe entomologique de France, 2e ser., Tome III, p. 215-223, pi. Ill, No. II. Paris. GOETGHEBUER, M. 1911. Chironomides de Belgique. Annales de la Societe entomologique de Belgique, Tome 55, p. 95-113- Bruxelles. 1913. Etudes sur les Chironomides de Belgique. Memoires de l'Academie royale des sciences, des lettres et des beaux-arts de Belgique, 3 fasc, 6 Deer., p. 1-26, pi. I-V. Bruxelles. Guerin, F. E. 1833. Notice sur les metamorphoses des Ceratapogons, et description de deux especes nouvelles de ce genre, decouvertes aux environs de Paris. Annales de la Societe entomologique de France, Tome II, p. 161-167, PI. VIII. Paris. Heeger, Ernst. 1856. Neue metamorphosen einiger Dipteren. Sitzungsberichte der Kaiserlichen Akademie der Wissenschaften. Bd. XX, p. 335-350, pi. 4. Wien. Holmgren, N. 1904. Zur Morphologie des Insektenkopfes. 1. Zum metameren Aufbau des Kopfes der Chirono- mouslarve. Zeitschrift fiir wissenschaftliche Zoologie, Bd. LXXVI, p. 439-477, pi XXVII, XXVIII. Leipzig. JOHANNSEN, O. A. 1903. Aquatic nematocerous Diptera [of New York]. Bulletin of the New York State Museum, University of the State of New York, LXVIII, p. 328-441, PI. XXXIII-L. Albany, N. Y. 1905. Aquatic nematocerous Diptera, II: Chironomidae. New York State Education Depart- ment, New York State Museum Annual Report, Bulletin LXXXVI, p. 76-327, PI. XVI- XXXVII. Albany, N. Y. Kellogg, V. L. 1902. The development and homologies of the mouth parts of insects. American Naturalist, Vol. XXXVI, No. 429, p. 683-706, 26 fig. The Athenaeum Press, Ginn & Co., Boston. 1905. American insects. 674 p., 812 fig., 13 pi. New York. Kieffer, J. J-, and A. Thienemann. 1908. Neue und bekannte Chironomiden und ihre Metamorrmose . I. Neue und bekannte Chiro- nomiden, von Kieffer. II. Chironomidemetamorphosen, von Thienemann. Zeitschrift fiir wissenschaftliche Insektenbiologie, Bd. 4, p. 1-10, 33-39, 78-84, 124-128, 184-190, 214-219, 256-259, 277-286. Berlin. Knab, Frederick. 1905. A chironomid inhabitant of Sarracenia purpurea, Meiriocnemus knabi Coq. Journal of the New York Entomological Society, vol. XIII, p. 69-73, pi. VI. New York, N. Y. 1914. Ceratopagoninae sucking the blood of caterpillars. Proceedings, Entomological Society of Washington, Vol. XVI, p. 63-66. Washington, D. C. 1914a. Ceratopagoninae s.ucking the blood of other insects. Ibid., p. 139-141. Laboulbene, Alex. 1869. Histoire des metamorphoses du Ceratopogon dufouri. Annales de la Societe • ntomologique de France, 4" ser., Tome IX, p. 157-166, pi. 7. Paris. Lauterborn, R. 1905. Zur Kenntnis der Chironomiden-Larven. Zoologischer Anzeiger, Bd. XXIX, p. 207-217, 15 fig. Leipzig. Long, William Henry, Jr. 1902. New species of Ceratopogon. Biological Bulletin, Marine Biological Laboratory, Woods Hole, Mass., Vol. Ill, p. 3-14. Press of New Era Printing Co., Lancaster, Pa. Lyonet, Pieter. 1832. Recherches sur 1 'anatomie et les metamorphoses de differentes especes d'insectes. Ouvrage posthume. Publie par M. W. De Haan. 5Sop.,54pl. Paris, Roret. AQUATIC MIDGES AND SOME RELATED INSECTS. 6l Malloch, John R. 1915. The Chironomidae, or midges, of Illinois, with particular reference to the species occurring in the Illinois River. Bulletin, Illinois State Laboratory of Natural History, Vol. X, Art. VI, p. 275-543. Urbana, 111. Meinert, Fr. 1882. Om retractile antenner hos en dipter-larve, Tanypus. Entomologisk Tidskrift, Argangen 3, p. 83-86. Stockholm. 1886. De eucephale Myggelarver. K. Videnskabernes Selskabs Skrifter, Argangen III, p. 373- 493, Pis. I-IV. Copenhagen. Miall, Louis Compton. 1895. The natural history of aquatic insects. 8 vo., 395 p., 116 fig. Macmillan and Co., London and New York. Reprinted 1903, 1912. Miall, Louis Compton, and A. R. Hammond. 1900. The structure and life-history of the harlequin fly (Chironomous). 196 p., 129 fig. Oxford. Mundy, Arthur Terry. 1909. The anatomy, habits, and psychology of Chironomus pusio Meigen (the early stages), with notes on various other invertebrates, chiefly Chironomidae. 56 p., VIII pi. Leicester, England. Needham, James G. 1906. The egg-laying of Chironomous annularis. Science, New Series, Vol. XXIV, p. 299. The Macmillan Co., New York. 1908. Notes on the aquatic insects of Walnut Lake, with especial reference to a few species of considerable importance as fish food. In A biological survey of Walnut Lake, Mich., by Thos. L. Hankinson, p. 252-271, 1 pi. Lansing, Mich. Needham, James G., and J. T. Lloyd. 1916. The life of inland waters. 8 vo., 438 p., 244 figs. Ithaca, N. Y. Perris, Edouard. 1847. Notes pour servir a l'histoire des Ceratopogon. Annates de la Societe entomologique de France. 2" ser., Tome V, p. 555-569, pi. 9, No. III. Paris. Pettit, R. H. 1900. A leaf-miner, Chcironomous sp., in water lilies. Michigan Academy of Science, First Report, p. no-Ill. Lansing. Taylor, T. H. 1905. Notice of a chironomous larva. Zoologischer Anzeiger, Bd. XXIX, p. 451-452. Leipzig. Thienemann, August. 1909. Orphnephila tcsiacea Macq. Ein Beitrag zur Kenntnis der fauna hygropetrica. Annales de Biologie Lacustre, Tome 4, p. 53-87, pis. viii, ix. Bruxelles. Tilbury, Mary Ruth. 1913. Notes on the feeding and rearing of the midge Chironomus cayugae Johannsen. Journal of the New York Entomological Society, Vol. 21, p. 305-308. New York. Willem, Victor. 1908. Larves de chironomides vivant dans des feuilles. Bulletin de la classe des sciences de l'Academie royale des sciences, des lettres et des beaux arts de Belgique, p. 697-704, pi. Bruxelles. Zebrowska, A. 1914. Recherches sur les larves de chironomides du Lac Leman. Lausanne. EXPERIMENTS IN THE CULTURE OF FRESH- WATER MUSSELS. By ARTHUR DAY HOWARD, Assistant, I'. S. Bureau of Fisheries. CONTENTS. Part- Introduction 63 Methods and plan of investigation 64 Observations on growth of juvenile mussels 66 Growth in floating crates 68 Growth in aquaria, tanks, and troughs 72 Growth in cement-lined ponds 74 Growth in earth ponds 77 Growth in pens 77 Structure and development of juvenile mussels 78 Habits and habitat of juvenile mussels 80 Discussion and application of results 83 Artificial propagation 84 The cultural method 84 Protection 84 Planting 86 Commercial possibilities 86 Literature cited 87 INTRODUCTION. The coming of fresh-water mussels to a position of commercial importance in America resulted in a special demand for information as to methods of propagating them. In response to this demand the U. S. Bureau of Fisheries undertook an extensive inves- tigation of the commercial fresh-water mussels. This led to the adoption of a method of propagation that promised effectively to increase the supply of mussels. This method, briefly, is the infection of suitable fish with the young mussels in the parasitic stage. These fish are then released to spread the mussels at large under natural conditions. The investigations have been continued for the purpose of extending the application of the methods now in use, the testing of new methods, and to secure more complete information on the life history of the mussels used in pearl-button manufacture. Since Leydig's (1866) discovery that the young fresh- water mussels are parasitic on fish, many attempts have been made to raise them in captivity. No particular difficulty has been experienced in carrying certain species through the parasitic stage, but up until the time of the present investigation there seem to have been no records of the rearing of these under observation through what is called the juvenile stage. 63 64 BULLETIN OF THE BUREAU OF FISHERIES. In aquaria, either balanced or supplied with running water, they did not seem to thrive. Even in tanks out of doors supplied with water from their usual habitat the results were negative. The majority apparently at the very beginning of their free life were eaten by predacious forms, or, if by chance they escaped these enemies, they continued their existence dwarfed. Something in the environment was unfavorable to them. Among European investigators who have attempted to rear young mussels are the following, with the results attained as to time carried under culture: F. Schmidt (1885), 4 weeks; G. Schierholz (1888), 4 to 5 weeks; W. Harms (1907), 7 weeks; and Karl Herbers (I9I3>), about 2 months, or to a size of 3.13 millimeters. In America we have the following records of artificially reared mussels. Lefevre and Curtis (19 12) found a young mussel two years after a plant had been made in a tank. Similar results were attained at the U. S. Fisheries Biological Laboratory at Fairport, Iowa. In this case two mussels, Lampsilis venlricosa (Barnes), were obtained in a pond one year after a recorded plant had been made. In these two instances no observations of the mussels were made in the period between the planting and finding of the mussel at an advanced stage of development. A. F. Shira (report in MS.) reared the Lake Pepin mucket in a balanced aquarium to a size of 4.4 millimeters. As a part of the general plan mentioned above, the experiments described in this paper were carried on to test the possibilities of artificial culture of mussels from the earliest stages up to the mature adult. The studies were carried on at the U. S. Fish- eries Biological Laboratory at Fairport, Iowa, under the direction of Dr. R. E. Coker, in charge of the investigations upon the fresh-water mussels, and later under A. F. Shira, his successor. The author wishes to acknowledge here courtesies extended and assistance rendered in the conduct of these studies to the Crerar Library, of Chicago, for use of their excellent facilities; to Bryant Walker, Detroit, Mich., for assistance in determination of mussels; to Caroline Stringer, Omaha, Nebr., and Ruth Higley, Grand- view, Iowa, for determination of Rhabdocools; to Prof. Edwin Linton, Washington, Pa., for assistance in the determination of Turbellaria ; and to Prof. F. B. Isley, Fayette, Mo., for suggestions of methods. METHODS AND PLAN OF INVESTIGATION. After some little experimental study of developing mussels it was realized that there must be some vital deficiency under artificial conditions to account for the many failures in attempts to raise mussels. It seemed that a promising line of attack in solu- tion of the problem would be to find some way which would depart from the natural habitat only so far as the necessity of mechanical control demanded. To rear at least one brood of the young seemed to be an objective of prime importance. Success in this would answer some unsolved questions as to growth, as well as furnish a starting point for more artificial methods if these were desirable. In our situation, where we take the mussels from the Mississippi River, the most practicable solution that offered itself was a floating crate containing baskets made of wire cloth of sufficient size to hold the fish and of a mesh small enough to retain the miscroscopic mussels. A crate held at the surface accommodates itself to' the frequent rise and fall of the river, is convenient of access, and removes the young mussels from many of their enemies prevalent at the bottom. Another advantage of a surface location is the fact that the precipitation of silt there is at a minimum. The first crate used (fig. 74) was constructed Bulu U. S. B. F., 1921-22. Fig. 5S.— Improve: 1 float employed in experiments in mussel culture showing one ol crates on "deck" opposite 11 - berth. Fig. 59.— iJanie float as in fig. 58 anchored in the river showing three crates in position supported by adjustable iron hangers which are visible above the float. Bull. U. S. B. F., 1921- Fig. 60. — A crate of improved pattern showing outer screens of 1-inch mesh and inner detachable screens of copper cloth, one of which is completely removed and the other turned in to show manner of insertion. Infected fish are held in the crates until the parasitic mussels are shed. The copper cloth prevents the escape of the mussels in the early minute stages. Fig. 61. — Concrete ponds used for mussel culture experiments. In the dry pond on the left is shown the method ol dividing into smaller units by means of screens. Bridges are shown over the two ponds on the right These furnish shade lor the fish and prevent their jumping over the screens as well as serving the purpose of bridges for the operators when seining the fish. Earth ponds and shed-covered troughs appear in the background. CULTURE OF FRESH-WATER MUSSELS. 65 from a floating fish car to which were added barrels to give greater buoyancy. Four baskets (fig. 75) of rectangular shape, il/i by 2% feet, were made to fit inside. These consisted of a framework of galvanized iron attached to a bottom tray of the same material , both of which were painted with two or three coats of asphaltum to prevent corrosion. On the frame was stretched copper cloth 100 meshes to the inch. In the baskets were placed the fish infected with mussels. In order to reduce the length of time necessary' for retaining the fish in such narrow confines, they were not placed in the crate until a few days before the end of the parasitic period of the mussels and were removed as soon as the mussels were shed. Plants of the following species of mussels were made from time to time: The washboard, Quadrula heros (Say); the mucket, Lampsilis ligamcntina (Lamarck) ; Lake Pepin or fat mucket, L. luteola (Lamarck) ; the yellow sand-shell, L. anodontoidcs (Lea) ; and the pimple-back, Quadrula pustulosa (Lea). Modifications of the floating crates were introduced from time to time with a view- to improvement of conditions for both fish and mussels and economy of operation. The latest form of float (figs. 58 and 59) adopted is made from two cedar telegraph poles held apart by crossbeams, 4 by 4 inches, at a distance sufficient to suspend lengthwise seven crates having dimensions 3K by ilA by \l/i feet. The crossbeams are placed at 4-foot intervals, and to them are bolted strap-iron hangers by means of which the crates are suspended. On the crossbeams over the telegraph poles are nailed 2-inch planks, 10 inches wide, forming a walk on each side the full length of the float. From this walk two operators can conveniently raise the crates in which the infected fish are placed. A float of this form was devised to protect the crates from wave wash and to give greater stability in stormy weather, when a shorter and smaller float would be tossed about. The crates or baskets (fig. 60) in the improved type are constructed of cypress lumber, being made as light as the demand for strength permits. The bottom or floor is made of matched lumber and tight enough to prevent the escape of the microscopic mussels. The superstructure consists of a framework, on the outside of which is nailed galvanized screen of one-fourth-inch mesh. Fitted inside of the frame and outer screen are the inner screens, which consist of wooden frames to which copper cloth is fastened with copper tacks. The inner screens are removable, held in place by buttons or other locking devices. The removable screens are so provided with overlapping strips as to give a joint sufficiently tight to prevent escape of the small mussels. In the use of removable wire screens the following objects were in view: It facilitated the cleaning of the copper cloth and provided an opportunity to enlarge the mesh of the screens as the mussels increased in size, thus giving them a freer flow of water and economizing the higher-priced fine-meshed copper cloth. The use of wood instead of metal as employed in the first baskets provided distinct and obvious advantages. Metal was objectionable wherever the young mussels might come in contact with it, was less durable, and was more expensive. Metal cloth could not be dispensed with entirely, because other fabrics will not last under water. The increase in size of the crates or baskets was of marked advantage in providing more room for the fish, thus permitting use of greater numbers with less mortality. The whole assembly of float and crates provided a convenient and economical means of operation greatly improved over the first crates, in which the raising of the much smaller baskets was necessarily done from boats and in comparison was awkward 66 BULLETIN OF THE BUREAU OF FISHERIES. and difficult. The improved float because of its form is more readily towed and handled in the current than the very much smaller floats first constructed and may be easily drawn out of the river by a team of horses when necessary, as for winter quarters. Other methods were employed in the investigation and, in a way, carried parallel for comparison to test the possibilities of the equipment already installed at the biological laboratory at Fairport. These were aquaria and indoor tanks and troughs, cement ponds, and earth ponds. Each of these was supplied with running water except in the case of special experiments with balanced aquaria. The water for the most part was taken from a reservoir receiving its supply by pump from the Mississippi River. Thus the water was, as a rule, practically unmodified. In some experiments with balanced aquaria filtered river water was used in order to eliminate the predacious animals which prey on the early stages of the mussels. For the same purpose, as well as to reduce the amount of sedimentation in river water, specially devised settling tanks were employed for supplying aquaria. The cement ponds (fig. 61) were of reinforced concrete 50 feet long, 10 feet wide, and averaging 2% feet deep, having perpendicular sides and constructed for the temporary retention of fish. An accumulation of mud and a specially prepared bottom of gravel, together with an abundance of water plants, furnished conditions which proved suitable for some of the most delicate species of fish. It was assumed that these conditions were as suitable to the needs of the mussels as they could be made under the circumstances. The earth ponds were from 41 to 61 feet long and 24 feet wide, varying in depth from 4 inches at the intake pipe to 4 feet at the well. An abundance of water plants furnished food and shade for the fish. The cement and earth ponds as compared with the floating crate do not so readily furnish the means for frequent observations of early stages. In using them it was planned to test their possibilities of rearing clams by a comparison of older juveniles grown in them. Thus the probable disadvantage of frequent disturbance necessary in making observations on younger juveniles would be avoided. Plants of young mussels were made from infected fish in each of the culture devices mentioned. A modification of the cement pond was used in one instance for the purpose of securing a current comparable in rapidity to that to which the river mussels are accustomed. A flow of 50 gallons per minute was supplied to a trough 16 inches wide by 12 inches deep by 50 feet long, giving a current of 0.1 mile per hour. This is by no means equivalent to the 2 to 3 miles per hour of the Mississippi, but was planned to imitate the conditions of the river more closely than that of the ponds in which the flow is inappreciable. OBSERVATIONS ON GROWTH OF JUVENILE MUSSELS. In this investigation studies upon growth have been made with a view to securing data upon general conditions as well as upon the more specific methods of rearing under artificial environments. The species tested were chiefly heavy-shelled river mussels, which include most of those that are considered of commercial value, as distinguished from the thin-shelled pond-dwelling forms. The latter apparently offer no particular difficulties. The most complete results were obtained from a species which selects a habitat somewhat intermediate between these extremes, in that it dwells in lakes and 64 65 Lake Pepin rmtcket, LamPsWs luteola (Lamarck), at various stages from young to adult. Figs. 62 and 63.— An adult gravid female, age about three years. Natural size. The right shell (fig. 65) has been removed to expose the viscera. At m is shown the marsupium in which the young arc carried from theeggtotheglochidial stage. Mussels grown under control in the experiments here described equaled this one in size at the age of first breeding, two years and three months. Fig. 64. — Glochidia or parasitic stage in the young as they appear on leaving the parent mussel. One with valves open may be seen at the middle left margin of the field. Photomicrograph. X27. Fig. 65.— Gill fdaments of a black bass infected with the glochidia of Lam psilis luteola 14 days after infection. Photo- micrograph. Fig. 66.— Lell to right: Young mussels oi one, two , three, four, five and one-half months of age, respectively. Natural size. CULTURE OF FRESH-WATER MUSSELS. 67 the quieter waters of rivers. This was the Lake Pepin mucket, Lampsilis luteola (Lamarck). In this mussel a surprising amount of growth took place during one season. The other species fared less well, in some cases apparently surviving only a short period. Since satisfactory positive results were attained with L. luteola, the experiments with this species furnished a basis for comparison of the methods in reference to their influence on growth. As the results with this species may have been largely due to inherent qualities, a short account of its natural history and development seems desirable. The Lake Pepin or fat mucket, as it is generally called, has a shell of excellent quality and possesses a good reputation as a pearl producer. It is probably the most widely distributed of the fresh-water mussels used commercially. Simpson (1900) gives its distribution as follows: Entire Mississippi drainage southwest to the Brazos River, Tex.; St. Lawrence drainage; entire Dominion of Canada east of the Rocky Mountains. The author has found it under the most varied conditions — from those of the marshy slough of a small creek to the deep waters and 'wave-beaten beaches of the Great Lakes. These observations would indicate that the form is adaptable to widely varying environment and would, perhaps, explain its thriving condition in this experiment where other species fared less well. 67 Figs. 67 and 68. — A young mussel on: to three days after leaving the fish, in outward form like the original glochidium but internally (tAt is. inside the shell) showing organs developed. Drawn with a camera lucida. X 140. 67.— Ventral view with valves apart, from specimens stained and cleared. 68. — Side view; a narrow growth of the new definitive shell may be seen bordering the glochidial shell. This species belongs to the bradytictic group called winter breeders. The glochidia are produced in the late summer or fall and are carried through the winter in the dis- tended marsupial gills (see fig. 62) of the female. The glochidia (fig. 64) are favorable for infection, because their comparatively large size makes it easy to follow the progress of infection (fig. 65) and subsequent shedding. Unfortunately, the number of glochidia produced is relatively small. The gravid mussels for this experiment were obtained in Lake Pepin, Minn., about May 15, 1914, and shipped to Fairport, Iowa, by express. On May 21 ripe glochidia were taken from three of the live mussels for the experiment. Some dozen different species of fish were infected and of these, six proved susceptible and carried the young mussels through their metamorphosis. Before the young mussels began to be shed eight infected largemouth black bass were placed in basket No. 2 of the floating crate. Some very rough weather followed, tossing the crate about in such a way as to make 6S BULLETIN OF THE BUREAU OF FISHERIES. the conditions severe for the fish and killing five of the eight. On June 10, 20 days from the date of infection, most of the young mussels were found to have been shed from the three remaining fish. On the same date shedding was found to have taken place from infected fish placed in the cement ponds and aquaria. The time of shedding for the earth ponds was not observed. The young mussels were secured at this early stage from the aquaria. At the time of shedding there is apparently no growth of shell beyond that of the original glochidium, but the young mussel (see fig. 67) internally has for the most part the organs of the adult in contrast with the simple structure of the larval glochidium. Growth of the shell begins at once (see figs. 67 and 68), as shown, and in the figure a narrow border of the new shell is already visible. GROWTH IN FLOATING CRATES. Two weeks after obtaining the plant of young mussels from the bass, evidence that they were thriving in the crate was obtained. A small sample of sediment from the bottom revealed some half dozen or more. These had already a considerable growth of shell, the largest having an increase in surface of at least three times the size of the original glochidium (see fig. 69). At various intervals throughout the summer and autumn the author readily obtained specimens, making observa- tions on rate of growth and preparing material for studies of development. Figure 66 shows individuals illustrat- ing the amount of growth from month to month. The last examination was made about November 20, when the whole plant in the basket (fig. 75) was photographed under water. Later they were removed from the mud, a census was taken, and more photographs were made (fig. 70) . After completing such observations as were feasible upon the whole plant of living mussels they were returned to a crate and placed in a pond to spend the winter. The series shown in figure 66 represents about the average ' growth from month to month. These, with the exception of the third, were removed from the basket on the dates given in Table 1, page 69. By inspection it is obvious that the rate of increase in growth as represented by these is not uniform throughout. This is due partly to the fact that in some cases small numbers only were removed at a time. In this way the average size was not secured in each instance. In one case only was a voluntary selection possible, and this was the last, made from several of nearly equal size. The specimen in the series for the second month (fig. 66, second from left) was probably smaller than the average. It will be noted that by months the increase is much more rapid at first, so that the rate is a decreasing one. Fig. 69. — A juvenile mussel 15 days after the beginning of free-living stage, or about two weeks older than that of figure 68. View of right side. Drawn with a camera lucida. X 140. 1 These were selected at random in most cases and so probably approximate the average. Bull. U. S. B. F., 192 ^•^IK^ #1 * r oHV ^Hiu ^^k- <*V WWW WW Fig. 70. — The contents oi one propagation basket at the end oi a season's growth of five and one-half months. The mussels were of microscopic size when shed in the basket by the fish. The arrangement in series shows the amount ol variation at this age under the prevailing condi- tions. Reduced to five-twelfths natural size. CULTURE OF FRESH-WATER MUSSELS. 69 Table i. — Increase in Length op Juvenile Mussels in a Floating Crate During the Growing Season of the First Year, 1914.' Date collected. Lernnh. Increase in length. Pate cullected. June 10.. July 18 . Aug. 17. Mm. Per cent. 3-95 8.8 1,580.0 209.5 Length. Sept. 12. Oct. 10.. Nov. 24. Mm. 22. i 27. 2 32 Inctease in length. Mm. 9-3 4.9 48 Ptr cetit. 71-5 21.0 17.6 ' The mussels measured were taken at random, with the exception of the last one. which was selected as the maximum. The length of 32 millimeters at the close of the season (1914) is one hundred and twenty-eight times that of the original juvenile at the beginning of free life. This cer- tainly compares favorably with the total length of 3 millimeters reported by Herbers (1913), which was the largest in his culture of juveniles, while the mussels in the experi- ment of which this paper treats were still alive and vigorous at the end of the season. Figure 70 is a photograph of the contents of a basket at the end of the season reduced to five-twelfths natural size. The mussels range in size from 32 to 15.5 millimeters. The variation is considerable, but it should be noted that less than 27 per cent are under three-fourths of the maximum size. The last mussel in the series, and one of the smallest, is deformed, probably restricted in growth by lodging in a crevice. Two more small mussels were found when the mud was passed through a sieve. Of these one measured 14. 1 millimeters and the other the remarkably small size of 6.9 millimeters. The latter was living at the time of removal from the river. These few cases of dwarf- ing are doubtless due to lodgment in unfavorable locations — under crowded conditions — in the basket. During the last month, from October 20 to November 20, a record of growth was taken to determine to what extent growth takes place as the water temperatures fall. Measurements of 10 mussels from the basket were taken. After marking and measuring each they were returned to the crate. The results are presented in Table 2, following which are given the water temperature averages, maximum, minimum, and range for the period. It will be seen that the growtli for the period was very slight. Table 2. — Increase in LENGTH of Juvenile Mi SSELS In a Floating Crate During the Last Month of the Growing Season, 1914. Specimen number. Length. Increase in length. Nov. 20. .!/»,:. Mm. Mm. O. I 93-3 23 4 24. 6 24 S . 2 24.2 2 1 < •3 25 2 25 < •3 11.3 91.3 .0 39. 0 29. 0 . 0 Length. Specjmen Dumber. Oct. 20. Average. Mm. 95-4 23 3 29 4 93 s Nov. Mm. 95 8 23. 6 29 7 34. o Increase in length. o. 4 ■ 3 • 3 • 5 water TEMPERATURE For period of MEASURED GROWTH. Average : °f. For 1 1 days, Oct. 20 to 31 54. 9 For 10 days, Nov. 1 to 10 50. 9 For 10 days, Nov. 1 1 to 20 43. 2 For whole period , Oct. 20 to Nov. ->o 49-2 Maximum for whole period , Oct. 20 to Nov. 20 60 Minimum for whole period, Oct. 20 to Nov. 20 32 Range for whole period, Oct. 20 to Nov. 20 28 7631G0— 22 2 7° BULLETIN OF THE BUREAU OF FISHERIES. On consulting the temperature averages the assumption is natural that such growth as occurred took place before the temperature fell. It is obvious that for the whole period (Oct. 20 to Nov. 20) growth was much less than in the warmer months. Compare the maximum of 0.5 millimeter for the period with the growth of 4.9 millimeters shown in Table 1 (p. 69) for the period from Sep- tember 12 to October 10. The desire to secure these records resulted in the postpone- ment of the date for removal from the river until a time dangerously late. On the night of November 19 ice floes bore down on the crate. Only by the rarest good fortune was the whole plant saved. The ice instead of destroying the crate or carrying it away landed it on shore, where the mussels were extricated without injury. A count of mussels grown in the basket follows: Alive in basket Nov. 20 „a Dead in basket Nov. 20 6 Removed from basket June 25 to Oct. 30 -r Total living for season 3l- As the original plant from the three surviving bass was an estimated 2,400 juveniles, it would give a survival of something better than 8y$ per cent. The mortality would be indicated by the difference in the figures of the original plant and the final crop. Observations upon growth were continued during the second and third summers. The results of measurements taken from month to month on marked mussels are indicated in Table 3. In figure 71 is plotted the increase of growth per month for 18 months, with the graph of the average water temperature. The data are taken from observations on mussel No. 3 in Table 3, as the record for this mussel is the most complete. Absence of growth from November to the middle of April, though not shown in the table, was observed and is supplied in the graph. Lack of observation for May, 19 15, is supplied from another brood of the same age giving an approximation to the true figure sufficiently close for our purpose. This would give the following increases in millimeters for each month: May, 1.7; June, 6.1; July, 9.1; August, 7.1; September, 3.9; October, 1.5 ; May (1916), 1.9. The growing season seems obviously to be correlated with the rising temperature of summer. In a general way, doubtless, it is dependent upon the phytoplankton, and the plankton is controlled to a large degree by the tem- perature (Kofoid, 1903, p. 572, par. 18). Table 3. — Growth op Mussels in a Floating Crate in the Second and Third Years. Specimen. Length in millimeters. Weight Num- ber. Sex. Mark. Apr. 19. 1915- June 10. 191s- June 22. 1915- July 22. 1915- Aug. ii, 1915- Sept. 25, 1915. Oct. 26. 1915- May 3i.lAug. 15, 1916. 1 1916. Oct. 6, 1916. grams, Oct. 6, 1916. I II III IV V VI VII VIII IX X 31-6 30. 6 n-s >i. 7 25. I 29- S 26. 4 21. I 14. I 26. O 35-3 36.9 35-3 29.3 48. 5 44.4 39- 6 47. 0 53-9 SI- S 46.9 S2-4 57-5 55-4 Si- 8 57-4 53-8 S&9 52- 7 58.8 5S-S 59-6 72. 0 ■7.-8 74.6 74-9 73- 0 53-5 67.0 49.6 57-8 3 4 5 6 Male 7 g 33-4 27- 7 3°- 7 SI- 5 43-4 47-2 SI- 4 S5-7 5S-9 58-2 516 52- 7 58.6 78.8 69-6 65- 0 80.0 S6-5 57-6 44-6 61.6 29. 1 41.9 9 Female Male 55- 8 51. I 1 No growth indicated here. Decrease perhaps due to breaking of periostracum. CULTURE OF FRESH-WATER MUSSELS. The second summer yielded one individual measuring 62.8 millimeters (2.47 inches) in length, the maximum, and many over 55 millimeters (2.16 inches) in length. From one of these were cut 16-line buttons 2 lines thick (see fig. 72). Although this is not a favorable size for cutting, the fact that the shell in two seasons' growth is almost suitable for commerciaUuse is of significance and far exceeds expectation. Growth during the third summer, when the adult stage was attained, determined by the first breeding, reached a maximum length of 85 millimeters, weight 63.1 grams, in the male, and a length of 77 millimeters, weight 66.5 grams, in the female (gravid). Length, average male 79.1 millimeters, average female 71.5 millimeters. The growths of the 1915 brood during their second summer compared with that of the 1914 brood for their second summer show a very striking difference. Although the 1915 brood 1915 •Jon. Feb. Ho Mfly Jun< July Auj. Sept. Oct 1916 Nov. Dec, ijan. Feb. Mor. Apr. May June MM. A 1 / p\ 1 f- / / \ \ 1 1 / / 1 \ \ 1 — /- ' / \ / 1 \ \ 1 1 1 1 \ 1 1 1 / 1 1 \ \ 1 1 1 1 J \ \ / / / / \ v / / / "Temp 70 GO 50 40 30 FlG. 71. — Growth of a fresh-water mussel in relation to temperature: . mean monthly water temperature (F.) in the Mississippi River at Fairport, Iowa; , monthly increase in growth of a IrcsliwuUr mussel in its second year, in millimeters. Zero represents the line of no growth omi the coordinates represent the increase for each month taken separately. (See p. 70.) began the second summer very much smaller, averaging n.6 millimeters in length, compared with 25.7 millimeters for the 19 14 brood, at the end of the season the former had increased 475 per cent while the latter had gained only 212 per cent.2 This dis- parity in growth brought the brood of 19 15 to a size — their second year — equal to that of the 1914 brood at the end of their second year in the face of a large handicap. This difference may be ascribed to difference in season which is, perhaps, the simplest explanation. The summer of 191 6 had higher water temperature, higher water stages, and less wind than usual. Flood stages, generally speaking, have been found unfavor- able to plankton production as determined by Kofoid in the Illinois River (Kofoid, 1903). The rapid growth this season occurred on falling stages but at an unusually sustained high level. As this high level was not due to local precipitation, it would seem that the conditions were consistent with (an assumed) high plankton production at the point of observation. The absence of wind as an important cause of turbidity would be favorable to the feeding of mussels. 'The small size of the 1015 brood was due to a late planting and partly, doubtless, to a less favorable growing 9e " ~2 BULLETIN OF THE BUREAU OF FISHERIES. Another explanation of this difference is the possibility of the existence of an in- herent controlling factor in growth, whereby an average growth may be obtained by the end of the second year. That is, in the case of a small first year's growth there would be compensative additional growth the second year. This phenomenon is not of uncommon occurrence in organisms. Barney (,1922) in studies of growth in terrapins finds "runts" selected in I9i3 in 1917 exceeding in growth larger selected individuals of 1913. A plant of yellow sand-shell, LampsiUs anodontoides (Lea\ was not as successful in numbers, but yielded three juveniles which survived the summer, and the largest attained a size of 8.3 millimeters in 6 months. The second summer it attained a length of 41 millimeters and a weight of 5.8 grams.3 GROWTH IN AQUARIA, TANKS, AND TROUGHS. A plant of juveniles from two bass, Micropterus salmoides, and one calico bass Pomoxis sparoidcs, was obtained in a rectangular glass aquarium. The young were readily found within a day or two after their escape from the fish, but later than this only shells of the earliest stages could be found. It is possible that the absence of growth in this instance was due to the destruction of the young mussels by enemies to be men- tioned later. Another test of the possibilities of aquaria was made by placing in them rapidly growing mussels taken from the floating crate at a more advanced stage and comparing their growth with the growth of mussels remaining in the crate. The growth in milli- meters and the increase is shown in Table 4. While in the aquarium the same individuals were measured each time, the measurements of growth in the crate were not based upon particular mussels, but upon different examples taken as representative of the lot. Observations were made in this way, because the recovery of marked mussels in the crate entailed danger of too much disturbance to the whole plant. Table 4-Comparative Growths op Juvenile Mussels ix Aquarium and in Floating Crate. Place of growth. Length in milli- meters. Increase in milli- meters Aug. 17. Place of growth. Length in milli- meters. Increase July 27. Aug. 17. July ,;. Aug. 16. meters Aug. 17. Aquarium I ?■ 7 Lost. 4- 2 ' | ( (6) (S- S) 1 0) 13 13.8 10. I 7 ,,. 7-3 7.1 The figures, although only approximate, are sufficientlv accurate to represent fairly the great difference in growth that has been shown in many experiments in other ways. The total growth from the beginning of the juvenile stage, June 10 to August 17, is 7 millimeters for the largest of three mussels placed in the aquarium for three weeks, while it is 10. 1 for the smallest of three taken from the crate on the same date. This gives a difference of 3.1 millimeters where the influence of the aquarium is exerted only for the relatively short period of three weeks. 'Attention is called to the employment of thegarpikes, Upisostos osseus (L.)and L. i,hto«„mus Raf. . as hosts for the mussels m th.s expenment. These are the only fish found of many tested which wil. carry the giochidia of this mussel (Howard." T4b)" Bull. U. S. B. F., 192 1-2: Fig. 72. — Mussels at the ane of one year and four months, and buttons cut from them. These mussels were the product ol artificial infection and rearing by the crate method. Photographed by J. B. Southall. CULTURE OF FRESH-WATER MUSSELS. 73 Young mussels of various sizes from one-half inch up placed in tanks and aquaria indoors at various times have shown a negligible amount of growth. Likewise, nega- tive results have been secured in plants of young mussels made in the following types of aquaria indoors supplied with flowing river water which was unmodified so far as known: Wooden tanks or troughs, tanks and troughs lined with galvanized iron painted and unpainted, and cement tanks and troughs. Two systems of water supply have been tried. In one the river water was pumped direct, in the other it was pumped first into a reservoir, from which it was distributed by gravity flow. Later results seemed to indicate a difference to be discussed below under cement ponds. In order to eliminate the destructive turbellarians and other predacious forms that might be introduced with the water, balanced aquaria, large and small, filled with filtered river water were tried. Here, too, the mussels survived for only a short lime. More recent experiments in rearing young mussels in a type of container of com- paratively small dimensions have been conducted with considerable success, first by F. H. Reuling (1920) at Fairport and later by the author and others. The conditions were so different from those of the experiments just described that they should throw light on controlling factors in the development of juveniles. Their convenient size made them admirably suited for experimental pi r] OSes where a considerable number of units are required. The equipment consisted of galvanized-iron troughs 14 by 8 inches by 8 feet, painted with asphaltum. The troughs were protected from the sun by a shed roof of wood; otherwise they were uninclosed. (See center background of fig. 61.) The water supply was derived from the surface of a pond containing vegetation. This arrangement yielded water of comparative clarity even when the river supplying the pond was turbid. The point of intake at the surface probably insured a minimum of animal enemies, such as Turbellaria, which might prey on the mussels. Additional precautions were taken against enemies by further straining through ordinary cloth and later close-meshed metal fabric. Broods of Lamps-Ms lulcola and some one-half dozen L. liganu niina, the river mucket, were reared in these troughs the first summer. In 1919 successful results were secured with three species approximately as follows: Yellow sand shells, L. ano- dontoides, 2,000; Lake Pepin mucket, L. lutcola, 3,000; the river mucket, L. ligamentina, .500. The dwarfing effect observed in aquaria and tanks indoors is a condition the causes of which have not been entirely determined. There is reason to suppose that reduced light and excessive precipitation of silt are possible factors, assuming that the water supply is the same as that of the river, ponds, or out-of-door troughs. Any such assumption is unwarranted, however, until comparative determinations of water condi- tions and contents have been made. Lack of growth suggests that the plankton, sup- posedly the principal food of the mussels, or other elements are for some reason wanting. The following evidence indicates the nature of some of these constituents which con- ceivably may be lost in part from water standing in reservoirs. Detritus, including dead organic matter, forms a considerable proportion of the food of mussels, according to A. F. Shira and Franz Schrader. (Coker, Shira, Clark, and Howard, 1921, pp. 88 and 93.) Wilson and Clark (19 12), in the examination of the stomach contents of river mussels, find a proportionally small amount of plankton 74 BULLETIN OF THE BUREAU OF FISHERIES. combined with what is apparently a larger quantity of nonliving organic and inorganic material appearing like the mud in which the mussels are embedded when in their natural habitat. Mussels are supposed by some to act as scavengers in consuming sewage. The evidence indicates, however, that, as a rule, they flourish better in waters of natural purity. (Linnville and Kelly, 1906.) It seems not unlikely that mussels may derive considerable nutriment from substances in solution. Churchill's (1915) experiments on the absorption of fat by mussels seem to support such a view. Consideration of the finely balanced conditions found necessary for the welfare of other lamellibranchs, including marine clams, to the growth of which considerable study has been given, removes any wonder at negative results with fresh- water mus- sels that have been subjected to highly artificial environments of aquaria and tanks. Complete success in the use of aquaria and such more or less artificial containers can hardly be expected until the factors of growth and their control are more thoroughly understood. GROWTH IN CEMENT-LINED PONDS. The cement ponds (see p. 66 and fig. 61), because of their location, size, and shape, were found very convenient in the experimental work for temporary holding of fish. The perpendicular sides permitted of ready subdivision by screens and easy control of fish, such as removal, transfer, etc. For the planting and culture of juvenile mussels, how- ever, their usefulness is still somewhat a question. Many unsuccessful trials led to the assumption that the cement bottom and sides presented an environment unnatural and unsuited to the life of the mussel ; but later results seemed to indicate that by proper control of conditions in them fair results might be obtained. Variations in bottom were tested, together with changes in depth and flow of water, in order to take into account the special needs of given species so far as known. The kinds of bottom employed were gravel, sand, mud or loam, and the uncovered cement. The gravel, sand, or loam were evenly distributed 1 to 3 inches deep over the cement. In addition to this a greater or less deposit of silt always accumulated from the water, the maximum precipitation occurring at the end where the supply pipe entered. The plants of juveniles were made from their fish hosts with the following species of mussels: Lampsilis luteola, L. ligamentina, Quadrula plicata, and Q. pustulosa. After one plant of L. luteola on mud bottom at the end of the growing season in November, 19 14, an examination was made to determine the results as to growth. The whole bottom contents of the pond were passed through a sieve of 3-millimeter mesh. Two mussels only were present out of a plant of several thousand. These measured only 1 1.4 and 15.3 millimeters, respectively, and the appearance of their shells gave evidence of unfavorable conditions. Many tests with the different species were made on a bottom of sand or mud. Another variation tried was the narrow cement pond in which large plants of the pimple-back mussel, Quadrula pustulosa, were made. In these ponds, as has been described (p. 66), a current of water over gravel and sand was kept up during the growing season. There was no opportunity for fish to disturb them, as the host fish (channel cat, Ictalurus punctatus) were removed as soon as the mussels had been shed from their gills. CULTURE OF FRESH-WATER MUSSELS. 75 Absolutely negative results were obtained from these experiments, as no trace of mussels could be found in screenings from a series of sieves in which the minimum mesh was 2 millimeters. (There is no doubt that the presence of any mussels approaching normal growth of two seasons would have been revealed by this search.) In these ponds normal aeration of the water and sunlight were more certainly provided for than in tanks and aquaria indoors. In contrast with these results, largely negative, was a plant of Ouadrula pustulosa, in which the outcome was more satisfactory. In one pond, in its first year used — i. e., the first year the cement was submerged (19 13) — infected fish were placed in the lowest divi- sion— i. e., nearest the outlet and farthest away from the inlet pipe. This division was reserved for channel catfish for the purpose of simplifying the history of this section in case any results were obtained. The pond as a whole was employed as a stock pond. A continuous supply of water was kept up summer and winter with a view to giving any mussels that might be obtained opportunity to reach a size that could readily be found. During four years the water was drawn down only a few times. On these occasions the lowering of the water was not allowed to an extent that would be injurious to any mussels that might have started. Only a cursory examination was made for mussels that might have reached a size to be readily detected. Purposely the treatment of this pond was varied from that accorded to the other ponds which, one or two years after plants had been made, were subjected to close inspection by sieving of the bottom soil. Had the same regimen been followed in this case the young mussels would certainly have been found even the first year, and it was an odd chance that the mussels prospered in this one pond where the "let-alone policy" was carried out. As this policy was different from that accorded to all ponds only in respect to the second to fourth years of growth it had no particular bearing upon the question as to how a set was obtained the first year. In seeking an answer to this question we may find a clue by considering wherein the conditions differed from the other ponds. In respect to two features, or rather a combination of two (possibly more, of course), the conditions here seem to have been unique for this type of pond. In the first place the division in which the catfish were held was practically free of bottom soil, there being an exceedingly thin layer only, if any, on the cement. In the second place, this division was farthest removed from the intake pipe, around which there was con- siderable subaquatic vegetation, with the result that the water reaching the lower end of the pond was comparatively free of silt which had been unloaded in the upper division. It is pretty certain that juveniles of many species in the earliest stage can not thrive where silt is precipitating rapidly, and it is quite probable that certain species of Nai- ades, like some marine pelecypods, require a clean bottom and possibly a hard substra- tum. It is somewhat difficult to avoid silt precipitation in ponds supplied with water pumped from a turbid river. In this case the form of the pond, the vegetation, and the position of the mussels presumably brought about the result. Another probable factor in the successful "set" was the "newness" of the water supply system and the consequent nonestablishment of predacious species which are found under usual pond conditions. Rhabdoccels are abundant in the ponds but not in the river water. Since the reservoir which supplies the ponds was filled first only the 76 BULLETIN OF THE BUREAU OF FISHERIES. previous fall and this pond was filled for the first time a few days before the plant was made, it seems likely that rhabdoccels and similar enemies had not yet become established. The number of successful sets observed in the case of newly established ponds (see earth ponds) leads to the conclusion that this factor of "newness"4 may be very important. In Table 5 below are given the measurements of 10 of these shells, including the largest and smallest. There is given the increase per year as indicated by the winter rest line. Table Irowth of 10 Mussels Quadrula Pustulosa, During Four Years in a Concrete- lined Pond. Specimen number. Yearly growth in millimeters. Total length in milli- meters. Specimen number. Yearly growth in millimeters. Total length in milli- meters. 1913 1914 1915 1916 ■913 1014 1915 1 1916 4-3 4.6 S 4.8 4-3 4-5 6.2 7-2 6. 1 3-6 3-8 6. r 6.2 6.8 5-8 6.4 5 S 5-3 S-6 S- 7 3 2 4-5 4- 7 22 24. 2 22.6 IS 17.6 20.3 4 4.8 4-4 3-7 6-5 6.9 5-3 4.6. 5 4-4 5 4 2-3 •• 3 16. 4 15.6 10 6 4-44 563 5-5' 4.21 '9-79 The largest mussel of this series reared in a pond is considerably smaller than a mussel of about the same age grown in the river, as shown by the following figures: Pond grown, length, 24 millimeters; weight, 1.9 grams. River grown, length, 28 milli- meters; weight, 4.6 grams. The retarding effect of the artificial conditions is obvious enough in this comparison, where the advantage of selection is all in favor of the pond- grown shell and in which the river-grown shell is a few months younger. In the summer following the discovery of this "set" of juveniles experiments were carried out to determine if the results could be repeated. The conditions as to bottom and clarification of water and source of water supply were made to coincide as closely as possible with those of the successful ' 'set." In one respect only as far as known was there a difference, namely, in regard to the factor of "newness" or absence of pond conditions. The water was taken from the same reservoir which, having been in use four years, had in a measure acquired the characteristics of a standing body of water. This difference was realized, but it seemed best to make use of the established system of supply as long as its suitableness was not disproved. Three species of mussels were used and several plants made with each. These species were Lampsilis ligamentina, L. anodonloides, and L. hdeola. The results were negative except with L. luteola, which, as indicated elsewhere, is not a typical river mussel and has yielded successful sets in almost all instances under the conditions prevailing in the ponds at the Fairport laboratory. These results would seem to indicate at least that the conditions provided were not decisive factors in the one successful set of Quadrula pustulosa, and that possibly the one factor in question, namely, the water supply, is the one which was responsible for success or failure. A review of the results attained in this type of pond, with its successful plants among the failures, holds out some hope still for the solution of the problems of rearing the true river mussels. The line of procedure indicated would seem to be the provision of a water supply direct from the river and a rigid exclusion of established pond condi- 1 The condition of the water supply beiore typical pond conditions have time I" develop. Bull. U. S. B. F., 1921-22. qjl <£> *»> ffe H^bWwO^L Fig. 73. — Juveniles of 20 species of mussels found in the artificial ponds at the U. S. Fisheries Biological Station within two years from the time of construction of the ponds. All reproduced natural size excepting the two right-hand figures in top row which are reduced one-hall. (Photo- graphed by J. B. Southall.) Reading from left to right these mussels are: Top row: Anodonta imbecillis, Anodonta corpulent a, Anodonta suborbiculata, Arcidens confragosus. Second row: Strophitus edentulus, Symphynota com plan ata. Lampsilis alata, Lampsilis laevisshna. Third row: Lampsilis caPax. Lampsilis gracilis, Lampsilis veniricosa, Lampsilis luteola. Fourth row: Lampsilis subrostrata. Lampsilis parva, Lampsilis ligamentina, Obovaria -Ihpsis, Fifth row: Plagiola donaciformis, Obliquaria refiexa, Quadrula plicata, (Juadrula undata. CULTURE OF FRESH- WATER MUSSELS. 77 tions. It should be possible to maintain such conditions by thoroughly cleaning the walls and bottom each season and, so far as possible, excluding pond plants and animals during the critical period when the young mussels are escaping from their hosts.5 GROWTH IN EARTH PONDS. A large plant of Lampsilis luleola was made in an earth pond in 19 14 from crappie of two species, Pomoxu annularis and P. sparoidcs, and the sunfish, Lepomis pallidas. The following spring an examination of the bottom yielded some eight mussels, the largest 24 millimeters in length, the smallest 12. The growth was not as great as that in the floating crate, but compared favorably. The number surviving, however, compared with the thousands introduced into the ponds by means of the fish, was disproportionately small. In lowering the water level of the pond there were found afewsheepshead, Aplodinotus grunniens, whose presence was quite unexpected and contrary to the plan of the ex- periment. As this is a mussel-eating fish, its presence might explain the disparity in numbers of the young mussels. Fortunately, a smilar plant was made the same season by the fish-cultural staff at the suggestion of the director. Since the pond was larger and the total number which was recovered was greater, it will better rep- resent the results by the pond method. A number of black bass were infected with Lampsilis luleola in the fall of 1913. In the spring they were placed in one of the large earth ponds, 0.S43 acre in extent, used for propagation. The following November (1914), when the pond was drawn, some 60 mussels were picked up from the bottom. In the spring of 1915 more were recovered, making a total of 150. These were examined and measured. They had attained about the same growth as the mussels in the floating crate. The largest measured 35 milli- meters in length, the smallest 15.5." The greater length would be explainable as due to the longer growth period; having been on the fish during the winter, they would in all probability have completed their parasitic development some time before June 10, the date on which the plant in the floating crate was made. As compared with the small pond, the size doubtless contributed to the maintenance of more favorable condi- tions. We have in such a body of water conditions closely approaching the habitat of L. luleola in nature. Whether the distinctively river-growing mussels would thrive in such a pond in the absence of a current has not yet been satisfactorily determined. However, the fact that, in spite of many failures with some of these species, a few of these (represented in fig. 73) have been found in the ponds, for the most part of unin- tentional or sporadic occurrence (see Coker, Shira, Clark, and Howard, 1921, p. 165), leads one to believe that favorable results might be obtained by a proper control of conditions. GROWTH IN PENS. Recently a device was employed by Roy S. Corwin (1920) at Lake City, Minn., which gave very satisfactory results with the Lake Pepin mucket. A box 10 by 10 feet square and about 8 inches high was surmounted by chicken wire and the whole * Experiments planned to conform as closely as equipment permitted to the conditions proposed were carried through the season of 1919. Precipitation of silt occurred in large Quantity, which doubtless accounts for failure to secure a plant of river mussels. A plant of lake mussels (L. luteola) was obtained. 'Measurements of these mussels after a second summer's growth. Deer, 191.;. give for trie largest a length of 65.6 millimeters. From two of these were cut r6-line buttons 2 lines thick. (See footnote, p. 71.) 78 BULLETIN OF THE BUREAU OF FISHERIES. sunk in a protected part of Lake Pepin. In the pen thus made it was possible to retain a considerable number of fish carrying heavy infections. At the end of the season the wooden bottom was floated to the surface, and an examination revealed a total of 1 1,000 small mussels as reported. This is to date the greatest quantity production of mussels yet attained in an inclosure. This method has several obvious good features for situa- tions in which it may be employed. It approaches natural conditions more closely than the other methods described. The suitable depth for both fish and mussels is more readily obtained than in a crate, as well as more ample range in other directions. It seems doubtful if it can be used in a river where the current would remove the young mussels or the silt deposit cover them too rapidly. This season (1920) a test of the device is being made in the growing of river mussels in the Fox River, where the mucket mussel (Lampsilis ligamentind) is abundant and apparently thriving, since young mussels are readily found. The water of this stream is clear so large a part of the time that a protected location devoid of current should prove suitable. It is difficult to see how such a pen could be employed in a turbid river like the Mississippi, since at points devoid of current the precipitation of silt would bury the young mussels. The habitat of juvenile mussels in the Mississippi has been found to be a current-swept gravel bottom, always clean despite the almost constant presence of mud-laden waters. STRUCTURE AND DEVELOPMENT OF JUVENILE MUSSELS. The rearing of these mussels through the juvenile stage presented for the first time the opportunity to determine the structure at almost any age and processes of develop- ment during this period in the life history of fresh-water mussels. The investigations by Herbers (1913) and Harms (1909) have recounted in detail the development during that period for the Anodontas, Margaritanas, and Unios. In these cases, however, the juve- niles were obtained for the most part free in nature, and therefore their age could not be given with certainty. As no detailed account has been published for the develop- ment of the large and valuable group of mussels included under the Lampsilinae, the description of complete development in these would be a distinct contribution to our knowledge of mussels. However, because of other features demanding more attention at the present time, the intention of this paper is to mention only a few prominent points in the development, reserving the detailed account for another publication. Upon beginning free life the shell of the young mussel, as has been stated above, is that of the larva. When closed, therefore, no striking difference between the young mussel and the glochidium is noticeable. Like the glochidium, it is for the most part colorless and transparent. If, however, the young mussel is alive it soon extends its foot, and in its use quickly demonstrates it to be an organ well developed for the purposes of locomotion. The foot is somewhat cleft at the apex, so as to give a bilobed appearance, and is clothed with cilia, all of which are in rapid motion during extension. On smooth surfaces like glass it has the power of adhesion, a property apparently not held in the adult, at least not to the same extent. By means of this organ the young mussel is able to move about rapidly. These peculiarities in the foot of the early juve- nile are soon lost, and during the first month the foot assumes the characteristic form of this organ in the adult. CULTURE OF FRESH-WATER MUSSELS. 79 The gills are in the form of papillae, of which at this stage there are three or four on each side of the foot, the longest being anterior, since it is the oldest or first developed (see figs. 67 and 68). They are long, slender processes slightly recurved at the ends. These increase in number with age and later become united to form the continuous lamellae of the inner gill. The outer gills become visible between the first and second month or at a length of between 3 and 5 millimeters. Schierholz's (1888) determina- tion of the time as the second and third year for Anodonta and third and fourth for Unio has been shown by Herbers (19 13) to be incorrect for Anodonta, and will probably be found to be rather late for Unio. Other prominent features in the youngest juveniles are the liver and the adductor muscles. The liver, because of its dark color, becomes quite prominent before the young mussel leaves the fish. It furnishes in this manner a ready index for the degree of development when examined alive. The adductor muscles also become conspicuous, but in another way. Because of their form and an index of refraction higher than that of the surrounding tissues they appear as bright spots. The stomach and intestines seem to become functional at once, the latter at first with a few turns comes graduallv to the tortuous condition in the adult. The heart and kidney can not readily be made out in whole mounts. Herbers (19 13) by sectioning finds their development pretty well advanced in Anodonta cclensis at a length of 2.59 millimeters, corresponding to the second month in Lampsilis luteola. The mantle is a direct derivative of the same organ in the glochidium. The coming of free life marks a change in its function. Where in the glochidial and parasitic periods (in this species) no increase of shell occurs, in the juvenile stage a phenomenal growth takes place. Beginning as a delicate microscopic membrane lining the glochidial shell, it increases with the growth of the mussel until, as we have seen, it is increased in size thousands of times in a single summer and eventually produces the heavy shell, the protective armor of the grown mussel. The shell of juveniles up to the second month has two features that are characteristic of this early period. In consistency it is like horn, being transparent and less hard than later, when it becomes calcareous. The surface is uneven owing to a series of regular and relatively high undulations, knobs, etc., which are characteristic for each species (fig. 73). These are designated as "umbonal sculptures" by conchologists in describing the adult mussel, in which they are not infrequently found well preserved. A structure to which special attention is called is the byssus, an organ that is charac- teristic of the juvenile stage in certain groups of fresh-water mussels. It consists of a hyaline thread produced by the byssus gland located on the ventral and posterior median edge of the foot. The first instance of it observed in the present culture was at an age of about 38 days, when the smallest of the mussels collected had a length of 4 millimeters (other cultures 1.9 millimeters). In this same species in nature the author has seen it present at a size of 2.8 millimeters. In juveniles of Ouadrula heros, at an age of a few days, there is apparent a tough mucous-like secretion that serves to anchor the young mussel. Near the end of the growing season byssi were found on mussels of over 1 inch in length. The strength and caliber of the threads are appropriate to the size of the mussel. When the mussels were removed from the water at a temperature near that of freezing on November 20, attachment by byssi was not noted. However, 80 BULLETIN OF THE BUREAU OF FISHERIES. the circumstances of their removal from the river rather than the change in tempera- ture may have caused them to become detached. An examination in March of the following spring revealed the byssus present in most of the individuals, and it was present until June 10, after which date it could no longer be found. The disappearance at this time near the middle of the season's growth requires some explanation. It comes at the beginning of the period of most rapid growth, which is, perhaps, a decided physio- logical change, although very gradual, coming as it does after two months of spring growth. The observations to be recounted of a byssus in adult mussels would lead one to expect the persistence of the byssus under favorable conditions. On August 14, 1914, the author found an adult Plagiola donacijormis on a byssus, and later E. A. Martin showed the author a still larger individual. The byssi in these cases were strong enough to support the weight of the mussels. In this species {Plagiola donacijormis), then, we find the byssus habit not confined to the juvenile stage. The development of the reproductive glands in fresh-water mussels was clearly made out by Herbers (1913) in Anodontas and Unios. He was able to distinguish early stages of the glands in Anodontas of 5.7 millimeters length. The maturity of these organs would mark the adult stage. In collecting various species of mussels in the field one occasionally discovers remarkably small individuals breeding. As these are so uncommon they are undoubtedly examples of precocity and exceptional. The author has not found gravid individuals of Lampsilis luteola under what was apparently the third year. In the cultures here described sexual differentiation in secondary characters appeared the second summer. Modifications of the gills to form the marsupia appeared in the female, together with the corresponding fullness of the shell over that organ. The males were marked by the more pointed posterior portion of the shell. In the middle of August of the third summer the first gravid mussels were found. This, the first observed date of breeding, was 2 years, 2 months, and 24 days from the date of implantation of the glochidium. All females as far as examined were found to be gravid, which indicates that breeding is general at this age. The glochidia were mature in some individuals on August 14 and near maturity in others, which from the date of last observation would fix the time of ovulation as July. Mature glochidia from these mussels were taken and an implantation obtained on a number of fish. The first free juveniles after metamorphosis were obtained in 10 days, others remained as late as the 18th day, a rather long period of shedding. The juveniles obtained represent the second generation of mussels, but the life cycle was completed when glochidia were obtained, as that was the stage with which the experiment began. HABITS AND HABITAT OF JUVENILE MUSSELS. The juvenile or postparasitic period begins with the release of the young mussels from encystment on the host. Because of the small size of mussels at this stage infor- mation regarding their habits and environment must depend largely upon studies under conditions of control or experiment. Obviously, it is entirely impracticable to count on finding them thus early in nature. The watching of the process of separation from the host has been found practicable only by making cuttings of infected gills from living fish and by examinations under the microscope. The first sign of the change is a repeated opening and closing of the shells. This is followed by extension of the foot, CULTURE OF FRESH-WATER MUSSELS. 8 1 the movements of which become gradually more vigorous until this remarkably motile organ sweeps an arc in the plane of the valves included by the three sides of the mantle cavity (the anterior, posterior, and ventral opening of the shell). The cases observed by the author took several hours, but under the conditions of observation the difficulties are greater than when normal in the living host. There seems to be an adhesion of the shell to the host's excised tissues that is due, very likely, to coagulation. In some cases the process was so prolonged that, before escape was effected, considerable decomposition of the host tissue was apparent. The juveniles, therefore, exhibited a remarkable resistance to the products of decay toxic to most animals. The free juvenile under conditions of observation appears at times very active. In moving from place to place the foot is extended a distance fully equal to the length of the shell, becomes fast to the glass or some object, then contracts, bringing up the remainder of the animal. This is repeated again and again, thus accomplishing a kind of creeping motion which carries the small organism across the field of the microscope in a suprisingly rapid manner. The presence of cilia all in rapid motion upon the foot and edges of the mantle add to the effect of vigorous vitality. It seems probable that the young mussels do not move about much if they find a suitable bottom. Time and again the author has looked for them on trays set on the bottom of the aquarium to catch them as they fall from the hosts, but all in vain before washing off the sediment. When this accumulated sediment in which they were lying was removed, they could be seen, and after being left for a few minutes without dis- turbance the}- would extend the foot and begin the migration reactions mentioned above. Often one finds considerable debris adhering to their shells. In one species delicate hair-like processes were observed. A covering of bottom sediment doubtless serves as a shield from enemies. The mortality at this age is very high as may be seen by the number of empty shells and the scarcity of live mussels a few days after the beginning of free life. Their chief enemies, so far as noted under cultural conditions, are very small rhabdoccels, turbel- larians that are extremely abundant during the summer in the water as it comes from the reservoir. These swarm over the bottom of the aquaria, and examples may readilv be found through the transparent body walls of which may be seen the mussels they have eaten. These have been observed in both the glochidial and early juvenile stages. The species of Turbellaria as determined by Caroline Stringer were Microstomum sp., Stenostomum leucops, and 5. tenuicauda. Specimens of the Microstomum were preserved with the young mussels still inclosed in their relatively capacious intestines. Another enemy which it has been possible "to arrest with the goods still on him" is a small chaetopod, apparently Chaetogaster. Neither of these worms is more than 0.4 millimeter wide, so that after the mussel attains three weeks' growth it must be safe from their ravages. The food of the very young juveniles seems to be similar to that of the adult; i. e., at least in part, microscopic plants and animalcules taken in through the incurrent siphonal aperture. In small juveniles one can watch these as they enter. The author once observed a considerable deposit of excreta containing the skeletal remains of such forms as diatoms. This debris was lying in a heap outside beneath the excurrent siphonal opening. 82 BULLETIN OF THE BUREAU OF FISHERIES. The floating crate method furnished an unusual opportunity for the study of living juveniles after the first two weeks, but as it was the first successful trial, for fear of dis- turbing the plant, the occasions for raising of the baskets were reduced to a minimum. Such incidental notes as were taken while tending the cultures may be of interest, inasmuch as so few observations have been made upon the habits and habitat of the juvenile Naiad. In relating observations upon the habits of the culture attention is directed to the conditions prevailing during the experiment. The arrangement of the crate and baskets is described under Methods, page 64. (See also figs. 74 and 75.) The crate, being placed in the river channel, received a current of 2 to 3 miles per hour. In the individual baskets when at the surface no current could be detected. The fineness of the mesh was chiefly responsible for this. Much of the time, however, owing to a slight sinking of the crate, water to the depth of an inch flowed over the top. Thus the mussels, although probably never in a continuous current comparable to that in the river, received a constant renewal of the water supply. A gathering of the mussels at the sides of the basket was very marked. This might be construed to indicate that they found there conditions more favorable than at other points. Doubtless at the bottom of the basket the freshest supply would be at the outer edges. During a greater part of the summer flood conditions prevailed in the river, so that the content of suspended silt was very high. The checking of the current on reaching the baskets resulted in the deposit of this silt at the rate of over 1 inch per week. This is considerable when the conditions are considered. Thinking this might bury the young mussels, the silt was removed weekly by washing through the sides of the basket. Later this regimen was abandoned, being considered too violent and an unnecessary disturbance for the minute mussels. At the end of the season in November the silt in the bottom had accumulated to a depth of 3 inches. This sedimentation, however, covered a long period, most of which was not in time of high water. The first collection of Lampsiles luteola from the crate numbered 7 at an age of 15 days. Three of these were built into the mosaic tube of a caddisfly larva, and of these three, two were still alive. The larva finding a scarcity of sand grains and similar building material had evidently made use of the mussels. The predacious worms mentioned above as so abundant and destructive of mussels were not found in the crates. They are apparently a bottom species, and thus the position of the crate on the surface fore- stalls their ravages. One of the most conspicuous species associated here with the mussels was the larvae of the Ephemerid mayflies. As they are vegetarian they could be destructive of young mussels only in a competitive way, but ordinarily in crate culture they would not develop in time to be troublesome. The presence of these and like insect forms is doubtless due to the development of eggs deposited by the adult insects in the crate itself. Some other forms observed were numerous Hydra and Polyzoa, together with the free-swimming forms which make up the plankton of the main river. The byssus was first observed in mussels of 38 days. The attachment was to such objects as could be found in the mud at the bottom of the baskets, some on the filaments of Cladophora and other algae growing in the basket. One was found attached to the tarsus of a dead spider. The byssus increased in diameter and length with the growth of the mussels. When the latter were large enough to be readily seen, it was surprising Bull. U. S. B. F., 1921-22. I 1 , 71 —A floating crate containing four baskets (ci fig. k) in which were placed the fish infected with mussel glochidia. The first successful attempt to rear mussels was made in this d< Fig. 75.— One ol the propagation baskets with the bottom -till submerged and photographed from directly above. < iwing to disturbance of the water supply the young mussels, as shown by their trails, have migrated considerably. Such migrations apparently do not occur under ordinary conditions. Reduced to two ninths natural size. CULTURE OF FRESH-WATER MUSSELS. 83 to find that they, like adult mussels, were usually buried in the mud, a small portion only of the posterior end of the shell reaching the surface. In fact, the only exception observed was in a case of interference with the water supply coming to the mussels, a discussion of which will be taken up later. No evidence of migration, by tracks or other signs, was seen. It would seem, therefore, that for this species when on a mud bottom the byssus would serve chiefly as an anchor for emergencies and would not frequently be called into service (cf. Isely, 191 1). A change of position in the mud was noted when, owing to the presence of a small catfish that had escaped from another basket, the mussels burrowed deeper. At the time of removing the mussels from the river for the winter the basket con- taining the brood of Lampsilis luteola was placed in a tank and the fresh supply of water cut down; then the mussels began to migrate, as can be seen by their tracks in the photograph (fig. 75). When the water was entirely drawn off, those on the surface fell over and closed their shells. Altogether by observations for such brief periods the author did not note a varied number of locomotor reactions. The fact is, mussels when thriving and undisturbed seem to be comparatively inactive. Experimentally, doubtless, there would be a varied number of reactions depending upon the variety of stimuli applied. At present in our campaign to preserve the mussels and to increase their numbers we are particu- larly interested in the reactions manifested under natural conditions. We have some evidence of adaptations to depth of water and migration determined by river stages. There are indications also that some breeding reactions are influenced by light, others by temperature, chemical action, etc. The reactions of mussels when caught on sand bars by receding water vary with the species. The hieroglyphics of their wanderings under these conditions are sometimes very elaborate. The discovery that the parasitism of mussels is limited in some species to one or a few species of hosts suggests the possibility of specific reactions in these by means of which the infection of the host is insured. (Howard, 1914a, Conditions of Infection in Nature, p. 39). This particular phase of their habits did not come within the range of this investigation, but it is suggested that in these ecological relations of parasitism the student of animal behavior may find that the ordinarily inactive fresh-water mussel will furnish a varied and interesting subject for study. DISCUSSION AND APPLICATION OF RESULTS. In considering the results of the foregoing experiments attention is directed particu- larly to those which seem applicable immediately to the campaign for mussel conser- vation. Prof. J. L. Kellogg (19 10) points out that there can be practically no conservation without culture or cultivation. Extinction has been the unvaried fate of useful forms, plant or animal, where the natural supply has been depended upon. In the more primitive human societies all food is obtained from the public domain, but civilization, with increase of population, has survived by assigning individual property rights from the public domain, thus encouraging and making cultivation possible. To give an example in a field closely allied to that of fresh-water mussels, this principle has been strikingly illustrated in the history of the oyster and clam fisheries. Those States, as Rhode Island and Connecticut, which framed laws encouraging the culture of oysters 84 BULLETIN OK THE BUREAU OF FISHERIES. increased immensely their production, while a constant decrease was observed where the natural reefs were depended upon without sufficient encouragement to cultivation (Massachusetts Commissioners of Fish and Game, 1907). This may be an extreme view, but it has been often true. It may be said, on the other hand, that although game protection (of fish, birds, and mammals) has been so frequently only a name in America, there are cases known to all where wild species have thrived under efficient protection combined with restocking in cases of depletion. Past efforts in the conservation of mussels have been largely confined to this limited type of protection. The work has consisted in "artificial propagation" (definition of which follows) and a certain amount of protection by law instituting open and closed seasons to fishing. The closing of certain streams for a number of years, thus creating preserves, has been advocated. Further assistance to nature in recovering from the effects of depletion is suggested in a system of culture, including protection and planting like that employed in restocking with fish. The experiments here described furnished practically the first positive data contrib- uting to the development of a system for the culture of fresh-water mussels. It seems worth while to consider whether cultural methods, which the present investigations indicate to be quite feasible, might add anything to the methods now in use. In using the term culture we distinguish from propagation. ARTIFICIAL PROPAGATION. Artificial propagation as it has been applied to mussels is a method which, as indi- cated above, has been employed by the Bureau of Fisheries some eight years past. The larval mussels are brought in contact with and allowed to infect the host fish, which are then released to spread the mussels under the usual conditions prevailing in nature. In the effort to secure increased production of mussels this artificial infection has the following advantage : Whereas in nature the number of mussels which succeed in finding lodgment upon a fish is, as a rule, comparatively small, by artificially bringing parasite and host together the fish is made to carry a much greater number than would otherwise succeed in finding a host. Thus, the number of mussels reaching the juvenile stage is increased. The place of shedding of the young mussels from the fish is to a large extent doubt- less a matter of chance. As among marine clams probably those only survive which fall on, or subsequently reach, a favorable bottom. These considerations are largely responsible for the present investigations in the effort to supplement artificial propa- ga 10n' THE CULTURAL METHOD. The cultural method as suggested by the present experiments would consist in carrying protection through the second critical period in the life of the young mussel and in planting in favorable localities the mussels obtained. PROTECTION. In almost all successful attempts at rearing animals or plants protection in critical stages is the important factor. An example from fish culture is the raising of trout. In agriculture the plant or animal is placed under the best environment attainable and protected from destructive forces of all kinds at all stages until used. If finally con- CULTURE OF FRESH-WATER MUSSELS. 85 sumed for human use, provision is made to insure the perpetuation of the stock. In nature the dominant animals are the mammals which apply the principle of protection in the care of their young. Likewise among plants, those lines that have adopted this economy have attained dominance. By the "artificial propagation method" the young mussel is carried through one critical event (infection) only. Liberation from the host and the early juvenile stage are equally if not more critical. Evidence showing this has been given above, and corrobora- tive of this is the following testimony of Prof. Kellogg (1910) regarding the correspond- ing stage in the soft clam : Probably not even the swimming stage is more critical for Mya than this period of creeping which is of longer duration. It is exposed to numerous enemies and has little defense against them , for its trans- parent shell is still very thin and brittle. Lefevre and Curtis (1912, p. 192) say regarding this stage of fresh-water mussels: It is to be supposed that only a very small proportion of individuals thus liberated would succeed in reaching maturity, as they would be exposed to the same destructive agencies as are encountered under natural conditions. The results attained in the present investigation seem to indicate that a culture carried at least through the early juvenile stage and possibly to the adult stage would be economically practicable. In the floating crate method and the ponds ' we seem to have found methods of protection. The proportion of survivals (8 plus per cent) in the crates is apparently greater than from those raised in the ponds (according to the best records we have) and doubtless can be greatly improved upon. Compared with the number under analogous conditions in nature it is tremendous. For example, Prof. Mobius (1877) finds that a young oyster has 104^000 of a chance to survive and reach maturity. The same is true among practically all forms in which the young are early exposed to the vicissitudes of a free life. In the culture of sea clams the operator is dependent for planting upon such seed clams as are obtainable from a purely natural and thus somewhat uncertain supply. There is this decided advantage in operations with fresh-water mussels, that the necessary glochidia can be obtained with practical certainty as long as adult mussels last. Protection at other than the two critical periods mentioned would be included in a complete system of culture. During the parasitic period it would consist in the proper care of the host fish. It may be noted that the fish when infected demands reasonable care, as the attaching glochidia cause a certain amount of laceration of the gills which subjects the fish to possible infection from fish mold (saprolegnia) and doubtless to some exposure from bacterial invasion.8 Culture for the adult mussels consists in providing the best environment for growth as well as economical means of protection and recovery. Experience has shown that other things being equal more rapid growth and development of heavier shell occurs in flowing water than where a current is lacking. (For other factors, see "Habitat," p. 94, Coker, Shira, Clark, and Howard, 192 1.) 7 The recent results described under troughs and pens have yielded even larger percentages. 8 Bacterium columnaris Davis has caused considerable mortality in experimental work at Fairport. 86 BULLETIN OF THE BUREAU OF FISHERIES. PLANTING. The planting of mussels in nature by dropping from the host fish, although conceiv- ably controlled to a certain extent by natural factors favorable to the mussel (Howard, 1914a, p. 39), is doubtless for the most part a haphazard process. Those which fall on unfavorable bottom must perish, and there is every reason to believe that successful mussel beds are the results of a precise combination of conditions at a given place. The investigations on sea clams and oysters show that myriads of the young develop to a given stage only to die if they are not on a suitable bottom. Great accumulations of these young clams on unsuitable ground may be saved by transplanting. When so employed in cultural operations, they are designated as seed clams. In the case of fresh-water mussels an artificial planting likewise would doubtless be more economical of mussels — at least than the planting in nature by fish allowed to go at large. In restocking either privately controlled or publicly owned waters the general procedure that suggests itself is to rear the young mussels to an age of 2 or 3 months or more and then to release them on bottoms that are known to be favorable. COMMERCIAL POSSIBILITIES. If the natural supply is not maintained by the means described and the price of shells continues to advance, there will possibly come a time when the rearing of mussels by a complete system of culture will become commercially profitable for individuals and privately owned corporations, whereas now carried out only by Government agencies.9 A few experiments have been made to test the palatability of fresh-water mussels. Incomplete and inexhaustive as these tests have been they have yielded encouraging if not completely satisfactory results. Reports of edible species have been received. The use of the mussel for food in addition to the present use of the shell alone would aid greatly in making the culture of mussels commercially profitable. The successful culture of the marine mother-of-pearl shell (Margaratifera var. maxatlantica) has been described by Dr. C. H. Townsend (1916). Seflor Gaston J. Vives, on Espiritu Santo Island, in the Gulf of California, reared these shells on a scale commercially profitable. The development of an industry of this nature in the culture of fresh-water mussels might be dependent upon the acquirement of property rights on river bottoms suitable for rearing mussels. Precedents for such allotments of water-covered areas are familiar in the leases for oyster beds on our sea coasts. However, the experiments thus far carried out indicate that the culture of mussels may differ to this extent from that com- monly employed in America ln for edible oysters, in that mussels can more conveniently be grown in crates or containers of some sort rather than on open bottom. This is true at least because of the potential migratory nature of the mussel as compared with the sessile habit of the oyster; i. e., it is necessary in the former case to provide for a possible loss of a plant by migration. However, that the recovery of fresh-water mussels may be comparatively easy under some conditions is the testimony of Prof. Isely (19 14). He 9 The possible alternative is the practical extermination of the mussel through excessive fishing either for the mussel itself or its host fish, or both. There have been well-known examples in history of complete extermination of useful species. 10 In the more intensive cultivation of oysters and clams in Europe containers called ponds, which are in thenatureof sluice- ways through which flowing water is conducted, have been extensively employed. CULTURE OF FRESH-WATER MUSSELS. 87 was successful in finding a very large percentage of marked mussels months and even over a year after planting. It will be noted that his work was done in small rivers where collection was possible by wading during low water. Solution of some of the problems still unsolved can doubtless be more directly and economically reached by further investigations of the life history and habitat of mussels and especially of the early juvenile stage. A few of the recent studies along these lines not mentioned in the foregoing pages follow: Allen (1914, 192 1) has made a study of the food of lake and river dwelling species. Baker (19 16, 1918) has made extensive ecological investigations in Oneida Lake. The author has had the opportunity of making an ecological survey of a portion of the Mississippi River, where mussels are abundant. A report is in preparation putting forth the results of this study of conditions control- ling the development of mussel beds and the growth of mussels under such an environ- ment. In general, much more detailed information is required concerning the various elements in an environment favorable to the mussel (see Coker, Shira, Clark, and Howard, 192 1), including water content, as substances in solution and in suspension, both food and gases; temperature variations; depth and flow; amount of light; and kind of bottom. That the conditions favorable to juveniles sometimes differ from those for adults has been indicated. If this is the case, as it is well known to be for many animals, a more complete knowledge of the requirements of the critical postparasitic stage in mussels will certainly contribute to their culture. Perhaps the most pressing problem is the securing of a complete knowledge of their enemies and means of combating these. There are still some commercial species for which the appropriate host is yet undeter- mined, and in most cases where the host has been determined practically nothing of the manner of infection and like ecological relations is known. The solution of these problems is difficult because dependent upon the observation of phenomena occurring in a medium different from our own. In the case of river mussels this medium owing to turbidity is not readily penetrated by sight. In spite of such difficulties, however, we must agree with Lefevre and Curtis (19 12) that among invertebrate animals the Unionidae, for the variety in economic and scientific interest of the problems they present, are scarcely excelled. LITERATURE CITED. Allen, William Ray. 1914. The food and feeding habits of fresh-water mussels. Biological Bulletin, Marine Biological Laboratory, Woods Hole, Mass., Vol. XXVII, No. 3, p. 127-146, 3 pis. Lancaster. 1921. Studies of the biology of fresh- water mussels. Biological Bulletin, Marine Biological Labora- tory, Woods Hole, Mass., Vol. XL, No. 4, p. 210-241. Lancaster. Baker, Frank Collins. 1916. The relation of mollusks to fish in Oneida Lake. Technical Publication No. 4, New Vork State College of Forestry, Syracuse University, Vol. XVI, No. 21, p. 1-366, figs. 1-50. Syracuse. 1918. The productivity of invertebrate fish food on the bottom of Oneida Lake with special ref- erence to mollusks. Technical Publication No. 9, New York State College of Forestry, Syracuse University, Vol. XVIII, No. 2, p. 1-253, figs. 1-44, 2 pis. Syracuse. Barney, R. L. 1922. Further notes on the natural history and artificial propagation of the diamond-back terrapin. Bulletin, U. S. Bureau of Fisheries, Vol. XXXVIII, 1921-22, p. 91-111, figs. 76-84. Washington. 88 BULLETIN OF THE BUREAU OF FISHERIES. Churchill, E. P., Jr. 191 5. The absorption of fat by fresh-water mussels. Biological Bulletin, Marine Biological Lab- oratory, Woods Hole, Mass., Vol. XXIX, No. 1, p. 68-86, 3 pis. Lancaster. Coker, R. E.; Shira, A. F.; Clark, H. W.; and A. D. Howard. 1921. Natural history and propagation of fresh-water mussels. Bulletin, U. S. Bureau of Fisheries, Vol. XXXVII, 1919-20 (1921), p. 75-181, pis. V-XXI, 14 text figs. Washington. Corwin, R. S. 1920. Raising fresh-water mussels in enclosures. Transactions, American Fisheries Society, Vol. XLIX, No. 2, p. 81-84. Columbus, Ohio. Harms, W. 1907. Die Entwicklungsgeschichte der Najaden und ihr Parasitismus. Sitzungsberichte derGesell- schaft zur Beforderung der gesammten Naturwissenschaften in Marburg, p. 79-84, 4 fig. Marburg. 1909. Postembryonale Entwicklungsgeschichte der Unioniden. Zoologische Jahrbiicher, Abteil- ung fur Anatomie und Ontogenie, Bd. XXVIII, p. 325-386, Taf. 13-16, 9 fig. Jena. Herbers, Karl. 1913. Entwicklungsgeschichte von Anodonia cellensis Schrot. Zeitschrift fur Wissenschaftliche Zoologie, Bd. CVIII, Hft. 1, p. 1-174, 104 fig. Leipzig. Howard, A. D. 1914a. Experiments in propagation of fresh-water mussels of the Quadrula group. Appendix IV, Report, U. S. Commissioner of Fisheries, 1913, 52 p., 6 pis. Washington. 1914b. Some cases of narrowly restricted parasitism among commercial species of fresli-water mussels. Transactions, American Fisheries Society, Vol. XLIV, No. 1, p. 41-44. New York. Isely, Frederick B. 191 1. Preliminary note on the ecology of the early juvenile life of the Unionidse. Biological Bulletin, Marine Biological Laboratory, Woods Hole, Mass., Vol. XX, No. 2, p. 77-80. Lancaster. 1914. Experimental study of the growth and migration of fresh-water mussels. Appendix III, Report, U. S. Commissioner of Fisheries, 1913, 24 p., 3 pis. Washington. Kellogg, James L. 1910. Shell-fish industries, p. i-xvi, 1-361, figs. 1-67. Henry Holt & Co., New York. Kelly, H. M. 1899. A statistical study of the parasites of the Unionidse. Bulletin, Illinois State Laboratory of Natural History, Vol. V, Art. VIII, p. 390-418. Urbana. Kofoid, C. A. 1903. Plankton studies. IV: The plankton of the Illinois River, 1894-1899, with introductory notes upon the hydrography of the Illinois River and its basin. Part I: Quantitative investigations and general results. Bulletin. Illinois State Laboratory of Natural History, Vol. VI (1901-1903), Art. II, p. i-xviii, 95-629, pis. I-L, 2 text figs. Gazette Press, Cham- paign, 111. Lefevre, George, and Winterton C. Curtis. 1912. Studies on the reproduction and artificial propagation of fresh-water-mussels. Bulletin. U. S. Bureau of Fisheries, Vol. XXX, 1910, p. 105-201, pis. VI-XVII. Washington. Leydig, F. 1866. Mittheilung iiber den Parasitismus junger Unioniden an Fischen in Noll. Tubingen, In- augural-Dissertation. Frankfurt-am-Main. Linnville, H. R. and H. A. Kelly. 1906. A text-book of general zoology. 462 p. London and Boston. Massachusetts Commissioners of Fish and Game. 1907. A partial report upon the shell fisheries of Massachusetts. 68 p. 25 pis. Mobius, Karl. 1877. Die Auster und die Austernwirtschaft. 126 p. ,1 Karte, 9 fig. Hempcl und Parey, Berlin. [Translation, "The oyster and oyster culture," by H. J. Rice, occurs in Report of Commis- sioner, U. S. Commission of Fish and Fisheries, for 1880 (1883), p. 683-751. Washington.] CULTURE OF FRESH-WATER MUSSELS. 89 Reuling, F. H. 1920. Experiments in the artificial rearing of fresh-water mussels in troughs under conditions of control. Transactions, American Fisheries Society, Vol. XLIX, No. 3, June, 1920, p. 153—155. Columbus, Ohio. SCHIERHOLZ, C. 1888. Uber Entvvickelung der Unioniden. Denkschriften der Kaiserlichen Akademie der Wissen- schaften, Mathematisch-Naturwissenschaftliche Klasse, Bd. 55, p. 183-216, Taf. 1-4. Wien. Schmidt, Ferd. 1885. Beitrag zur Kentniss der postembryonalen Entwickelung der Najaden. Archiv fur Natur- geschichte, Jg. 51, p. 201-234, Taf. n-12. Berlin. Simpson, Charles Torrey. 1900. Synopsis of the Naiades, or pearly fresh-water mussels. Proceedings, U. S. National Mu- seum, Vol. XXII, No. 1205, p. 501-1004, PI. XVIII. Washington. 1914. A descriptive catalogue of the Naiades, or pearly fresh-water mussels. In 3 parts, 1540 p. Bryant Walker, Detroit, Mich. [Ann Arbor Press, Ann Arbor, Mich] Townsend, Charles Haskins. 1916. Voyage of the Albatross to the Gulf of California in 1911. Bulletin. American Museum of Natural History, Vol. XXXV. Art. XXIV, p. 399-476. New York. Wilso.v, Charles B., and H. Walton Clark. 1912. The mussel fauna of the Kankakee Basin. U. S. Bureau of Fisheries Document 758, 52 p., 2 figs., 1 pi. Washington. FURTHER NOTES ON THE NATURAL HISTORY AND ARTIFICIAL PROPAGATION OF THE DIAMOND-BACK TERRAPIN. By R. L. BARNEY, Director, U. S. Biological Station, Fairport, Iowa. CONTENTS. Page. Introduction 91 Brood stocks of the experimental farm 92 Original Carolina brood stock . 93 Second Carolina brood stock 94 Texas brood stock 95 Ratio of sexes and fertility 9° Growth 98 Attainment of salable size 100 Culling ioo Winter feeding <°i Space requirement 102 1909 brood 103 1910 brood io4 191 1 brood IOS 191 2 brood I°6 1913 brood 107 1914 brood IQ8 Mortality 108 Summary no INTRODUCTION. There appears to have been sufficient progress made in the experimental work on the artificial propagation of the diamond-back terrapin, Malaclemmys centrata, at the United States Fisheries Biological Station, Beaufort, N. C, since 19 15 to warrant the drawing up of a report covering such information on this subject as has been collected to date and has remained unpublished. In Economic Circular No. 5, revised, of the U. S. Bureau of Fisheries,1 the results of observations up to and including 19 15 are cited with methods outlined for construction of pens, selection of brood stock, care of eggs, young, and adults, and some notes on the growth of the terrapins. Much information has been collected since that time by continuing observations on many of the same terrapins considered in the 191 7 report and also through further studies with different purposes begun in more recent years. 1 Hay. W. P.: Artificial Propagation ol the Diamond-Back Terrapiu. Economic Circular No. 5. revised. U. S. Bureau of Fisheries, Washington, iyi;. 91 92 BULLETIN OF THE BUREAU OF FISHERIES. The terrapin propagation study has been directed since its beginning by several investigators. Originally Dr. R. E. Coker gave his attention to its possibilities and prepared a report of his results for the North Carolina Geological and Economic Survey.2 At the same time Prof. W. P. Hay began similar investigations in Chesapeake Bay. In 1909 these were transferred to Beaufort, where Dr. Hay took charge of the experi- mental work and continued giving it his direction from 1909 to 1915. During this time H. D. Aller planned and carried out the feeding of yearling terrapins during the winter in a warmed nursery house. Lewis Radcliffe relieved Mr. Aller in 1912 and was later followed by S. F. Hildebrand. Some of the material herewith discussed is from experiments begun by the two last-named investigators, but left unfinished because of their removals from Beaufort. The present paper is based on the unorganized notes of each of the above-mentioned investigators and also on the systematic observations carried on under their supervision by Charles Hatsel, the terrapin culturist stationed at Beaufort, N. C, since the experimental work was begun. The large share of credit for the continuity and the accuracy of the observations of the entire experimental terrapin propagation project is due Mr. Hatsel for his exceptionally careful, energetic, and faithful work. The writer has had the direction of the experiments since the fall of 1919. B. J. Anson has assisted in organizing and tabulating the data discussed in this paper and J. B. Southall has prepared the graphs. BROOD STOCKS OF THE EXPERIMENTAL FARM. The terrapins of the original brood stock, which are either the parents or grand- parents of all the Carolina terrapins that are held in captivity and under observation at the Beaufort station, were purchased in two lots, the so-called " original lot of North Carolina breeders ' ' and the ' ' second lot of North Carolina breeders. ' ' To these was added later, but kept separate, a number of adult Texas terrapins as brood stock. The pro- duction in eggs and young throughout the years of captivity of these terrapins is herewith tabulated (Table 1) and shown in graphic form (fig. 76). Table i.— Records of Breeding Stocks of Terrapins in Captivity at Beaufort, N. C. Stock and year. Original stock: 1909 1910 191 1.. ..... . 1912 1913 1914 1915 1916 1917 1918 >9"9 Males. Females. Eggs. Number. 288+ 598+ 688 732 736 923 921 722 >757 '834 Rate per female. Young. Number. (?) 6.5+ ■3- 9+ 16. o 17.0 18.8 23. 6 23. 6 ■ 8. s 1 194 ' 21-3 288 598 538 610 594 836 813 639 "675 b 757 Rate per female. (?) 6.5 13- 9 12.5 14. 1 15.2 21.4 20. 8 16.3 6 17-3 l> 194 Per cent eggs hatched. " Five males taken from this lot for experimental purposes in 1911 were returned to it in 1919. 1 Estimated. (?) (?) (?) 80. 90. 88. 88. i>89 'Coker, R. E. : The Cultivation of the Diamond Back Terrapin. Raleigh, 1906. Bulletin No. 14, the North Carolina Geological Survey. THE DIAMOND-BACK TERRAPIN. 93 Table i.— Records of Breeding Stocks of Terrapin in Captivity at Beaufort. N. C— Con. Stock and year. Males. Females. 45 7° 45 7° 4S 70 45 70 45 29 29 28 70 64 63 63 27 b jo 50 89 d 33 34 d 29 34 d 29 34 d 29 34 d 12 25 d 12 25 d 12 25 5 II 5 11 Eggs. Number. Rate per female. Young. Number. Rate per female. Per cent eggs hatched. Second stock 19" 1912 1913 1914 i9'5 1916 1917 1918 1919 1920 Texas stock: 1912 1913 1914 1915 1916 1917 1918 1919 1920 649 673 74S 958 871 " 973 "731 ° 773 1, 172 I-'7 292 412 399 421 497 (') 270 138 9.0 9 4 10. 4 13- 4 136 2 154 1 11. 5 'I5-4 c 13- I 3-7 8.5 12. I II 7 16.8 19.8 M 24.5 <■■-' 5 583 606 724 876 783 805 a 670 a yo2 1. 133 IOI 281 376 366 383 439 (') 8.3 8.6 10. 2 12. s 12. 2 12. 7 d 10. 6 014.0 c "2.7 2.6 8.2 11. o 10. - ■5- 3 ■ 7S (•) 22. 4 89.8 90. o 97.1 91,3 89.8 '82. 7 693.0 &90.8 96.6 79.5 96- 2 91.2 91.7 9O.9 88.3 91.4 97- o « Estimated. t* The 1920 record represents the combined production of the original and the second brood stocks. The penning together of individuals of both stocks made it impossible to ascertain the production of either slock. c Decrease in production probably due to destruction of eggs by rats. d This record for males includes five Carolina original stock males used in hybridization studies e Records for production not obtained on account of storm which destroyed egg beds and washed many small and adult terrapins from their inclosures. ORIGINAL CAROLINA BROOD STOCK. Considering the original or first lot of breeders it will be noted that highest egg production occurred in 1915 and that since then the egg rate per female, with exactly the same number of females laying, has diminished by from three to five eggs. The percentage hatched has varied but slightly. This, however, would be reasonably expected with the same number of males on hand and the number of eggs to be fertilized somewhat less. In view of the fact that egg production has fallen off since the 1916 production and remained under that high mark now for four years through 1920, it seems probable that the period of maximum egg production in this brood has passed. However, the slight increase during 1918 and 1919 perhaps means that the brood may still reach greater egg production than its maximum egg record of 1915. Still beneath the maximum mark, it may also indicate that certain females are about to reach maximum production while others have passed this point. The heavy falling off in recorded egg production in 1920 (see figures for 1920 under second stock in Table 1) was due to the depredations of rats which dug up many nests and destroyed hundreds of eggs before control methods were effective. The 1920 figures, therefore, do not represent the possible egg production and hatch, for the record of eggs laid would doubtless have been very much higher had it not been for the destruction caused by the rats. The average size of the females of this lot in 191 1 was 154 mm.3 Considering what is known of the history of some of the individuals of this lot since 1902, their size then and their growth since, it appears probable that at the time of their measurement in 191 1 they averaged close to 20 years of age. The estimated age of these terrapins is 3 Approximated from measurements recorded in inches. 94 BULLETIN OF THE BUREAU OF FISHERIES. arrived at from a knowledge of their size at the time of their purchase and of the num- ber of years during which they have been captive in the experimental pens. They were all adults at the time of their purchase, and the approximation of their age must be quite accurate. From the knowledge that it takes at least much longer than nine years for hibernating terrapins to reach an average length of 142 mm. (see Table 3) and that after the length 142 mm. is reached the average annual growth increment, is not more than 1.5 mm., it appears reasonable that the age of the terrapins of this brood was at least 18 years in 191 1 (the year of measurement). Years. To reach average length of 142 mm 9 To grow from 142 to 1 54 mm. (the 191 1 average length) 9 From 191 1 to 1921 IO Total average age in 192 1 2& 2A- 20 16 12 8 -^, — / / / / \A w / \ \ / / / __/ V / / / / 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 Fig. 76. — Egg production per female in original and second Carolina and Texas brood stocks. , Original Carolina brood stock: , second Carolina brood stock; , Texas brood stock; * combined egg production per Icmale o( original and second Carolina brood stocks. Marked decrease in egg production probably due to destruction of eggs by rats. Inasmuch as maximum recorded egg production for this lot occurred in 1915, we may presume, then, that maximum egg production occurs about the twenty-fifth year of a terrapin's life. The actual maximum production of young, however, would, of course, depend on the presence of sufficient males among the breeding stock. SECOND CAROLINA BROOD STOCK. The second lot of North Carolina breeders, with average measurement of 141 mm.4 in 191 1 and probable average age of 9 years, has shown, with the exception of the years 1918 and 1920, a general increase in egg production since the beginning of its laying. The 1918 record showed a dropping off in egg production of 2.1 eggs per female. The reason for this decrease is problematical, but the exceptionally severe winter of 191 7-18 and the following late spring, with its resulting longer hibernation period and its retard- * Approximated from measurements recorded in inches. THE DIAMOND-BACK TERRAPIN. 95 ing influence on normal spring feeding and growth in the terrapins, may possibly be the causes of the decreased productiveness, though a similar decrease is not found in the egg production of the original brood stock. The 1920 decrease shown in Table 1 is due in part to the destructiveness of the rats above referred to and probably also to a very late spring, the first eggs being laid on May 17, a rather late date. Many of the brood terrapins, however, did not lay their first eggs of the season until the mid- dle of June, more than a month after the usual first egg-laying date. In 1917 certain of the females of the second lot of breeders were set aside for experimental purposes, and this had a tendency toward diminishing the actual number of young produced, while the experiments yielded information which is of value from other aspects and will be discussed later. Because of this experimental work the figures for "young" in the table are estimates and are included only tentatively in the 1918 and 1919 records. The mixing of part of the first lot with the entire second lot of breeders in 191 8 and 19 19 made it necessary also to estimate the number of eggs laid by both these lots, and these records have been so noted. The estimation was possible in view of the fact that a considerable proportion of the original brood stock was not mixed with the second brood stock. The method followed to obtain the estimated records was this. One lot of the original brood stock (lot A) was held in a separate pen during the year, and from the egg and young produc- tion of this lot was computed an average egg and young production for the entire original brood stock. From this computation it was possible, then, to figure the production of those terrapins (lot B) of the original brood stock which had been penned with the second brood stock. The egg and young production of the entire second brood stock (lot C) was ascertained by subtracting the egg and young production of lot B from the total egg and young production of lots B and C combined, lot B plus lot C representing the mixed lots of brood stock. In 1920 the terrapins of the first and second lots of breeders were so mixed that it was impossible to estimate at all accurately the egg production of either lot, and for this reason the egg production and hatch of these broods for 1920 are combined. The combined egg and young production record, though lower than that of either lot in the preceding year, is not significant, however, because of the heavy egg destruction caused by rats. TEXAS BROOD STOCK. The Texas brood stock, with average length in the spring of 1920 of 177 mm.'1 and probable age of 30 years, has shown an increasing productiveness annually since its first laying in confinement until 1920, when doubtless, as above, the decrease in the egg and voung production record was due to destruction of eggs by rats. In 1919 from 1 1 adult females of this stock which had been used in hybrid studies with Carolina males there were obtained 270 eggs, or 24.5 eggs per female. In 1916 all but 12 of the Texas terra- pins were returned to Texas. The 1 2 that remained were the finest and largest females of the original Texas lot, and the exceptional egg production of the 1 1 mentioned above mav be due to this fact. It is understood that when the entire Texas brood stock was at Beaufort the average number of eggs produced per female was lower than for either of the Carolina lots'. B Approximated Irom measurements recorded in inches. 96 BULLETIN OF THE BUREAU OF FISHERIES. It is noteworthy in this connection that this maximum production for the Texas stock is greater than that of the Carolina breeders. The maximum records of the two stocks are as follows: Carolina, egg rate, 23.6; young rate, 21.4. Texas, egg rate, 24.5; young rate, 22.4. The excessively cold winter of 191 7-18, which apparently slowed down output among the feecond brood stock of Carolina terrapins, did not effect any retardation in the productiveness of the Texas stock. The heavy decrease in recorded productiveness of this brood stock in 1920 is due to destruction of eggs by rats, as in the case of the first and second lots of Carolina breeders. RATIO OF SEXES AND FERTILITY. The number of males in a given stock of brood terrapins in each of these experimental lots has been about one-third to one-half the number of females present. There appears to be a negligible difference in the rate of young hatched per female in the different broods with differing percentages of males present. A normal hatch appears to be about 90 per cent of the eggs laid, no matter how great a number of males may have been pres- ent. It is needless to say that scarcity of males would, of course, increase the number of infertile eggs laid. This percentage of infertile eggs is much larger at the beginning of the laying period (fig. 79) of the terrapins and at the beginning of captivity (Table 1). The cause of the high infertile egg rate among terrapins which have laid for the first time under our observation and have been penned with male terrapins of exactly their age throughout their lives may be found in the fact that possibly the males do not reach sexual maturity as early as the females. This is indicated in a study of certain lots of terrapins experimented with during 1919. Seventy-eight female terrapins of the 1914 brood which had never laid fertile eggs, due to the fact that males had never been penned with them, were separated into two equal lots which were kept in separate pens. With one lot of 39 females were placed 3 males of the original brood stock at least 25 years old; with the other lot were placed three 5-year-old males of the 1914 brood. This division of the lot and introduction of males occurred on the same dav in earlv spring, so that there might be plenty of opportunity for fertilization to occur before the egg-laying period arrived. The production was as follows: Pen 9 . . Pen 19. Females. Males. Age of of males (years). Eggs laid. 245 187 Eggs hatched. Per cent fertile eggs. 54 152 32. O 81.2 Per cent infertile eggs. 7S.0 18.8 It is suggested from these data that males of 5 years are less potent than much older ones and that maximum fertility may not be expected where young males, just reaching sexual maturity, are used. In reviewing the entire matter of the most desirable numerical relation of males to females in this species, it should be pointed out that mating among terrapins is pro- miscuous. Copulation in one year may mean the production of fertile eggs for more than that year alone. To cite a case under our observation, in 1914, 10 females of the second lot of Carolina breeders which had been producing young were set aside in a separate pen without males- With no further association with males, these terrapins THE DIAMOND-BACK TERRAPIN. 97 laid fertile eggs each subsequent year until and including 1918. The record production and hatch of these 10 terrapins through 19 18 is as follows: of egg Eggs. Young. Percent infertile eggs. Eggs. Young. Per cent infertile eggs. 129 116 128 102 0. 7 12. 0 130 108 39 4 1916 96. 2 In the spring of 1919, 5 males were introduced into this lot, and the fall produc- tion was 137 young from 146 eggs. From this experiment it appears that female ter- rapins may retain live spermatozoa in a health)' condition after a single copulation as long as four years, and under such conditions some eggs laid even in the fourth year may be fertile. It is apparent also that fertilization may occur immediately after copulation. In further consideration of the proper ratio of sexes for maximum fertility we have the records of several domestic broods (Table 2) which, it happens, have contained fixed ratios of males per 100 females throughout their existence. This set of observa- tions includes lots in which the males number 5, 9, 12, 24, 32, and 50 per 100 females, and the records give some suggestion of what may possibly be the most desirable ratio of males to females to produce maximum fertility. Table 2. — .Sex Ratio and Fertility of the Diamond-Rack Terrapin in Captivity Egg-laying year. 1909 hibernated brood First Second Third Fourth Filth Sixth 1910 hibernated brood First Second Third Fourth Fifth i9ro selected brood: First Second Third Fourth Filth Sixth Males. Fe- males. Males per 100 fe- males. zoo 200 50 .10 Eggs i id 96 11R 9* 119 138 140 38 2tto I f>40 42" »J 4.X ."9 678 4": Per cent fertil- ity. Egg-laying year. 77.8 73-4 95- 7 94- 7 89.2 97 4 »;. 4 91- -' 75- o 61. 2 81. s 74.0 91. 2 9v 5 91. 5 95- 7 1910 winter -fed brood First Second Third Fourth Fif.h Sixth 191 1 winter led brood First Second Third Fourth I-ath Sixth 1912 winter-fed brood First Second Males Males. Fe- per 100 F.ccs males. fe- laid. males. 10 «40 8 12 10 107 9 348 10 107 9 57* 10 104 9 606 9 103 8 9." 9 >°3 8 298 10 • 12 8 IO 81 7 IO 12 470 10 81 12 68a 10 12 083 10 82 12 621 18 43 24 II! 18 43 24 188 Icrtil- >'y. 83.3 78.6 90.5 75- 4 45-9 100. o 0.0 94 1 93 S 88.8 91.4 8s. ■ In viewing the records of these broods it is necessary to bear in mind that maximum fertility does not occur in the first year of laying in any brood unless there happens to be laid only a very few eggs which may have, by chance, become fertilized. Accepting 90 per cent as normal fertility, it will be noticed that this per cent of fertility was reached in the 19 10 hibernating brood in the third year of laying, and that this brood contained only 5 males per 100 females. However, in the following year, when there were about 200 more eggs laid, the per cent of fertility dropped to 75. The fifth year, though showing a decrease in fecundity in the terrapins, shows another considerable lowering of percentage fertility. A similar drop in the per cent of fertility is also found in the fifth and sixth years of laying of the 19 10 winter-fed brood after it had reached at least 90 per cent fertility in previous years. It was accompanied in the fifth year by an 7i;374° — 22 2 93 662 Mm. 18.0 38.0 Mm. 199 33-8 Mm. 40. 0 51.0 No. 34 -J28 Mm. 34- J 43.7 The maximum length of any terrapin under our observation kept through one winter and until the following May in the nursery house is 81 mm.* The greatest length of a winter-fed terrapin at approximately 2 years of age (measured in September) is 104 mm.,0 or slightly more than 4 inches. The offspring of domestic stock appear to do better in captivity than those of "wild" stock (fig. 78). From measurements • These measurements are taken on the lower shell, following the commercial method of measuring a terrapin: 81 mm. Oj'i inches ; 104 mm. O 4'i inches. 102 BULLETIN OF THE BUREAU OF FISHERIES. taken in the autumns of 1919 and 1920 the following information on this point was brought to light : Treatment. October, 1919. Average Offspring of— Number. Age in years. Average length in milli- meters. Maxi- mum length in milli- meters. length in milli- meters, October, 1920. 95 99 3 3 65. 3 74-4 87 I 66. S ..do MM. 90 80 70 60 50 4-0 30 / — ■ — — - — / ' / / / / / V 1916 1917 1918 1919 1920 Fig. 78. — Growth of 1916 offspring of domestic and wild parentage. Each lot was fed one winter and selected for large size in the fall of 1917. — — — , Offspring of 1909 domestic stock; , offspring of original Carolina wild brood stock. SPACE REQUIREMENT. The 191 7 brood had two lots of ioo each chosen from it in 1919. One of these lots, with average measurement of 52.5 mm., was placed in a large so-called fish pool which measured 38 by 24 feet. The other lot with average measurement of 60.1 mm. was placed in a small pen with dimensions of 29 by 12 feet. Observations on the growth of the two lots have been as follows : Square feet per terrapin. Number of terra- pins. Average length in millimeters. Increased length in milli- meters, May to October, 1919. Average length in milli- meters, Oct. 5, 1920. Increased length in milli- meters, October, 1919, to October, 1920. Treatment. May, 1919. Oct. 23. 1919. 36 9-3 IOO IOO 60. 1 52- 5 66.6 60. 9 6.S 8.4 70.3 73. 2 THE DIAMOND-BACK TERRAPIN. I03 A conclusive statement of space requirement or of the value of extensive running ground is not warranted from this single set of observations. It is of significance, how- ever, that the greater growth has occurred in two successive years in the larger pool. This may have been due, nevertheless, as much to the fact that the terrapins were smaller in this pen than in the " close-confinement " pen and may have just reached or were in a stage of rapid growth, whereas the other group may have passed that same period. That plenty of space has a tendency to increase fecundity in the terrapin is suggested by the 1909 brood. These terrapins, held in a pen 32 by 5 feet for six years, have had a very large average yearly egg production per female. This pen provides each terrapin with approximately 26.6 square feet of ground. The exceptionally high laying record of these terrapins may be due, in part, to the large space and uncrowded condition of their pen. Their size, of course, is large, but abundant space may be a contributing factor in causing increased productiveness. 1909 BROOD. This brood, the first terrapins hatched in captivity in the Beaufort pens considered in this paper, consisted of 1 2 individuals — 8 males and 4 females. Several of the males MM. ISO 140 130 120 110 100 90 80 70 60 50 40 30 20 10 IN. b 4i «J 1.1 s 2+S 1(1 248 ■IU 12! JU is a "i. S *♦■ r y ' - 4 V - 3 2 *i - - - 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 FlG. 79- — Growth of 1909 brood of original Carolina brood stock. Ecs production and batch per female per year expressed, respectively, by figures on the curve This brood has always had but four females. Estimated. of this lot have been used in other experiments, and since [915 there have been only 2 males with the 4 females under observation. They have been kept during this time in a small pen which, however, is large enough to support many more terrapins than these 6. This 1909 lot has hibernated each winter since its birth and shows what is probably a normal growth and development for terrapins held in captivity. The first eggs from the females of this brood were laid in 1915 (fig. 79) when the terrapins were 6 years old. The egg rate per female in that season was 24. Yearly since then there has been an increase, until in 1919 the egg rate reached 32.2 per female. This 1919 egg production was accompanied bv a hatch of 30 young per female and represented at that time the best record observed at Beaufort for average egg production and hatch. In 1920, how- 104 BULLETIN OK THE BUREAU OF FISHERIES. ever, the brood surpassed its best record again by laying 140 eggs, of which 125 hatched giving an average egg record of 35 per female and an average hatch of 31.2 young per female. There were 16 nests found in 1919 and 19 in 1920, indicating quite conclusively that all females of the 1909 brood laid at least four times and that three of them laid five times in 1 920. Growth after the seventh year is small, but it is attended by increasing fecundity. The cause of the high percentage of infertile eggs in the first year of laying of this brood may be due to the fact that the males of the brood were not mature. However, the following year the larger part of the eggs by far was fertile. 1910 BROOD. Two lots of the 19 10 brood were set apart in the fall of 1910, one fed the first winter, the other allowed to hibernate. A comparison of the average growth of these two lots indicates that the first lot, winter-fed, by average measurement, arrived at the 5-inch MM. 150 140 130 120 110 100 90 80 70 60 50 40 30 20 10 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 19^0 Fig. 80. — Growth of 1910 brood of original Carolina brood stock. Egg production and hatch per female per year expressed, respectively, by figures on the curves. ■ — , Fed two winters and selected for large size in the spring of 1914; , fed one winter; . hibernated. mark and to egg-laying one year before those that hibernated, and also that the egg-laving began at a higher rate and continued higher than that of the hibernated group (fig. 80). In early spring (1915) 8 males and 25 females were selected for size from the fed lot and kept in a separate inclosure. Their egg production in the second year of their laying was 16.7 per female, a much greater productiveness than was made either by the ordi- nary lot of fed terrapins from which the selected individuals were chosen or by the lot of hibernating terrapins three years later in 19 19. These facts indicate that for the commercial terrapin culturist it would pay more in a series of years to hold the largest and fastest growing stock as breeders rather than to pick here and there for his commercial sales. He could hold such brood stock over a long period of years and feel certain that maximum production would not come before at least 15 or 20 years. He would know, also, that the egg production of his breeders was as large as could be obtained. Thus, each year a certain number of the best grown females of 3 vears or older could be chosen to be held as the established brood stock. IN. e 1 + .3 16.7 ■11.8 a* 38.6 21.0 183 27.1 214 18.6 17.8 5.7- 4.4 9.3- 6.7 2.8- 1.3 5 HI -7 7 5.4- 4,0. ^ JT — - — — - — / '' /.I £' ~A .4 2.9 ■XA 5.0 ■3.9 7.1 ■S.2 4.7- 23. '* ' ? ' - 3 /■ — "/ 2. 1 / / r - y '' - - — THE DIAMOND-BACK TERRAPIN. I05 The balance could then be used as salable stock whenever their size was great enough to make them marketable. This would tend, then, to the selection by the terrapin farmer of his best producers and fastest growers and in the course of years lead to a race of quick -growing, large-framed, and highly productive terrapins. 1911 BROOD. The 191 1 brood has consisted of two lots of terrapins, one fed two winters and hibernating thereafter, the other hibernating each winter. The average growths of these two groups differ about 1 2 mm. at any season of any year. The tendency, however, as age increases, is toward a diminishing of this difference in the average growths. The evidence brought to light in the 1910 brood that winter feeding tends toward earlier productiveness is borne out in this brood also. In the fed lot the first egg laying occurred .09 ■0 .08 0 5.2 -5.0 8.3 -7.0 IIS- ae 8.2 7.3 2 0* 2 o« 5.5 ■21 • 4 / / / f - ]/ • > ■ s~ - - MM. ISO 140 130 120 110 100 90 80 70 60 50 40 30 20 10 1911 1912 1913 1914 1915 1916 1917 1318 1919 1920 FlC. 81.— Growth of 1911 brood of original Carolina brood stock. Ecs production and hatch per female per year expressed, respectively, by figures on the curves. , Fed two winters; .hibernated; .estimated; * no males in this lot previous to this year; *» fertility modified by experimentation. in the fourth year, probably by only one female. Substantial output occurred in the sixth year with an egg rate of 5.2 per female. The first production of the hibernating lot occurred in the seventh year, but this was negligible — 0.2 egg per female. The second egg laying in the hibernating stock was likewise small, the rate per female being 2 eggs. This only further points out the desirability of winter feeding. It indicates, also, when the results are compared with the 1910 brood lot which was fed only one winter, the futility and extra cost of feeding terrapins more than one winter. It appears that the 1 910 winter-fed brood has shown that egg production and growth from one year's winter feeding is much more desirable than the same from two years' winter feeding when selection is not made of the brood stock. It may be added in this general connection that winter feeding does not tend toward the development of weaker adults nor necessarily to animals more susceptible to disease. It is true that young terrapins in the nursery house are subject to disease, and there is occasionally considerable mortality from this cause. It apparently kills many of those that would probably die from inherent weakness at best. There are io6 BULLETIN OF THE BUREAU OK FISHERIES. many terrapins which suffer an attack of the disease in question and again recover their well-being. The subject of mortality among the young terrapins is discussed on page 108. It is of interest to note in further discussion of feeding two winters and its lack of advantage to the culturist the fact that, from average measurement, the hibernated terrapins reach the 5-inch mark approximately one year and one-half after those fed two winters. Referring again to the 1910 brood, the lot fed one winter reached the 5-inch mark two years before the 1910 hibernating group. It may then, perhaps, be that two winters' feeding may slow down growth rather than hurry development. It will be noted, too, that egg production in the 1910 brood lot fed one winter is negligibly different as regards the year of substantial egg production from that of the 191 1 lot fed two winters 1912 BROOD. The 1912 brood was fed the first winter and allowed to hibernate each winter there- after. In the spring of 1914 there was made a selection of 100 each of the smallest and IN. 6 5 4 3 2 1 .5 -.0 1.9- -.0 1.7- 4.4 ■1.0 -3.2. ^*^-' *- .»- 2.3 1.6 1.4 2.3 1.9 - * / / "" / / / / - - MM. 150 140 130 120 110 100 90 80 70 60 50 40 30 20 10 1912 1913 1914 1915 1916 1917 1918 1919 1920 Fig. 8j. — Growth of 1912 broods of original Carolina and Texas brood stocks. E^g produc- tion and hatch per female per year expressed, respectively, by figures on the curves. , Offspring of Texas brood stock, fed one winter; , offspring of Caro- lina brood stock, selected from winter-fed lot for large size October 1, r9ij; , offspring of Carolina brood stock, selected from winter-fed lot as runts October 1, rc>i.?. largest grown terrapins of the stock. These two lots were kept separately, and their growth and egg production to 1920 have been observed. The lot selected for large size after the first two years did not exhibit as unusual growth as it had in the first winter, while the "runts" after 1915 showed relatively much faster growth. Their average length in 191 7 was about 10 mm. greater than the lot which had been chosen originally for its early rapid growth. Both lots produced eggs in the same year (19 19) when they were 7 years old. It is of interest to note in this connection that the "runt" group averaged 2.3 eggs per female, while the "selects" averaged 1.6 eggs per female. It is suggested, then, from this brood stock that selection with a view toward earh' attainment of salable size or early and increased egg production took place at too THE DIAMOND-BACK TERRAPIN. 107 young a stage in the development of the terrapins to be of any advantage to the terrapin eulturist. A comparison of this 19 12 study in the effect of selection with that of the "selects" of the 1910 brood emphasizes this point. Selection for size occurred in the 1910 brood when the terrapins were 4 years old. The egg production from these "selects" was especially large. In the 1912 lot the egg production is not above normal for either group of terrapins fed one winter and remaining unselected. Selection as early as the second year is premature,, since the terrapin at that age has not reached one-half its adult size, and there may be many influences after the second year to retard growth in what then appears as an exceptionally healthy and rapidly growing terrapin. The Texas brood of 1912, numbering 24 in 1916, 1917, and 1918, and 14 from 1919 to 1921, has shown greater average growth and produced eggs in 1918, a year earlier than the Carolina terrapins of the same age. The possible earlier arrival of the offspring of Texas stock at sexual maturity may be hereditary in character. All the antecedents of this Texas stock were from the marshes of Texas, where the longer growing and laying season with the very limited hibernation period would normally tend toward the occurrence of an earlier maturing animal than would be found in nature in North Carolina. This early maturing characteristic may have become inherent in the Texas stock. 1913 BROOD. The 1913 brood was originally divided into two lots — one hibernated while the other was fed. Of the fed lot the largest 100 terrapins were selected for further MM. 150 140 130 120 110 100 90 80 70 60 50 40 30 20 10 1913 1914 1915 1916 1917 1918 1919 1920 Fig. ??. — Growth ol 1913 broods of original Carol ma and Texas brood stocks. Keg production and hatch per female per year expressed, respectively, by figures oil the curve. — , Offspring of Carolina brood stock, selected for large size in the spring of 1914; .offspring of Texas brood stock, fed one winter; , offspring of Texas brood stock, hibernated study. These selected individuals at 6 years of age did not average 4K inches, though, of course, there were some over 5 inches in length in the lot. Neither had they laid any eggs. There was, however, in the seventh year a small produc- tion of eggs and young in this stock. Their slowness of growth can not be well IN. S 4 - 4 sa- - — -- -i: a .i 3 2 7> t r -' 1 / • - ioS BULLETIN OF THE BUREAU OF FISHERIES. accounted for. The severe winter of 19 17-18 may have retarded their develop- ment somewhat, but the retardation in the average growth curve occurs before that winter. It seems only further evidence that selectio'n of the large terrapins at the beginning of the second year does not necessarily mean that those terrapins will be the largest or best producers in the fifth or sixth year; that, as has been well shown in the 1912 "selects" and "runts," selection of brood stock as early as the second year is premature. The Texas brood of 191 3 does not show normal growth. Its average growth has, however, been better than that of the Carolina brood which received the same treatment. No eggs have been laid by the Texas stock. It may be that the retarded average growth curve of the Texas stock is due to the fact that a number of the larger terrapins of this lot were lost in a heavy storm in the summer of 1918. This would also explain the late egg laying of this group. 1914 BROOD. This lot of Carolina terrapins selected in the spring of 19 15 from thos.e terrapins that had been fed during the previous winter has not shown more than ordinary growth and did not produce eggs during 1919, the fifth year of its life. In 1920, the sixth year, IN. 5 ■*•"* ** r*** ■ ^ ^ _ / //" — ,/ '/ - / ' / t, - — M M. 130 120 110 100 90 80 70 GO 50 4-0 30 20 10 1914 191.5 1916 1917 1918 1919 1920 Fig. 84. — Growth of 1914 broods of original Carolina and Texas brood stocks. Egg production and hatch per female per year expressed, respectively, by figures on the curve. , Offspring of Carolina brood stock, selected for large size in the spring of 1915; — 1 off- spring of Texas brood stock, selected for large size in the spring of 1915. however, there was a small production of eggs. Early selection and the brood's later retardation in average growth, which has been evidenced in lots previously discussed, is further borne out here. The Texas brood of 1914 has been rather insignificant, in- asmuch as it consisted of only five terrapins in 1919, the others having been shipped to Texas. These five averaged somewhat larger than the Carolinas of the same stock, but had not produced eggs up to 1920. To make room for other experimental lots they have been shipped to Texas to be liberated. MORTALITY. As has been pointed out in the report of 191 7 mortality among adult terrapins is very small. It does not amount to more than one-half of 1 per cent. In the young the mortality runs higher, especially if the terrapins are winter-fed. The death rate THE DIAMOND-BACK TERRAPIN. 109 among young terrapins hibernating the first year after birth is, however, very small. There may be an increase during the year following, but it is negligible. The death rate in the nursery house, however, is a matter of importance, though in certain years it is small. The cause of heightened mortality in the winter-fed individuals is a disease, rather cancerous in nature, which attacks the head, flippers, and especially the tail of the terrapin. The disease rots off the tail and spreads to the body proper, probably affecting the spinal cord and causing paralysis and death. Often terrapins become blind or unable to eat because of the disease. Its cause is unknown, though it probably is due to a microorganism. The disease seems most virulent to those nursery boxes which are best heated or in the sun, while those which remain throughout the cold season in the shade or at a considerably lower temperature than the others do not suffer so high a death rate. The disease does not attack the weak and small terrapins alone, but often kills some of the best grown. The kind of food used seems to have no connection with the prevalence or the virulence of the disease. Antiseptic washes and thorough cleanliness in the house and boxes apparently have some advantages, but the disease will prevail even when the most scrupulous care is taken. Possible value of antiseptic treatments seems to be borne out to some extent in the table on page 1 10. Treatments have been with potassium permanganate solution applied to the nursery boxes at intervals of a day or every few days. The solution is poured into the boxes and thrown upon the sides, allowed to remain some time, and then drawn off. Early in the fall before young terrapins are housed a thorough treatment with formalin or copper sulphate solution is given the boxes and walls of the house. The floor, being sand, is treated with lime. The first year that any antiseptic was used in the nursery house was 19 15. In 1914 the percentage of mortality among terrapins housed for the winter was 29. In 1915, with the disinfecting treatments being used, the mortality dropped to 5 per cent. However, this dropin mortality rate may have been due to causes other than the use of the disinfectants, for since 1915 there has been a continued high mortality rate even though disinfectants have been used regularly. Apparently the infecting organism can live in either salt, brackish, or fresh water, as the disease, once started, spreads when the water in the nursery boxes is changed from fresh to salt, or vice versa. Certain terrapins that are attacked recover after several weeks, but the majority die. What possibly is another symptom of the same disease, or perhaps another disease altogether, is the so-called "softening of the shell" among the young terrapins. It is found among the terrapins that make no growth and is due, perhaps, to faulty absorption of the yolk sac or to possible infection of the body wall during this process. Such infection could be readily conveyed to the internal organs and destroy the health of the terrapins so affected. The "softening of the shell" symptom is always accompanied by loss of the power of growth and sometimes by the loss of pigment in the carapace or by a deadened appearance of the entire shell. Terrapins that develop this latter symptom rarely live. The offspring of the Texas stock are as susceptible to the disease as those of the Carolina. Hybrids (Carolina- Texas cross) are none the less susceptible to the disease. From a study of the winter-fed terrapins during the winter of 1919-20, it is sug- gested that high temperature and direct sunlight on the nursery boxes in which are held the several lots of yearlings are correlated closely with the prevalence of the disease. It appears that the greatest mortality occurred in those boxes closest to the stove and no BULLETIN OF THE BUREAU OF FISHERIES. in such locations in the nursery house that direct sunlight fell upon them. Greatest growth also occurs in those terrapins held closest to the heater, and therefore there appears to be a direct relation between rapid growth and increased prevalence of the disease. The mortality of winter-fed terrapins, as it has occurred during the period from November of one year to March of the next since 19 12, is as follows: Mortality of Winter-fed Terrapins. Year. IQI2- 1913. 1914. I9IS- Mortality. Number fed. Number. Per cent. 546 4° 7 569 41 7 1. 771 522 29 '.493 87 S Year. 1916 1917 1918 1919 Number fed. 2. 126 2.038 (") 2 937 Mortality. Number. Per cent. 227 554 37 18 " All terrapins of the rt>r8 brood were liberated. SUMMARY. The egg production and the hatch of the original and second Carolina and of the Texas brood stocks during their captivity at Beaufort have been reviewed. It appears that the probable average age of individuals of the two Carolina brood stocks in 1921 is 28+ and 18+ years, respectively. Egg production in domestic terrapins has occurred as early as the fourth year. In terrapins fed one or two winters egg pro- duction generally begins in the fifth or sixth year. In hibernating terrapins it rarely occurs before the seventh year. Egg production immediately after penning is small but increases to normal in about the third year of captivity. At least for the first six years of sexual maturity, probably for much longer, it is greater among the fed terra- pins of a certain brood than among those of the same brood allowed to hibernate. It is estimated that maximum egg production occurs when a terrapin is approximately 25 years of age. Terrapins in captivity have been observed to lay as often as five times in a single season. Average annual egg productions as high as 23.6 and 24.5 per female have been recorded for the original Carolina stock and the Texas brood stock, respectively. The maximum average annual egg production of any female under observation has been 35 eggs; the maximum hatch per female, 31.2 young. This record was obtained in 1920 from the 1909 offspring of the "wild" stock. Rats, because of their burrowing into lately made nests and destroying large num- bers of eggs, are serious enemies of the terrapin. The proper numerical relation of males to females for maximum fertility is not known exactly, and it is difficult to ascertain it accurately in view of the habit of pro- miscuous mating among terrapins. One mating, moreover, may give rise to fertile eggs for four years thereafter; that is to say, the life of the spermatozoon in the female after copulation may be at least four years. A 90 per cent hatch, which appears to be normal, can, however, be obtained among well-matured terrapins when the number of males is one-third the number of females. This average percentage of hatch is not increased by the presence of a larger proportion of males. It appears from the study of certain of the domestic broods that while egg laying is small a maximum fertility may THE DIAMOND-BACK TERRAPIN. I I I be obtained, but not always, from broods in which there are 5 to 12 males per 100 females. However, when egg production is substantially increased it appears that there must be more males to obtain maximum fertility. Broods in which the males numbered 24 to 32, respectively, per 100 females produced a normal fertility, even though egg production was large. It seems warranted, then, to conclude that from 24 to 32 males per 100 females are necessary in order to obtain a 90 per cent fertility when the females of a brood average perhaps from 12 to 24 eggs per season. Males are apparently a year slower than the females of the same brood in coming to sexual maturity. The growth of terrapins from birth to their maturity is recorded. The increasing number and percentage of individuals of a given brood reaching the 5 and 6 inch lengths, the marketable size of terrapins, as age increases has been pointed out. In the sixth year 60 per cent of a given winter-fed lot reached the 5-inch length and 6 per cent the 6-inch length. In the seventh year of a given hibernated lot 57 per cent of the brood reached the 5-inch length and 1 per cent the 6-inch length. A large number of each brood of winter-fed terrapins will have reached the 5-inch length by the fourth year. In the fifth year the average winter-fed terrapin will pass the 5-inch mark. Winter feeding not only hastens growth but quickens development of the sex organs and influences toward greater fecundity at an earlier date than occurs among terrapins allowed to hibernate. Offspring of domestic terrapins appear to do as well in captivity as the offspring of " wild " stock. The maximum growth for any terra- pin raised in captivity under observation has been 81 mm. for one year's and 104 mm. for two years' development. Small terrapins seem to make faster growth when fed oysters than they do when fed salt or fresh fish. Adults make good growth when fed fresh fish. Selection of brood stock should not occur before the third year, preferably later. Selection for this purpose should be of the largest individuals of a brood, since there is a positive correlation between size, age, and fecundity. Earlier selection than in the third year is undesirable, since rapid-growing 1 and 2 year old terrapins often have their growth retarded in the following years and at the fifth year are no larger and no more productive than those terrapins which were poorly grown in the first two years. It appears that the more space allowed terrapins in captivity the greater will be their growth within certain limits. Plenty of space may also have a bearing in its possi- ble influence on increased egg production. Mortality among yearling terrapins fed in a warmed nursery house during the winter varied from 7 to 29 per cent from 1912 to 1919. Mortality among the adult terrapins, as has been pointed out in another publication, is about one-half of 1 per cent. There is no doubt that many more than one-half of the young terrapins of any brood will live in captivity to attain a salable size. Yearling Texas and Carolina terrapins and the hybrids of these stocks seem to be equally susceptible to a disease, probably bacterial, which has killed many young terrapins in the nursery house at Beaufort. Disinfecting treatments of the nursery house and boxes have not proved to be a consistent control of the disease. NOTES ON HABITS AND DEVELOPMENT OF EGGS AND LARWE OF THE SILVERSIDES MENIDIA MENIDIA AND MENIDIA BERYLLINA. By SAMUEL F. HILDEBRAND, Assistant, U. S. Bureau of Fisheries. & Contribution from the U. S. Fisheries Biological Station. Beaufort, N. C. INTRODUCTION. The present paper embodies the results of observations made on eggs, larvae, and adults of the silversides Menidia menidia and Menidia beryllina. All observations were made on living or fresh material, in the immediate vicinity of the Fisheries Biologi- cal Station, Beaufort, N. C, and they extend over a period beginning in April, 1914, and ending in August, 1916. The eggs used in this work were artificially spawned and hatched in the laboratory. The descriptions and drawings are offered with the view of affording means of identi- fving the eggs and larva? with the adult. The two closely related species under consid- eration are compared and contrasted in order to show likenesses and differences in their habits and development. Menidia menidia, SILVERSIDE. ADULTS. This fish belongs to the family Atherinidae, the silversides, which arc elongate shapely fishes with a silvery lateral stripe. Most of them are of small size, inhabiting fresh or salt water of temperate or tropical latitudes, and they usually run in I schools. The genus Menidia may lie distinguished from related genera by the strongly curved premaxillary, the narrow bands of teeth on the jaws, the short lower jaw, which is included in the upper when the mouth is closed, the rounded abdomen, and the smooth, firm scales. There are only two species of the genus known from North Carolina waters. The present species may be distinguished from Menidia beryllina, the other species, by the larger size, by the longer anal fin, which consists of one spine and 21 to 26 soft rays, by the more posterior position of the dorsal fins, and by the black peritoneum. This species is exceedingly abundant in the vicinity of Beaufort, X. C, inhabiting both salt and brackish water, and it is the only fish which occurs in large numbers in the shallow waters throughout the winter. Large schools may be seen along the shores of Pivers Island during the coldest days, when practically all other fishes have migrated to deeper water or to a warmer latitude. 7T5S10— 22 IJ3 II-l BULLETIN OF THE BUREAU OF FISHERIES. It was learned from the examination of large collections that the ratio of males to females is about equal. It so happens, however, for unknown reasons, that at times a school which consists almost wholly of females is taken, and again the reverse is true. The females are constantly somewhat larger than the males, their average length being about ioo millimeters, with a maximum length of 123 millimeters. The average length of the males is about 89 millimeters, with a maximum length of 112 millimeters. The food of this species consists of small fish, crustaceans, algae, and diatoms. SPAWNING. Spawning occurs from early spring to late summer, ripe or nearly ripe fish having been taken by the writer during March, April, May, June, July, and August. Eggs of several sizes are present in the ovaries at one time. When one lot is spawned, the eggs of the next lot are large enough to be seen with the unaided eye. The presence in the ovaries of several different sizes of eggs at one time -strongly suggests that spawning occurs more than once and perhaps several times during the season. An average-sized female produces as many as 500 eggs at one time, and the eggs can easily be hatched artificially. The writer has hatched them during cool weather by merely placing them in a shallow glass dish at the time of fertilization and leaving them undisturbed and without change of water until the young fish appeared ; but when the weather is warm an occasional change of water is essential. The eggs are deposited in shallow water along grassy shores, where large schools of fish collect for this purpose. Capt. Charles Willis,1 a resident of Morehead City, N. C, found a very large school of silversides spawning among eelgrass, in shallow water near Morehead City in May, 1915. The eggs were attached to the vegetation in clusters and became exposed at ebb tide. He then collected and preserved about a quart of the eggs, together with several specimens of the fish which he later exhibited to the author. EGGS. The eggs of this species are approximately \% millimeters in diameter and are slightly heavier than sea water. Their form is spherical when spawned, and they remain so until hatched. The eggs to the unaided eye appear to be separate when first spawned, but as soon as exposed to water, opaque threads of considerable length become visible at the upper pole of the egg. If the water is agitated, the threads become visible more quickly than they do in quiet water; likewise the eggs appear to become attached to objects in the water or to each other more quickly. The addition of the sperm, too, seems to hasten the formation of the opaque threads. A microscopic examination, how- ever, showed that the eggs are delivered in a transparent gelatinous mass, consisting of more or less definite strands, but, as already indicated, the mass does not become opaque and definitely threadlike until exposed' to the elements. The strands again become transparent after they have been exposed for a somewhat variable period of time, and then they are elastic like rubber and of very remarkable strength. It was necessary to use glassware and glass apparatus for incubating and handling the eggs, as they readily adhered to all other materials with which they were placed in contact. 1 Capt. Willis was employed lor several seasons by Mr. Russell J. Coles in the latter's investigations of the fishes of Cape Look- out. It was through this employment that Capt. Willis's interest in the natural history of fishes was stimulated. EGGS AND LARViE OF MENIDIA MENIDIA AND M. BERYLUNA. 115 It is obvious, then, that the purpose of these gelatinous strands is to afford ready means by which the eggs may attach themselves to vegetation or other objects in the water. It is likewise apparent that because of this provision the parents spawn in places where there is an abundance of vegetation. The eggs are yellowish green, as seen in a mass with the unaided eye, but when seen singly under magnification they are semitransparent, and the slight greenish pig- ment appears to be in the individual granules of the yolk. It is also seen that what appeared to the unaided eye to be a single thread by means of which the egg becomes attached is really a bundle of very fine strands of uniform size. A large fat globule, occupying a central position, is always present, and smaller globules, from a few to several in number, are variously distributed. There is a perceptible space between the egg membrane and the vitelline membrane which varies in width. It is broadest at the germ disk and narrowest opposite this point; that is, the yolk sphere occupies the upper part of the egg sphere (fig. 85). EMBRYOLOGY. The protoplasm becomes concentrated after fertilization has taken place at the lower pole of the egg, forming a cap on the yolk of the egg. This mass of protoplasm is the blastodisk. The first cleavage plane cuts the blastodisk parallel with the axis, passing through the upper and lower pole of the egg (fig. 86). The second is at right angles to the first (fig. 87). Cleavage in these eggs is quite regular (fig. 88) and proceeds rapidly in a relatively high temperature, but it is much retarded in a relatively low water tem- perature; for example, the stage represented in figure 89 was reached in about six hours in a water temperature of 84° F., but in a water temperature of about 400 F. the time required to reach the same stage was approximately 48 hours. The outline of the blastoderm on the yolk after an advanced cleavage stage is reached is only indistinctly visible, and the development can not be clearly observed. Within 12 hours after fertilization with a water temperature of 840 F., or within about 60 hours with the temperature of the water at approximately 6o° F., the outline of the embryo, however, may be seen (fig. 90). 3 It is curved with the periphery of the egg and is some- what less opaque than the remainder of the blastoderm. Within 24 hours after fertilization with a water temperature of about 840 F., or within approximately four days with a water temperature of about 6o° F., the embryo is distinctly formed and has about 12 somites. It now extends at least half the distance around the periphery of the egg, and only one large fat globule remains (fig. 91). Two days after fertilization with a water temperature of 840 F., or about seven days with a water temperature of 6o° F., the heart begins to pulsate and large blood vessels may be seen traversing the yolk. The blood at first contains relatively few corpuscles, which flow slowly, but their number and speed increase rapidly with the development of the embryo (fig. 92). Soon after circulation is well established large dark-green or brownish chromatophores appear on the yolk and smaller ones on the embryo. The embryo by this time has fully encircled the egg and is segmented through- out and capable of considerable movement, the tail being partly free. The period of incubation is about 16 days in a water temperature varying from 40 to 6o° F. 2 The author's attention was called to an irregularity iu this figure by Dr. Albert Kuntz, who kindly examined the illustra- tions and read the manuscript, suggesting that an abnormal eg*; was probably studied, as the outline of the advancing blastoderm should he regular and not broken as shown in the figure. u6 BULLETIN OF THE BUREAU OF FISHERIES fr*s~-> 85 87 86 ea — --' — — -os 90 Egg of Menidi a menidia, X 34. (Drawn by Mrs. E. 1J. Decker.) — Egg shortly alter fertilization: at. adhesive thread';: gd, germinal disk; [g, fat globules. — Egg in 2-cell stage. — Egg in 4-ceIl stage, surface view. — Egg in 32-cell stage, with it cells visible in side view. — Egg in advanced cleavage stage. — Egg in stage showing first outline of embryo: 3 ea, embryonic area; cs, cleavage space. (The blastoderm appears to project above the yolk of the egg more prominently in some eggs than in others.) Fig Pig Etc.. 86. 87. 88. ElG. 89. Mrs. 90. EGGS AND EARV.E OF MENIDIA MENIDIA AND M. BERYLLINA. 117 3V /— i % — --^X- — *v ac bv — £-*-:£ -V •>., — ». . . i . A h. 91 D2 93 94 Mmidia matidia. (Drawn by Mis. 1: B. Decka Fig. 91— Surface view ofegg 2 days after fertilization, water temperature, Si P.: es, embryonic shield; iv, third ventricle 01 the brain; 4V. fourth ventricle; ac, auditory canal. X 37. Fir.. 9i.— Egg slightly more advanced than fig. 91, showing large blood vessels (bv) transverstng the yolk, arrows indicating the direction of the flow; h. heart. X 37. Fig. 93.— Newly hatched larva. X 20. Fig. 94.— Youni; fish. 13 mm. in length. X 9- Il8 BULLETIN OF THE BUREAU OF FISHERIES. LARWE. The newly hatched larva is approximately 5 millimeters in length. Considerable coiling, obviously, is necessary in order that a fish of this length may be contained within an egg only 1.25 millimeters in diameter. The larvae are very slender, having an extremely long tail. They are very active, but for want of proper food or other causes they lived only a short time (usually about three days) in the aquarium after hatching. The yolk is completely absorbed by this time, but no important structural changes are evident. The newly hatched fry are highly transparent, only a few yellowish-green pigment spots being present. A few large spots occur on the upper surface of the head and a row of smaller ones along the base of the ventral fin fold. Circulation, due to the transparency of the newly hatched fish, can be nicely observed with a low power of the microscope. The blood may be seen flowing through the aorta to the tail, then curving sharply and returning through the caudal vein. One large blood vessel is still evident on the small yolk yet remaining. The vertical fins appear as continuous folds, sur- rounding the entire caudal portion of the body (fig. 93). Swimming appears to be accom- plished by the very rapid vibration of the tail. The young fish has assumed practically all of the characters of the adult when it reaches a length of 13 millimeters, and it is then easily recognized (fig. 94). Menidia beryllina, SILVERSIDE. ADULTS. This fish differs from M. menidia in the smaller size, in the shorter anal fin, which consists of 16 to 19 soft rays, in the more anterior position of the dorsal fins, and in the pale silvery peritoneum. It is much less abundant than M. menidia, and it inhabits only fresh and brackish water in the vicinity of Beaufort, N. C, although it is reported from salt water from other localities. It is fairly common in the mullet pond ?nd in the fresh or brackish water creeks along the inland waterway and above the "narrows" in Newport River. The females, as in the foregoing species, are somewhat larger than the males, reach- ing an average length of about 61 millimeters and a maximum length of 72 millimeters, while the males reach an average of only about 55 millimeters and a maximum length of 65 millimeters. The ratio of the males to the females appears to be about equal. The food of this species consists of small fish, small crustaceans, diatoms, and algae. SPAWNING. This fish, as does M. menidia, spawns throughout the summer, or from March to September. The author has taken ripe or nearly ripe fish during March, April, May, June, and July, but the small size of the young taken during October and November indicates that it spawns as late as September. This species also selects shallow water with an abundance of vegetation for its spawning ground. It is quite evident that this fish, as does M. menidia, spawns several times during one season, for it has ova of several sizes in the ovaries at one time, and when one lot of eggs is ripe those of the next largest size are big enough to be plainly visible with the unaided eye. DEDUCTIONS CONCERNING THE AIR BLADDER AND THE SPECIFIC GRAVITY OF FISHES. By HARDEN F. TAYLOR, Chief Technologist, U. S. Bureau of Fisheries. The function of the air bladder in fishes has been a subject of discussion since the time of Aristotle, all manner of uses having been suggested for it — flotation adjustment, sense organ, manometer, barometer, respiration reservoir, sound producer, steering device, and the like. The most obvious function is that having to do with adjusting the equilibrium of the fish in water, and, knowing as we do the bewildering diversity of modification of most other organs among the thousands of species of fishes which are known, it is not at all surprising that the air bladder has also been adapted to perform many functions other than the principal and obvious one. For a brief description of the air bladder of fishes and a discussion of its functions see Tower (1902), Goodrich (1909), and Giinther (1880). It is not necessary in the present connection to consider the various secondary functions of the air bladder. The mere fact of its presence necessarily affects the specific gravity of fishes, and it is on this point that this discussion centers. The specific gravity of the fat-free substance of salt-water fish (including backbone, but not the head and viscera) can be shown to be about 1.076. For the present purpose that figure will be taken for the whole fat-free fish. Full sea water has a specific gravity of about 1.026. A 10 kg. fish, fat-free and exclusive of air bladder or other spaces, [OOOO ,. , . . .,., . . , 9293 cc. In order to be in equilibrium with sea water would have a volume of 1. 076 An air bladder or IOOOO of the specific gravity mentioned, it should displace — - — - =9746 cc. other space of 9746 — 9293=453 cc. is necessary to give the fish the required displace- ment. For water of increasing salinity, from pure fresh water to most concentrated sea water, the following table shows the corresponding air-bladder volume necessary to float each 10 kg. of fish whose specific gravity is 1.076: Specific gravity of water. 1. 000 1.005 i. 010 1. 015 Volume of air bladder. Cc. 706 °S7 608 S59 Pressure in air bladder, millimeters mercury. 760 817 882 9S9 Specific gravity of watt-r I. 020 i. ais i. 020 1.030 Volume of air bladder. Cc. S" 403 4S3 416 Pressure in air bladder. millimeters mercury. 1,030 ■»I59 i,iS8 1,289 92486°— 22 122 BULLETIN OF THE BUREAU OF FISHERIES. This table shows that if fish live in and are adjusted to fresh water and if they travel seaward it will be necessary for the air bladder to become smaller. In those, fishes in which the air bladder is closed (as in all the Acanthopteri, or spiny-rayed species, which are typically marine) the volume may presumably be reduced by resorption of some gas in the blood and the discharge of it into the sea, or the gas volume may be compressed, in which case the pressures developed in the air bladder corresponding to the various salinities are shown in the table in terms of millimeters of mercury, the assumption being made that when the fish is adjusted in fresh water its air-bladder, pressure is i atmosphere (760 mm.). In nearly all the teleost fishes, except the spiny- rayed fishes, the air bladder is provided with a pneumatic duct connecting either with' the alimentary canal or with the exterior. Presumably, excess of gas may be expelled through this duct. Some of our most important species, such as salmon, shad, and herring, have this duct. If the fish lives in and is adjusted to sea water and travels in the direction of at diminishing salinity gradient, the conditions are entirely different, for in this case a migration toward fresh water will demand an enlarging air bladder. If, however, the air bladder is at 1 atmosphere when the migration begins, then the pressure must become less than 1 atmosphere, or a partial vacuum must be established in the air bladder, which seems quite improbable. As an alternative to this we may suppose the gas to be absorbed into the blood from the surrounding sea water and discharged into the air bladder. Apart from the physicochemical and physiological difficulties involved in this gas transference against pressure, it is obvious that the mere pumping of gas into the air bladder will be without influence on the specific gravity of the fish if it merely develops pressure and will be effective only in so far as it actually expands or stretches the fish to a larger size. This method of reducing specific gravity appears quite as improbable as the method that would involve a partial vacuum. Unless some other means is found the fish will have to maintain itself afloat by constant muscular effort if it goes to water of a lower salinity. The specific gravity of a fish varies with the amount of fat present in the tissue. In fact, Bull (1896, 1897) investigated the possibility of determining the fatness of fish quickly and simply by determining the specific gravity of the fish, and his results, while not altogether satisfactory, are promising. The foundation of this work is, of course, the fact that the specific gravity of fish fat or oil is less than 1 (usually about 0.925), while that of fat-free substance is greater than 1 (about 1.076), that is, fats float on water, while fat-free fish substance sinks. When amounts by weight (Wt and W2) of two sub- stances of different specific gravities (5, and S2) are combined, the resultant specific gravity of the whole (Si+2) is given by the formula' : ■5'+2 Wl+]v, ir,.s',+H',,s1 ' ( On a percentage basis (where \Vi + \V2= 100) : 100 S^g :il+3~W1Si+W3S1' 1 The formula for this relation given by Bull (1897, p. 641), ~ . is in error. In this formula F and/, Tand/, V-'andl are, respectively, the weight and specific gravities of fat, dry susbtance, and water. 120 BULLETIN OF THE BUREAU OI" FISHERIES. EGGS. The eggs of this species are not quite spherical when first spawned, and they are some- what smaller than those of M. menidia, their greatest diameter being approximately 0.75 millimeter. The action of the eggs after spawning, with respect to adhering to objects in the water, is identical in the two species; but under magnification it is seen that in the present species the gelatinous threads are comparatively few in number, and one of them is always much enlarged (fig. 95), while in M. menidia, as already shown, they are very numerous and of uniform size. The eggs of the two species appear to be identical in all other respects EMBRYOLOGY. Nothing essentially different from M. menidia in the embryology of this species was noted (figs. 95, 96, and 97). The eggs hatched in a water temperature varying from 78 to 820 F. on the eighth, ninth, and tenth day after fertilization. LARWE. The newly hatched larvae are approximately 3.5 millimeters in length, or 1.5 milli- meters shorter than the newly hatched larva- <>!' M. menidia, and it is only in size that they can be readily distinguished from the latter (fig. 98). Therefore, what was previously stated with respect to the larva:' of M. menidia also applies to the present species. The development of the young fry, too, appears to be identical, except that in the present smaller species the development with respect to size is proportionately slower, although from field observations it is evident that maturity is reached just as early. EGGS AND LARVAE OF MENIDIA MENIDIA AND M. BERYLLINA. II9 gd-— • at- ■--^A.-^g 7>* ac — U4~f£ Mcnidia berytUna. (Drawn by .Mrs. E. B. Decker.) Fig. 95.— Egg in j-cell stage: at, adhesive threads; gd. germinal disk; fg, tat globules. X 67. Fig. 96.— Egg in advanced cleavage stage. X 67. Fig. 97— Egg i'A days after fertilization, vrater temperature 820 F. : ac, auditory canal. X 67. Fig. 98.— Newly hatched larva. X 72. AIR BLADDER AND SPECIFIC GRAVITY OF FISHES. 123 Calculating the specific gravities of a fish on the two figures assumed above (0.925 for fat, 1.076 for fat-free substance), we have the following specific gravities of fish with increasing percentage of fat : Specific Per cent gravity of fat. of fish. O I . 0760 2 1 . 0726 4 1 . 0692 6 1.0656 8 1 . 062 1 10 1.0587 12 1-0553 Per cent of fat. 14- 16. 18. 20. 22. 2934 46.55- Specific gravity of fish. 0519 0486 °453 0419 0385 0260 0000 Thus, a fish which increases in fat content diminishes in specific gravity. At 29.34 per cent fat the fish without air bladder would be in equilibrium with sea water and at 46.55 per cent fat (if it were possible) would float in fresh water without air bladder. How does the air bladder react to those changing specific gravities? As the fish grows fatter the air bladder must occupy less and less space, as shown in the following table. This table begins with a 10 kg. fish without fat and shows how the air bladder must change as the fish adds fat, so that each weight has the percentage of fat indicated, the other constituents of the fish remaining constant. Fat in fish (per cent). Total weight of fish. Specific gravity Of fish substance. Volume of solid fish substance. <:) Displace- ment nec- essary to float fish in sea water, specific gravity= 1.026. V.I.026/ Air bladder volume necessary. c—d 4... 6.. 8... 10... 12.. . 14... 16... 18... 20.. . 22.. . 2934 G. 10,000 10, 204 10,416 10, 638 10,869 11,111 11.363 11,628 11.904 12,19s 12,500 12,820 14.152 1.0760 1.0726 1. 0692 1.0656 1. 0621 1. 0587 1.0553 1-0519 1.0486 1.0453 1. 0419 1. 0385 i. 0260 Cc. 9.293 9.513 9.741 9-983 10.233 10,495 10, 767 11.054 n.351 11,660 11,997 12.335 13.793 Cc. 9.746 9.945 10,152 10, 368 «0, 593 10,829 11.075 11.323 11,602 11,886 12,183 12,495 13.793 Cc. 453 432 411 38S 360 334 308 379 251 220 l86 160 These figures demonstrate clearly that as the fish becomes fatter the • specific gravity of the fish substance diminishes and the necessary air-bladder volume becomes smaller and smaller. Thus, in a fish of 22 per cent fat (which is not uncommon in herring and salmon) the fish would be in equilibrium with sea water (specific gravity = 1.026), with scarcely more than a third (160 cc.) as much air-bladder volume as would be required for a fish free from fat. At 29.34 Per cent fat the fish would be in equilib- rium with sea water without an air bladder. Thus, fat can take the place of the air bladder and make the latter unnecessary. We saw above that a fish must find some means of increasing its displacement if it is to migrate from salt water to fresh water and not sink. With the several incre- Specific Percent gravity of fat. of water. 14 1 . 0097 16 1 . 0084 18 1 .0062 20 I. 0040 22 1 .0025 25.3 I. 00OO 124 BULLETIN OF THE BUREAU OF FISHERIES. ments of fat the fish would be in equilibrium with water of the specific gravities shown below without any change of air bladder volume of 453 cc. per 10 kg. body weight. Specific Per cent gravity of fat. of water. o 1.0260 2 1 . 0239 4 1 . 02 1 7 6 1-0193 8. . : 1.0171 10 1. 01 58 12 1. 01 27 It appears from these figures that a fish living at sea and accumulating fat would find itself more at home physically in fresher water without the necessity of enlarging the air bladder. May this not be the influence which directs salmon and shad from salt water to the mouths of rivers? Indeed, it seems unavoidable to conclude that such difficulties in navigation as are introduced by 20 per cent and more of fat must have a profound influence on the movements of the fish. The changes that take place in composition of salmon have been extensively studied. The fluctuations of body com- position of the salmon at different stages of the life cycle were studied by Miescher- Ruesch (1880), Paton (1898), Greene (1914, 1919), and those of the herring by Hjort (1914). Briefly, the career of the Atlantic salmon at sea is as follows: Two or two and one-half years getting its growth, developing muscle and bony tissue (a period obvi- ously of high body specific gravity), then in the third year the accumulation of much fat, whereupon the fish moves to fresh water in the fourth year, when the spawning migration is performed. The large accumulation of fat is consumed in the develop- ment of the reproductive organs and in supporting the fish during the journey in fresh water when no food is taken. We do not know so much of the shad as we do of the salmon, but what information we have agrees in a general way with the above. Apparently, therefore, the fish can not well go into fresh water before a sufficient quantity of fat has been accumulated, because of difficulties in keeping afloat. After the fish has accumulated the fat there would seem to be a strong influence directing it to fresh water. There is another possible means of overcoming the excessive buoyancy of fat in sea water. That is, the fish may descend until the pressure of the water, by reducing the volume of the air bladder, reduces the displacement of the fish to the necessary extent. This reduction of displacement, with increase in specific gravity, must always occur, in any event, when a fish containing an unprotected air bladder descends. The deeper the fish goes into the water the more easily it descends, the excess weight of the fish becoming greater and greater. If the fish begins to rise, the excess weight over displacement, which must be overcome by muscular exertion, becomes less. The effect to be realized from this cause depends, however, on the volume of air bladder present when the fish is at the surface, for, obviously, when, say, 500 cc. of gas is compressed to 250 cc, a greater difference in specific gravity will result than when 100 cc. of gas is compressed to 50 cc, though the same amount of pressure would be required in either case. Therefore, in a fish whose air bladder had been reduced (if such reduction really occurs) in response to AIR BLADDER AND SPECIFIC GRAVITY OF FISHES. 125 accumulating fat, the effect on specific gravity of diving would be correspondingly reduced. In the following table there is shown for the several percentages of fat the correspond- ing depth to which a fish must descend in fresh and salt water, respectively, so that the pressure of the water would equalize the displacement by compressing the air bladder. There is also given the excess weight in grams over the displacement which the body of a 10 kg. fish would acquire by diving 10 m. in water whose density is 1.026 if the fish is in equilibrium at the surface. Per cent fat . Depth (or equilib- rium, fresh water. Depth for equilib- rium, sea water, specific gravity = 1.026. Load in excess of displace- ment, 10 meters depth. Per cent fat Depth for equilib- rium, fresh water. Depth for T„ . ■ «"""» e^cess'o"! Cm. Cm. G. 253 Cm. 486 644 831 1,094 J,S8: 1,881 Cm G. 474 IS8 6j8 50 :o5 182 266 368 49 102 177 259 JS8 211 16. 810 130 6 18 8 186 1.542 96 1,833 The air bladder is present in the great majority of fishes. In the Selachii it is absent, in the teleosts it is generally present, although the Heterosomata or flatfishes, Xiphias, the swordfish, Menticirrhus, all the Alepocephalidae, and a few other families or genera are without air bladders. In all the spiny-rayed fishes (which are typically marine) the air bladder when present has no outlet duct in the adult fish. Any reduc- tion in volume of the air bladder must therefore be accompanied, at least temporarily, by pressure. The pressure might be relieved by absorption of gas into the blood. It does not seem at all unlikely that the varying salinities of ocean water guide such fishes as the mackerel, tuna, herring, bluefish, sharks, and many others. Temperatures, oxygen and CO, content, plankton, and other food supply have been studied as directing influences, but it is difficult in every case to show what the immediate effect of the in- fluence is on the fish. In the case of specific gravity the direct effect is obvious and unavoidable. It would be exceedingly unsafe to make assumption as to what method the fish uses to maintain itself in equilibrium with the water. We know from the work of Tower (1902) and others that the composition of the gas in the air bladder of fishes varies. It contains more and more oxygen with increasing depths, so that fishes taken from great depths have nearly pure oxygen in the air bladder. It was shown that the air bladder very probably performs an important respiratory function. Certainly the loading into the blood stream or the removing therefrom of large quantities of gas could not possibly fail to have a profound effect on the physiological functions of the fish even if fatal embolisms did not occur. The following conclusions are drawn : 1 . Fish on migrating from water of low salinity to that of high salinity may adjust specific gravity by reducing the size of air bladder. In the reverse direction there is no apparent means for voluntary adjustment. 126 BULLETIN OF THE BUREAU OF FISHERIES. 2. As a fish puts on fat its body specific gravity diminishes, pari passu, and in pro- portion to the amount of fat present (a) its navigation in salt water is more difficult; (b) fresher water is better suited as a physical medium. Until a certain amount of fat is accumulated migration from salt to fresh water must be difficult or impossible. 3. Reduction of volume of air bladder may possibly be effected by (a) resorption of gas from bladder to blood and expulsion through gills; (b) direct expulsion of gas through pneumatic duct (except in Acanthopteri) ; (c) diving, whereby hydrostatic pressure reduces the volume. The effect produced by diving a given depth is propor- tional to the absolute volume of the air bladder. 4. Diminishing specific gravity consequent upon increasing fatness probably con- stitutes a strong directive influence governing the movements of fishes, both marine and anadromous. LITERATURE CITED. Bull, Henrik. 1896. Enkel Methode til Bestemmelse af den ferske Silds Fedtholdighed. In Aarsberetning fra Fors0gsstationen og Laereanstalten for Tilvirkning af Fiskeriprodukter i Bergen for Aaret, 1895-1896. Norsk Fiskeritidende, isde Aargang, 1896^.549-557. Bergen. 1897. Om Egenvaegt og Fedt i Sild, Torskelever, m. m. Ibid., i6de Aargang, 1897, p. 294-300. Goodrich, E. S. 1909. A treatise on zoology, edited by Ray Lankester, Part IX, Vertebrata Craniata (First fascicle: Cyclostomes and Fishes) (p. 223 and 359-362). 518 p. Adam and Charles Black, London. Greene, Charles W. 1914. The storage of fat in the muscular tissue of the king salmon and its resorption during the fast of the spawning migration. Bulletin, U. S. Bureau of Fisheries, Vol. XXXIII, 1913 (1915), p. 69-138, XI pis. Washington. 1919. Biochemical changes in the muscle tissue of king salmon during the fast of spawning migra- tion. The Journal of Biological Chemistry, Vol. XXXIX, 1919, p. 435-456. Baltimore. Gunther, Albert C. L. G. 1880. An introduction to the study of fishes (p. 142-149). 720 p. Adam and Charles Black, Edinburgh. HjORT, JOHAN. 1914. Fluctuations in the great fisheries of northern Europe, viewed in the light of biological research. Conseil Permanent International pour 1' Exploration de la Mer. Rapports et Proces-Verbaux, Vol. XX, 288 p., 3 pis. Copenhague. Miescher-Ruesch, F. 1880. Statistische und biologische Beitrage zur Kenntnis vom Lcben des Rheinlachsos im Siiss- wasser. Schweizer Katalogue der Fischerei-Ausstellung zu Berlin, 1880. p. 154-232. Leipzig. Paton, D. Noel [editor] and OTHERS. 1898. Report of investigations on the life history of the salmon in fresh water: From the Research Laboratory of the Royal College of Physicians of Edinburgh. Fishery Board for Scotland, Salmon Fisheries, iv+176 p., 14 pis. Glasgow. Tower, R. W. 1902. The gas in the swim bladder of fishes. Bulletin, U. S. Fish Commission, Vol. XXI, 1901, p. 125-130. Washington. o c p. ! E : < '"So - = 2 .E ^. 3 ■p ja -0 w ,- X3 "C a * - "o w> - I* 1 £ * s a £ - o - s a u o z; J2 w — 1 s 3 S $ - X Bull. U. S. B. F., 1921-22. (Doc. 922.) ^^■^flB SS^ssffl '*?•»->■. IB 1 ^^Mh^C v Li h ■e Hf Wr/" . ' 3?£j£ / » ■ MSEif^^ -"l* i w-g IgT'y3" >* fe 4- S . - *E i <* Mfc-^*->*v«r-* wBF rt^^Hr k Fig. 100 —The sea mussel Mytilus edulis Linnaeus, showing foot distended and attachment by byssus. Fig. ioi.— A bed of sea mussels in Menemsha Pond, Marthas Vineyard, Mass., exposed at low tide, of surface is covered with more than a bushel of the shellfish. Each square yard BIOLOGY AND ECONOMIC VALUE OF THE SEA MUSSEL Mytilus edulis.1 J- By IRVING A. FIELD, Late Professor of Biology in Clark University and Special Investigator, V . S. Bureau of Fisheries. Contribution from the U. S. Fisheries Biological Station, Woods Hole, Mass. CONTENTS. Page. Page. Introduction 128 Systematic and economic relations of mol- lusks 128 Anatomy and physiology of the sea mussel . . 131 The shell 131 Description 131 Histology 134 Formation 137 Attachment to the body 137 Chemical composition 138 The mantle 138 Digestive system 139 Anatomy 139 Histology 142 Physiology 145 Circulatory system 147 Heart 147 Arterial system 148 Venous system 151 Blood 154 Physiology 155 Muscular system 157 Foot and byssus 159 Anatomy and histology 159 Physiology 160 Chemistry of byssus 163 Respiratory system 163 Anatomy 163 Histology 164 Physiology 166 Excretory system 168 Anatomy 168 Histology 169 Physiology 169 1 After this report was completed and submitted to the Bureau but beiore it could be published Dr. Field 's untimely death occurred in February, 192 1. Consequently there was no opportunity for the author to review the final editorial corrections or to read the proof. The report was submitted to the faculty of Clark University. Worcester, Mass.. in partial fulfillment of the requirements for the degree of doctor of philosophy. 127 Anatomy and physiology of the sea mussel — Continued. Nervous system 170 Sense organs 174 Anatomy 174 Histology 177 Physiology 181 Reproductive system 182 Anatomy 182 Histology 182 Physiology 185 Embryology 186 Germ cells 186 Maturation and fertilization 189 Cleavage and formation of germ layers. . 190 Development of the trochophore larva . 192 Transition to the adult 194 Mantle 194 Shell 194 Alimentary organs 195 Muscles 195 Gills 196 Kidney 196 Nervous system 196 Sense organs 199 Pericardium 199 Genital organs 199 Growth 200 Food of the sea mussel and its significance . . 203 Enemies and parasites 215 Inanimate destructive forces 215 Active enemies 115 Starfish 215 Drills 216 128 BULLETIN OF THE BUREAU OF FISHERIES. Page. Enemies and parasites — Continued. Active enemies — Continued. Other gastropods 217 21S 218 219 219 219 219 Fishes Birds Mammals . . Passive enemies Eelgrass .... Algae Invertebrates ... 220 Parasites 220 Polydora ciliata 220 Haplosporidium mytilovum, n. sp. . 220 Uses and commercial value 222 Chemical composition and nutritive value . . 224 Pa?e. Seasonal changes in structure and food value 225 When mussels are unfit for food 227 Mussels and typhoid fever 227 Ptomaines 228 Peculiar poisons 229 Sources of poison 230 Chemistry of mussel poison 232 Conclusion 234 Cultivation of mussels 234 Duration of mussel beds 240 Efforts to develop a mussel industry in the United States 241 Summary, conclusions, and recommenda- tions 245 Bibliography 247 INTRODUCTION. The object of this report is to present as completely as possible the facts known concerning the biology and economic importance of the sea mussel Mytilus edulis Linn, and the possibilities of developing a mussel fishery in the United States. In a previous paper (Field, 191 1) the food value of the sea mussel was demonstrated to be equal to or greater than that of any other commercial shell-fish on our coast, and the mussel beds of our eastern and western coasts were shown to constitute one of our great undeveloped marine food resources. The importance of this sea mussel as a valuable source of food supply was considered so great that a more exhaustive study of the life history, distribution, and the commercial possibilities of utilizing the species was considered advisable. The results of this investigation show the possibility of adding to our food supplies millions of pounds of wholesome flesh food annually. The material entering into this report is based upon the review of an extensive literature verified and supplemented by a series of investigations carried on during seven summers for the United States Bureau of Fisheries at its biological station at Woods Hole, Mass., and by a reconnaissance of the mussel beds on a limited portion of the north Atlantic coast. A considerable portion of the work was done in the biological laboratories of Clark University. SYSTEMATIC AND ECONOMIC RELATIONS OF MOLLUSKS. The term mussel, as generally used, applies to either of two groups of bivalve mol- lusks one of which is restricted to salt and brackish waters, the other to a fresh-water habitat. The marine species belong to the genus Mytilus and other allied genera of the family Mytilidae. On our eastern coast there are five species of this family represent- ing three genera, the most important of which is the common sea mussel, Mytilus edulis (fig. 100, opp. p. 128), which ranges from the Arctic Ocean to Cape Hatteras. The horse mussel, Modiolus modiolus, is next in importance, ranging from the Arctic Ocean to New Jersey. Mytilus hamatus, the hooked mussel, is found from Chesapeake Bay south- ward and on the Gulf coast. Modiolus demissus (Modio/a plicalula), the plicated mus- sel, is a shallow water form found from Maine to Georgia. Modiolarm nigra is a north- ern form which inhabits the deeper waters from the Arctic Ocean to Cape Hatteras. SEA MUSSEL MYTILUS EDULIS. I 29 On our western coast the family is likewise represented by three genera and five species. Mytilus edulis occurs from the Arctic Ocean to San Francisco. Mytilus calijornianus is common on the California coast. Modiolus modiolus and the straight horse mussel, Modiolus rectus, are also present on the California coast. Modiolaria nigra occurs from Arctic waters to San Francisco. The marine mussels are characterized by a byssus which is secreted from a gland located at the base of the foot. The fresh-water mussels belong to the family Unionidae represented by Unio, Anodonta, Quadrula, and other allied genera. They are particularly common in most of the rivers of the central United States. They secrete no byssus in the adult stage. The phylum Mollusca comprises a great variety of forms; but there is a close re- lation between all the groups, which are merely modifications of the same type. It includes the chitons, Amphineura; snails, Gastropoda; mussels, clams, oysters, scallops, etc., Lamellibranchia; and the nautilus, devilfishes, and squids, Cephalopoda. The characteristic feature of these animals is a ventral, muscular foot which usually serves for locomotion, but is much modified according to habit. The body is soft and moist and usually more or less covered with a shell which is generally either univalve or bivalve; the shell is secreted by a glandular fold of skin called the mantle. The shell often consists of three layers; an outer thick, tough portion, or periostracum ; a middle prismatic layer, which is much thicker; and an inner mother-of-pearl, or nacreous layer, which is sometimes brilliantly iridescent. The adult forms show no sign of segmentation and the body cavity is more or less obliterated. The pericardium represents the chief portion of what is left of the true body cavity. Communication between the pericardium and the exterior is established through the nephridia. The respiratory organs consist of gills except in a few species, chiefly terrestrial, which possess a sort of lung. It is probable, also, that the mantle plays an important part in respiration. The nervous sys- tem consists of three ganglionic centers with connectives located respectively in the head, cerebral ganglion; in the foot, pedal ganglion; and on either side of the oesophagus, visceral ganglia. Sense organs of touch, sight, smell, and equilibrium may be present in the head region. In the development of the Mollusca segmentation of the egg is unequal and the larva? pass through a free-swimming or trochosphere stage which is also the characteristic larval stage of the Annelida. There is also a probable relationship with the Polyzoa. In distribution, we find the Mollusca occupying, in a general way, the whole surface of the earth at all latitudes and altitudes. They are found in the polar, temperate, and tropical regions; in the ocean; along the seashore; on land; and in fresh- water lakes, ponds, and streams. Certain snails of the suborder Stylommatophora have been found in mountains at a height of 15,000 feet; abyssal mollusks have been taken from a depth of 2,800 fathoms. There are pelagic species which are distributed over the surface of the sea, some live on the floating seaweed, while others descend many thousands of feet from the surface. It is within the Tropics, however, both on land and in the sea, that the Mollusca are most abundant both in numbers and varieties. Protective markings of a striking nature are characteristic of many mollusks. Most of the pelagic species are colorless or tinged with blue. The nudibranchs, which are found on the floating sargassum weed, are beautifully marked with yellow and brown like the weed itself. Other species are green or red in color, similar to the algae on which they live. The shellfish which live in the great depths beyond the reach of the faintest 130 BULLETIN OF THE BUREAU OF FISHERIES. ray of light are characterized by thin, colorless shells, a highly developed tactile sense, and the absence of visual organs. The length of life and age of attaining sexual maturity vary considerably for different mollusks. Mytilus reaches the adult stage in one year. The fresh-water mussels, Ano- dontidse, do not reach sexual maturity until they are 5 years old. Some mollusks, nudi- branchs, and the cephalopod, Rossia, appear to live for one year only, while others, as Mytilus and the oyster, may live 10 or more years; the periwinkle, Littorina, has been known to attain an age of 20 years in captivity, and the Anodontidae, which are remark- able for their long life, may reach an age of 25 or 30 years. The Mollusca is an old group whose fossil representatives are found in all Paleozoic deposits upward. As a group it has met the changing conditions of the world most successfully, as is clearly demonstrated by its present abundance and wide distribution. More than 28,000 living Mollusca have been described up to the present time, more than half of which are Gastropoda. The economic importance of the Mollusca is very great indeed. The group includes species of negative as well as of positive value. In the former class may be mentioned the so-called shipworm, Teredo navalis, a boring lamellibranch whose habits are ex- tremely destructive to the bottoms of wooden ships, to wharf piles, and to other sub- merged wooden objects, which are riddled by its borings. To prevent the destructive inroads of the shipworm it is necessary to incase the bottoms of wooden ships with a metal sheath and to coat such wooden objects as spars, buoys, etc., with verdigris paint periodically every six months. Among the gastropods are found many voracious species armed with rasping organs against which few shellfish are safe. They prey upon many species valuable to man, such as oysters, clams, scallops, mussels, etc., by boring holes through their shells and literally eating them alive. The destructive ravages of these snails on the commercial species of mollusks amount to many thousands of dollars yearly. The Cephalopoda are also carnivorous animals of very active and voracious habits. They dart into the schools of young fishes and feed upon them in great numbers. Young lobsters and other small crustaceans often fall prey to them. Molluscan species of positive value to man are numerous and represent every class except the Amphineura. Most of the cephalopods are good to eat and are uti- lized extensively as food in some countries. Although not used as such in the United States, there is no reason why they should remain a neglected food product. Squid is the most valuable bait known in the cod fisheries and for this reason often brings fancy prices. When abundant it is used for fertilizer. The cuttlefish furnishes the cuttle bone which is used as a food for canary birds, and formerly its inky secretion was sold as India ink or sepia, which was used for drawing purposes. The gastropods include species of food value, as, for example, the large, edible snail of Europe, the periwinkle, Littorina littorea, which is eaten by the ton in London, but, as yet, remains unknown as a food in this country. The abalone of our western coast is beginning to be appreciated as a food through the influence of the Chinese, who have developed the fishery into a business worth many thousands of dollars annu- ally. Its shell is remarkable for its great beauty and was formerly used by the Indians for making their money. In Europe it is used for making buttons, studs, and buckles and for decorating purposes. SEA MUSSEL MYTILUS EDULIS. 131 The class Lamellibranchia is the most important of all from the commercial stand- point. It includes the oyster, which furnishes the most valuable fishery of the nation, the receipts from this source alone amounting to one-third the total income derived from all the fisheries of the United States. The flesh of the oyster constitutes a most delicious morsel, and the shells are used in the construction of roads, as a food for poultry, as fertilizer, and as cultch for starting new oyster beds. About 25,000,000 bushels of oysters are utilized in the United States annually. Other valuable edible lamellibranchiate species are the clams Mya and Venus, which have made the New land clam-bake famous throughout the land, and the scallops, which are popular in every hotel and restaurant of our northeastern coast. The fresh-water mussels of our inland waters furnish pearls of rare value and shells especially adapted for the button industry and for the manufacture of articles of much beauty. So great has been the demand in recent years for the important species that the resources have been greatly depleted. Fortunately, however, the United States Bureau of Fisheries has been able to take up the problem, and, by the application of scientific methods, it is now propagating mussels to provide for the increased demand. Not only do the lamellibranchs yield products of commercial value, but in their daily functions they perform a service which has never been estimated in dollars and cents. Their habit of setting up currents of water which are continually filtered through the gill filaments serves to remove the bacteria and other microorganisms along with quantities of floating organic particles which, if left in the water, would lead to stag- nation. They constitute, therefore, one of the great purifying agents of our lakes, ponds, and streams. The United States Bureau of the Census reported the value of the mollusk fish- eries of the United States for 1908 as follows: Oysters S15, 713,000 Hard clams I»3I7. 000 Long clams 553, 000 Scallops 317,000 Fresh-water mussel shells 392, 000 Slugs and pearls 300, 000 Sea mussels 11, 600 Oyster and other shells 20, 000 Squid 43, 000 Cockles, winkles, conchs 35, 000 Total 18, 701, 600 ANATOMY AND PHYSIOLOGY OF THE SEA Ml/SSEL. THE SHELL. DESCRIPTION. • The sea mussel has a general form which may be described as triangular ovate. Anteriorly, in the hinge region, the shell presents its greatest breadth; posteriorly, it becomes narrower and flattened. The posterior edge of the shell is nearly semicircular in outline; in the dorsal region it forms almost a straight line up to the beginning of the hinge, where it bends obliquely downward at an angle of about 450 to the umbo, which is located at the tip of the shell. From the sharp point of the umbo the ventral edge of 132 BULLETIN OF THE BUREAU OF FISHERIES. the shell extends backward in an almost straight line. In specimens of mussels growing on wharf piles in protected situations the ventral edge sometimes presents a slightly convex outline, while, on the other hand, mussels growing on the rocks or mud where they are subjected to swift currents and exposure often show a decidedly concave under surface. The size of the adult mussel varies from 2 to 4 inches in length, from 1 to 2 inches in height, and from %" to 2 inches in breadth. Occasionally specimens 4^ inches long are found. The proportions of length to height and breadth vary with the age of the mussel. Individuals less than a year old show a length, breadth, and height which are to each other as 2.75 : 1.5 : 1; while in older ones they are to each other about as 2.25 : 1.15 : 1, indicating that in adults growth in breadth is proportionately more rapid than in length. The color of the shell varies from violet or blue-black to a pale blue. When dried it takes on a brownish hue. This change of appearance is due to the hornlike covering of the shell, the periostraeum, which is itself brown. The characteristic violet color of the shell comes from the thick prismatic layer which lies immediately below the thin periostraeum and which contains a deep blue pigment. The general hue of the shell is therefore due to a combination of the brownish, transparent periostraeum and the under- lying layer of deep blue calcareous matter. This results in a variety of color variations according to the thickness and density of the periostraeum and the amount and distrib- tion of the pigment in the prismatic layer of the shell. The arrangement of the pigment is in bands which run from the umbo in a radiating manner to the posterior end of the shell. Most commonly the stripes lie so closely together that it gives the shell surface a uniform dark blue color. Sometimes, however, the bands of color are few in number or entirely absent. In the latter case the mussel is colored a uniform brown or yellow- brown by the periostraeum, while in the former case it is marked with alternate blue and brown bands which radiate from the umbo to the posterior edge of the valve. The inner surface of the shell is divided sharply into two regions, an inner, glossy white or pinkish-white mother-of-pearl layer and an outer deep blue border about three- sixteenths of an inch wide. The line of demarcation between these two layers is sharp and may be either straight or serrated in outline. The blue layer is absent in the hinge region. Six impressions which mark the attachment of muscles are conspicuous on the inner white surface: (1) The largest and most prominent is more or less circular in form and located posteriorly near the dorsal border. It marks the point of attachment of the posterior adductor muscle (fig. 104, PAD, opp. p. 132). (2) Running anteriorly from the dorsal edge of .this impression is another, linear in form (fig. 104, PRet), which marks the point of attachment of the posterior retractor muscles of the foot and byssus. (3) A third, somewhat triangular impression lying just posterior to the impression of the posterior adductor muscle marks the insertion of the muscles of the anal membrane (fig. 104, An). (4) At the anterior end of the shell on the ventral border is an impression where the anterior adductor muscle is attached (fig. 104, A Ad). (5) Just above it on the dorsal edge is another which marks the point of insertion of the anterior retractor muscles of the foot and byssus (fig. 104, A Ret). (6) A long, narrow, linear impression extending along the lower edge of the shell from the impression of the anterior adductor muscle to that of the posterior adductor muscle and just inside the border of the nacreous layer forms the line of attachment for the pallial muscles (fig. 104, Pal). Bull. U. S. B. F., 1921-22. (Doc. 922.) LR' \Liq 102 y PRct PAd AR< An Aj\cI 104 'Pal Fig. 102— Interior view of hince ligament. Lig, ligament; LR, ligament ridge FlG. 103. — Interior view of anterior end of a valve showing four hinge teeth, T. Fig. 104. — Inner surface of a valve showing muscle impressions. .1.1/. anterior adductor muscle; An, anal muscle; A Ret, anterior retractor muscle; PAJ, posterior adductor muscle; Pal pallial muscle; PRet, posterior retractor muscle. Fir.. 105.— Dorsal surface of a shell from which periostracum and ligament have been removed. LC, ligament cleft. Fig. 106. — Ventral surface of a shell from which periostracum has been removed. ByC, byssus cleft; I.n. lunula. SEA MUSSEL MYTILUS EDULIS. 133 The two valves are attached at their anterior dorsal edges by means of a hinge plate over which the periostracum extends from one valve to the other. Teeth are present at the anterior end of the hinge in numbers varying from one to six, the most common number being three or four. The teeth of one valve are so arranged that they fit into the depressions between the teeth on the opposite side. In size they are small, rarely over a millimeter in length, and in form they are conical pointed knobs or wedge- shaped lamellae (fig. 103, T). The ligament (fig. 102, Lig), a straight, brownish colored, elastic rod, lies between the two valves just beneath the hinge band and helps to unite the shell edges. In cross section it presents the form of an ellipse with its long diameter lying in a horizontal position. It is bounded laterally by parallel ridges, the ligament ridges (fig. 102, LR), which have a very characteristic structure. They are chalky white in color and per- forated with numerous pores. Each ridge terminates in a fine point both anteriorly and posteriorly. The median surfaces are concave to fit snugly against the liga- ^^^2^^>v^-- //£ ment when the shell is normally open; ^tffZ S^sJ consequently when the valves are closed by contraction of the adductor muscles the ligament is compressed and its elas- ticity tends to counteract the action of the muscles (fig. 107). Asa. esult, when §p j >^---PL the adductor muscles relax the ligament "lik / J • forces the shell open again. This ex- plains why the shells of dead mussels are always open. \^ ^^. -Per The umbo is at the anterior end of the shell and forms a sharp beak, off the ventral side of which may be found, \QJ often hidden by the periostracum, a _ _ ^. , . „. .. .... J L Fig. 107. — Cross section of a shell 111 hinge region in diagrammatic Special Structure called the lunula (fig. lonn to show relation of ligament to hinge plate and valves. HL, IO6, LU, Opp. p. 132), which bears a hinge ligament; HP. hinge plate; ZJ?, ligament ric^c. XL, na- . , . ... creous layer; Per, periostracum; PL, prismatic layer. definite relation to the hinge teeth. It consists of a series of semicircular furrows and ridges which run out peripherally from the teeth and terminate at the umbo. Each furrow corresponds to a tooth; and each ridge, to a depression between the teeth. The lunula is conspicuous only in individuals where the teeth are comparatively large in size and number. When the valves of a normal shell are shut they form a complete closure. If, however, they are first treated with a solution of potassium hydroxide, which removes all the periostracum, it will be found when they are closed that there are two places where the edges fail to come in contact. On the ventral side in the middle of the shell there is a fissure through which the byssus may project, the byssus cleft (fig. 106, ByC). In the normal shell this cleft is hidden by a fold of the periostracum which incloses the marginal blood sinus. The structure is such as to act as a cushion which presses against the byssus when the shell is closed. Corresponding to the byssus cleft on the dorsal side in the hinge region is another opening between the valves, the ligament cleft (fig. 105, LC, opp. p. 132). In the complete shell this depression is covered externally by the periostracum and internally by the underlying ligament. 134 BULLETIN OF THE BUREAU OF FISHERIES. HISTOLOGY. If the shell of the sea mussel is broken or cut in cross section, three distinct layers, sharply defined from each other, are visible to the naked eye: An outer, thin, cuticular layer, the periostracum ; a middle violet-colored portion, the prismatic shell; and an inner glossy white or pink substance, which often reflects iridescent colors, the mother- of-pearl, or nacreous, layer. Under high magnification each of these parts shows a characteristic structure. The periostracum generally covers the whole outer surface of the shell and extends over the free edge for a short distance over the inner surface where it terminates in a fold of the mantle border. It is a smooth, glossy cuticula, thinnest in the region of the umbo, where it is often completely worn off as a result of exposure at low tide and the action of strong waves and currents. The hinge of the shell is formed from the periostracum which extends over from one valve to the other. In cross section the periostracum presents three layers (fig. in, Per, p. 136). The outer and inner portions consist of a clear, transparent, brownish substance which does not readily stain with any of the ordinary dyes. The middle portion is colorless and has an affinity for plasma stains. In the region of the mantle edge from which it is an outgrowth the layer con- sists of a single layer of cells, but as the outer surface of the shell is approached the cells disappear, leaving a series of cavities to mark the middle zone. The periostracum may, therefore, be divided into three distinct areas: An outer, middle or hollow, and an inner layer. The periostracum is attached to the layer of the shell lying imme- diately below it by means of trabeculae which are embedded in the calcareous substance. The blue portion of the shell or middle layer is composed of fine needlelike filaments of calcareous matter closely united into a single structure by an organic matrix of conchiolin (fig. 108, p. 135). When a valve is broken in cross section and examined with a hand lens this layer presents a series of alternate ridges and grooves with glisten- ing surfaces which extend across the shell. With higher magnification it is possible to see that the prisms are long and almost straight and so arranged as to form an angle of about 450 with the outer shell surface (fig. in, PL, p. 136). The pigment, which is more abundant on the peripheral surface of the layer, is deep blue or violet in color and is distributed in the form of parallel bands which run across the prisms at an angle of about 300 (fig. 112, PB, p. 136). Around the ventral and posterior borders of the shell there is no inner nacreous layer present, consequently the prismatic shell lies in direct contact with the outer fold of the mantle edge. The nacreous or mother-of-pearl layer covers the inner surface of the shell out to the mantle line as a boundary. It is thickest in the anterior and middle regions and thinnest at the border. This is the only layer which continues to grow in thickness throughout the life of the mollusk. The nacreous and prismatic layers lie in direct contact with each other without any intervening substance to connect them. The structure of the nacre consists of a series of thin lamellae with irregular edges placed one on the other with their surfaces lying horizontal to the surface of the shell. When seen in cross section of the shell under high magnification they appear as fine irregular parallel lines (fig. 1 11, NL, p. 136). If a portion of the nacreous layer is dried for some time or is treated with sodium hydroxide it becomes fragile and has a tendency to break up into flakes whose surfaces mark the line of cleavage between the separate lamellae. SEA MUSSEL MYTILUS EDUUS. 135 Microscopic examination ot the flat surface of one of these flakes near its edge will show distinctly the leaflike layers with their irregular edges (fig. 109). The transverse plane of cleavage always follows the zigzag edges. A surface view of a single lamella under very high magnification reveals a fine network with meshes of a polygonal form (fig. 1 10). The ligament when examined in cross section with an ordinary hand lens presents three distinct layers; the outer portion is marked by the dark brown periostracum, the middle part is composed of a yellow-brown homogeneous substance, while the inner layer is of the same color but marked with numerous horizontal dark brown lines. Under higher magnification four distinct and separate layers are visible. The outermost layer consists of the periostracum (fig. 113, Per, p. 136), which is really not a part of the ligament proper, although it is fused so closely with it as to form 108 110 Fig. 108.— Transverse section through prismatic layer of shell, showing needlilike prisms, which are hild together by an organic matrix of conchiolin. X 854. Fig. 109. — fragment of nacreous layer showing overlapping lamclhc. X soo. Fig. 1 10.— Surface view of nacreous layer very highly magnified. a unit body. What appears to be the middle layer under low magnification is divisible into two sharply separated parts when examined with stronger lenses. The outer por- tion of this layer presents a homogeneous structure similar to a film of gelatin, with a yellowish color often tinged with blue or green (fig. 113, HyL). The inner part of the middle layer is of a darker shade and strongly granular in appearance, due to an abundance of irregular masses of lime crystals, many of which are aggregated into starlike bodies. A few of these crystals may be seen scattered in the homogeneous layer close to its bor- der next the granular layer (fig. 113, GL). The inner layer is distinguished from the others by its cross-striped appearance (fig. 113, IL). It is marked by numerous fine vertical lines across which there run at right angles many dark brown bands of varying width. Crystals of lime salts are present in small numbers scattered throughout the substance of this layer. The crystals may be readily removed by treating the sections 136 BULLETIN OF THE BUREAU OP FISHERIES. —Per -PL -NL ; ■ V.-'- '■■: ->; -■■ '^•Ztecv^&F^ ---P5 112 P6T 113 114 Fig. hi. — Cross section of piece of shell. X s° approximately. Fig. 1 1 2.— Section through prismatic layer. X 45- Fig. 113. — Cross section through ligament. X 26. Fig. 114. — Cross section through edge of mantle. X 35. Fixed in Gilson fluid and stained with Delafield hematoxylin and congo red. Abbreviations. — GL, granularlayer; HyL, hyaline layer; IF, inner fold; 1L, inner layer containing a few groups of calcium crystals; MF, middle fold, showing origin of periostracum (Pit) from its outer edge; Mus, muscle fibers; AX, nacreous layer; OF, outer fold; PB, pigmented bands; Per, periostracum; PL, prismatic layer. SEA MUSSEL MYTILUS EDULIS. .137 with weak nitric or hydrochloric acid. The staining reactions of the decalcified sections of the ligament are characteristic; the middle layer has a tendency to take up the plasma stains, while the inner layer takes up the basic dyes. FORMATION. The periostracum is an outgrowth from a specialized portion of the mantle edge. If a cross section of the mantle is examined under 1owt magnification it will be seen that the edge of the mantle is divided into three distinct folds which run parallel with its edge. They may be designated by their position as the outer, middle, and inner folds. From the outer surface of the middle fold the periostracum arises as a thin cuticula which is secreted bv a layer of epithelial cells having characteristic, long, elliptical nuclei and fibrils which lie at an oblique angle with the surface of the periostracum. Numerous muscle fibers from the mantle terminate among these cells (fig. 1 14, Mus,p. 136). The youngest portion of the periostracum, which lies in contact with the secreting cells, is a thin, trans- parent, homogeneous structure, but as it extends out beyond the limits of the mantle edge it grows progressively thicker and becomes differentiated into the three layers which have already been described. The periostracum grows over the edge of the outer fold, beyond which it becomes attached to the outer surface of the prismatic layer of the shell. The prismatic or blue layer of the shell is secreted by the low columnar epithelial cells which cover the outer surface of the outer fold of the mantle edge. As fast as the material is built up along the edge of the shell its outer surface comes in contact with the outgrowing periostracum, to which it becomes attached. The epithelium of the outer mantle surface is composed of small cubical cells and gland cells which secrete the mother-of-pearl (fig. 1 14, p. 136). The process is continuous, so that as the animal grows older this layer continues to grow thicker, giving the shell the unusually firm and heavy character which is often noted in old mussels. Exposure to the rough action of waves and currents stimulates the cells to more rapid secretion. The ligament arises from a layer of tall columnar epithelial cells which lies imme- diately below it. ATTACHMENT TO THE BODY. The whole outer surface of the fleshy part of the body is more or less intimately connected with the inner surface of the shell. The epithelium of the mantle forms a rather weak attachment, while the muscles adhere most tenaciously at their points of union with the shell. The epithelial cells of the outer surface of the mantle lie in direct contact with the inner surface of the shell and are attached to it by the secretion of a soft, gummy sub- stance from which the shell is being formed constantly. The attachment may be likened to that of a label pasted on a bottle. In case of the muscle attachments, a very different type of adhesion is found. Here the epithelial cells of a highly specialized nature serve as anchoring organs. They are so intimately attached to the bundles of muscle fibers at their proximal ends that it is difficult to distinguish them from the contractile tissue without applying staining methods. Ordinarily muscle fibers stain more deeply than do the supporting epithelial cells. Distally the epithelial cells are embedded in the surface of the shell, making an attachment so strong that it is impossible to separate the mass from the shell without applying acid to dissolve away the calcareous substance in which they are firmly fixed. 138 BULLETIN OF THE BUREAU OF FISHERIES. CHEMICAL COMPOSITION. The shell of the mussel consists of an organic base infiltrated with mineral salts, as has been shown above in the description of its histological structure. The organic matrix is an albuminoid substance called conchiolin, the composition of which, according to Wetzel (1900), is carbon 52.3 per cent, hydrogen 7.6 per cent, nitrogen 16.4 per cent, and sulphur 0.65 per cent. It is readily obtained by mascerating the shells in hydrochloric acid and boiling the residue in sodium hydroxide in which the conchiolin remains undissolved. Treated with hot mineral acids it goes into solution. Wetzel (1900) found that the substance gives Millon's reaction, and from the decomposi- tion products formed in boiling sulphuric acid he obtained glycocoll, leucin, and an abundance of tyrosine. He assigns this compound to a place between casein and egg albumin. The inorganic constituents of the shell consist chiefly of calcium carbonate with which are present small quantities of sulphates, oxides, or carbonates of magnesium, iron, manganese, and silica. The following analysis by Mr. Adrian Thomas will serve to show the various elements and compounds: Composition of the mussel shell. Calcium oxide Magnesium oxide Iron and manganese oxides Silica Carbonates Sulphates Organic matter Water Phosphates, chlorides, sulphides Trace. Traces of sodium and potassium which probably came from unremoved sea water were also detected. THE MANTLE. The mantle is a fold of integument which almost completely envelops the body. It is composed of the two lobes which lie symmetrically placed on the right and left sides of the body. They arise dorsally as an outgrowth of the body wall, cover the entire inner surface of the shell, and terminate in a free ventral border which is firmly attached to the edge of the shell by means of the pallial muscles. The free mantle edges unite anteriorly near the posterior edge of the anterior adductor muscle. At the posterior end of the shell they are joined together by a triangular-shaped band of deeply pigmented integument, the branchial membrane (fig. 115, BrM , opp. p. 138). The exhalent syphonal opening (fig. 115, Exs, and fig. 116, Exs) lies just dorsal to the branchial membrane and is surrounded by a tough ring of heavily pigmented tissue. The mantle edges separate to pass on either side of this opening and converge forward to the apex of the shell, where they unite and terminate. Between the syphonal opening and their point of termination they are joined together by a continuation of the branchial membrane. The space between the mantle lobes lying just below the exhalent corresponds to the inhalent syphon of many lamellibranchs. The structure of the mantle lobes in young animals is quite simple. In cross section they are thin and membranelike, consisting of an outer layer of simple epithelial cells, an Per cent. 51.21 27 32 11 37 33 I 02 8 °5 5S Bull. U. S. B. F., 1921-22. (Doc. 922.) PAd PRet L ARet An Fm. M Mes < I IhiR Ft LP 117 9 Mlli -OLP ILP PAd --BrM -Exs -BrM ' Ins FlG. 115. — Dorsal view of an adult mussel w ith valves open normally. Fig. i 16. — Dorsolateral view of an adult mussel from which shell has heen removed. FlG. 117. — Median view of a mussel with right valve, ri^ht mantle lobe, and gills rem