Hormonal Control of Sexual Maturity in Octopus

During the last few years an intensive programme of experiments designed to correlate structure with function within the brain of cephalopods has been carried out by two groups of workers. Boycott and Young have studied the effect of brain lesions upon visual learning (for references see Boycott & Young, 1955; Young, 1958) and the present authors (Wells & Wells, 1957–8) have studied the effect of lesions on tactile responses. In both series of experiments animals have been found with enlarged gonads, and a preliminary note in which three cases (two males, one female) are cited as occurring after optic tract section has been published by Boycott & Young (1956). The much more frequent occurrence of enlarged gonads in experiments on tactile responses, in which blinded animals were used, led us to begin a systematic investigation of the causes of this phenomenon. We started collecting records of the condition of the gonads in our experimental animals in August 1954 and by the end of August 1957 had accumulated data from 487 individual octopuses, including samples from every month of the year except January. The present account is an analysis of these data and shows that maturation of the gonad is determined by a secretion from the optic glands ; production of this secretion is regulated by a centre in the supraoesophageal brain mass and ultimately by light.

Octopuses of from 200 to 1000 g., probably aged from 6 to 18 months (see p. 6), were collected from the Bay of Naples. These animals were killed by cutting out the brain and weighed immediately after death, weights being recorded to the nearest 10 g. Gonads were dissected out together with their ducts and weighed to the nearest 01 g. The majority of these records were from animals treated in various ways in connexion with work on learning in Octopus, in particular with work on the animal’s tactile system. For the latter experiments animals were subjected to operations in which they were blinded by section of the optic nerves and/or had parts of their brains removed. Descriptions of the operational techniques involved are given in Wells & Wells (1956, 1957a). After such operations animals were used in training experiments of variable duration before being killed; brain lesions were then checked from serial sections. At least two variables must therefore be taken into account when considering the effect of central brain lesions and the effect of optic nerve section upon the state of the gonads—the nature of the lesion made, and the time elapsed between operation and killing the animal. The possibility of a third variable, seasonal variation in the state of the gonads, was also considered but although records were collected in every month of the year (except January) no obvious seasonal differences were found. It is possible that with more data small seasonal variations would prove detectable, but this is unimportant to the present analysis since enlargement of the ovary, upon which most of the experimental results are based, is unmistakable when it happens and has not been observed to occur spontaneously at any time of the year in animals within the size range (200–1000 g.) considered, although such animals will copulate (see page 4). Males with spermatophores in their testicular ducts have, on the other hand, been observed at all times of year.

The experimental animals were kept separated from one another, whenever possible in individual tanks. In some experiments animals were kept in large tanks divided into stalls by partitions. This was not entirely satisfactory as a means of separating the animals which were sometimes able to touch one another and even copulate (see Pl. 2 a) through gaps left for sea-water circulation around the edges of the partitions.

In the text, tables and figures individual animals are referred to by the number with which they were identified in our original protocols. The prefix to this number A, B,C or D gives the year in which the animal was used, being 1954–7 respectively. Other animals are identified by a three-letter reference ; these are octopuses used in visual training experiments by Boycott & Young, who kindly allowed us to collect data on gonad and body weights from their animals and also placed at our disposal serial sections of the brains of those that we wished to examine because they had enlarged gonads. The method of mapping brain lesions used in this account is explained on p. 8.

Because octopuses are expensive to buy in Naples, a comparatively small number, thirty-two males and thirty-five females, were killed without first being used in experiments involving surgical interference with the nervous system. To these true controls we can add a further fifty-two males and thirty-seven female animals that died or were killed within a few days of operation, having gonad weights that did not differ significantly from those of the true controls (see Table 1). In the case of males this means animals killed within 7 days of operation; in females, where the gonad enlargement as a result of operation is much more spectacular, significant differences in gonad weight are found as little as 4 or 5 days after operation, and the limit of the period during which it seems justifiable to include records from operated animals amongst the controls is reduced to 3 days. The gonad weights from the total of eighty-four male and seventy-two female controls are summarized in Text-fig. 1. The weight of the testis is more variable and nearly always greater than that of the ovary from animals of the same size; moreover the testis weight shows a marked increase relative to the body weight over the range considered while the ovary to body weight ratio increases very little and for all practical purposes may be regarded as being constant at 0·002–1/500th of the total body weight.

In female octopuses from the 200‒1000 g. size range considered in this account the ovary is always small. On opening the mantle cavity it may be seen as a white spherical body about 1 cm. across (octopus of 500 g.) lying at the posterior extremity. From it the paired oviducts pass out under part of the kidney and along the sides of the longitudinal septal muscle that partially divides the mantle cavity. About half way along these ducts are the oviducal glands, spherical swellings of the same opaque white colour as the oviducts themselves, and, in control animals, about twice the diameter of the duct. Text-fig. 2 c shows the size and position of these parts in a control animal. In none of the control females examined did we find ovaries containing eggs large enough to be visible to the naked eye and it is probable that in the normal way female O. vulgaris of less than 1000 g. do not breed.

The males, on the other hand, often have readily visible spermatophores in their gonads when as small as 300 g. and will copulate or attempt to copulate with female octopuses of similar size. A male octopus will also attempt to insert its hectocotylized third right arm into the mantle cavity of other males. Both males and females resist this intrusion but the resistance is by no means as vigorous as Pelseneer (1935) suggests—’La femelle se défendant des bras et du bec pendant toute la durée de I’acte qui gêne sa respiration’^*—and situations such as that shown in Pl. 2a (where the female could readily have moved away) are not uncommon in aquaria. As in the case of Sepia (Bott, 1938; Tinbergen, 1939), Sepiola (Racovitza, 18946) and Loligo (Drew, 1911, L. pealii, and our own observations on L. vulgaris) the female appears to play a purely passive role in mating behaviour.

Very little appears to be known about the relationship between age and size in Octopus. The animals grow very rapidly indeed in aquaria; Lo Bianco (1908) records an octopus of 100 g. from the Naples aquarium that grew to 410 g. in 6 weeks and our own animals of approximately 500 g. not uncommonly doubled their weight during 3 months of experiments in which they were fed four 10 cm. sardines daily, by no means their full capacity. It seems probable that in the sea the growth of octopuses is, in the normal course of events, limited mainly by the availability of food. Octopuses are solitary animals and one would expect to find a considerable scatter in the size of individuals from broods hatched at the same time. The only clue to the age of our experimental material is therefore the availability throughout the year of animals of the 400–500 g. size wanted for training experiments. In three successive years in Naples animals of this size were particularly difficult to obtain during the latter part of January and February, while individuals of up to 200 gr. or of over 1000 g. were relatively common; in October-December it was often hard to get animals of less than 700 g. These impressions are based upon deliveries to the laboratory and periodic visits to the local fishmarkets and cannot be quantified since we were deliberately selecting animals from a limited size range and kept no record of animals that we did not accept for use in experiments. The observations do nevertheless suggest that animals of 500 g. used in midsummer were in all probability spawned during the previous summer and if left in the sea would have attained a weight of 1000 g. in the following year.

In any batch of animals brought into the laboratory the females tended to be rather smaller than the males and to grow more slowly in the aquarium under exactly the same conditions. A similar state of affairs has been reported for Loligo vulgaris from the North Sea (Tinbergen & Verwey, 1945) where the females grow more slowly and reach sexual maturity later than the males; and among Sepia collected and exhibited in the Naples aquarium males were consistently larger than females (they were also more numerous, but this is because females are retained by the fishermen to catch more males (see Boycott, 1958)).

Out of 465 animals used in experiments on tactile discrimination, 251 were males and 214 females. We deliberately selected females during one short period of this work, but it is unlikely that this made as much as 5% difference to the totals. Assuming that there was no selection of males and females by the fishermen (unlikely, since the fisherman who supplied most of the material was unable to distinguish the sexes with any reliability, although he claimed to be able to do so), these figures indicate a sex ratio approaching 1:1, with, if anything, a preponderance of males. Crew (1927) quotes a sex ratio of 1 male to 3 females for Octopus at hatching as ‘having met with general approval*. Presumably these figures do not refer to O. vulgaris; in order to get such information it would be necessary to rear the animals until their sex organs differentiated, as Montalenti & Vitagliano (1946) have done with Sepia (where, incidentally, they found a sex ratio of 1:1). Larval O. vulgaris, unlike Sepia which hatches from a much larger egg, are minute and pelagic and so far as we are aware nobody has yet succeeded in keeping them for more than a very few days. It seems therefore, improbable that Crew’s figure refers to O. vulgaris and there is no need to suppose the very heavy selective mortality of females during the first year of growth that it would imply if correct.

The brain of Octopus consists of a central mass surrounding the oesophagus broadly divisible into sub- and supraoesophageal regions, each of which is in turn divisible into a number of clearly defined lobes. The whole structure is enclosed in a cartilaginous box except for the optic lobes which project from either side of the supraoesophageal mass. Text-fig. 3 shows the location and relative sizes of these parts. Within the supraoesophageal mass a number of lobes may readily be distinguished. For present purposes we are concerned directly only with the vertical lobe and the structures immediately underlying it, the subpedunculate and dorsal basal lobes.^* The former is a discrete bilateral structure having its own walls of cells and neuropil (Pl. 1 a, b) connected by tracts with the neuropil of the dorsal basal lobe. It is also the source of nerves running to the subpedunculate tissue in the orbits (Boycott & Young, 1956) and possibly of nerves to the optic glands (see p. 23) ; it is therefore of particular interest as being the only part of the supraoesophageal mass, other than the buccal lobe, giving rise to efferent nerves. Immediately below the subpedunculate lobes is a less well defined region, the dorsal basal lobe, characterized by islands of cells intermingled with neuropil and tracts. This lobe is separated into anterior and posterior regions by an irregular hanging curtain of cells, and its neuropil is continuous with that of the underlying median basal lobe. The basal lobes are higher motor control areas and lesions in this region produce deficiencies in posture and in the co-ordination of movement (Boycott & Young, 1950). The rest of the supraoesophageal lobes (with the exception of the buccal lobe which controls the mouthparts) appear to be concerned only with sensory integration and with learning; the effect of lesions to these lobes is not generally apparent until attempt is made to train the animals to make visual or tactile discriminations.

The only non-nervous structures that need be mentioned are two small spherical bodies located on the optic stalks—the optic glands (Text-fig. 3 a). The anatomy of these glands is discussed on p. 18.

Where the optic lobes were removed by section of the optic stalks distal to the optic glands no checking other than examination of the excised lobes was considered to be necessary.

All other brain lesions were checked from serial sections. These were prepared using a modification of Cajal’s silver method given in Sereni & Young (1932, method ‘B’). The maps given in Text-figs. 5, 7 and 11 show the extent of the part removed assessed as accurately as possible from these sections and plotted upon a pair of standard diagrams representing longitudinal median and transverse sections through the supraoesophageal lobes (see Text-fig. 3).

In Text-figs. 5, 7 and 11 the state of the optic glands at death is recorded for each operated animal alongside the transverse map of the lesion made. R and L indicate enlarged and r, 1 normal sized optic glands on the right and left sides respectively. Where the condition of the gland was not recorded this is indicated –; absence of a record generally indicates that the optic gland was normal in size.

It will be shown below that section of the optic nerves or removal of certain parts from the supraoesophageal lobes from the brain of Octopus causes enlargement of the gonad. This enlargement is essentially a ripening process and the increase in size of the ovary in female animals is proportionally much greater than that of the testis in males, the ovary becoming relatively enormous as it ripens prior to egglaying. Most of the experimental work has therefore been done with females, where enlargement of the ovary produces a quicker and more sensitive means of detecting changes arising from the experimental treatments than enlargement of the testis in males. Because of this, experiments with females will be dealt with first, followed by an account of the fewer similar experiments with males showing that the mechanism of control of maturation of the gonad is essentially the same in both sexes.

Text-fig. 1 b shows the size of the ovary in seventy-two control animals with body weights of from 200–1000 g. In no case was the gonad/body weight ratio greater than 0·0032. When a survey was made of 105 animals kept for more than 7 days after lesions had been made to the supraoesophageal lobes of the brain and/or after the optic nerves had been cut it was found that in sixty-nine of them the gonad/body weight ratio was larger, sometimes very much larger, than this maximum. The experimental animals fall into 3 categories:

1. Animals having brain lesions including removal of the vertical lobe and of certain of the tissues lying immediately underneath it.

2. Animals with the optic nerves cut or the optic lobes removed.

3. Animals having central lesions of various types as well as having the optic nerves cut or the optic lobes removed.

These results are summarized in Text-figs 4 and 6 from which it can be seen that where enlargement is produced at all it is commonly very considerable, the ovary increasing from about 1/500th to as much as i/5th of the total body weight within 6 weeks of operation.

It is convenient to deal with the results in these categories separately although, as will be shown later, the results in category 3 can be reclassed into categories 1 and 2 on a basis of the lesions made.

Twelve animals taken from a long series of experiments by Boycott & Young in which part or the whole of the vertical lobe was removed in connexion with visual training experiments. Octopuses treated in this way were generally found to have ovaries of quite normal size when killed 2–6 weeks after operation but in a small proportion of cases very large ovaries were found, and the present sample of twelve animals includes six such cases as well as six of the more typical animals in which the ovary was not enlarged (Text-fig. 4). Maps of the lesions found in this sample of twelve animals are given in Text-fig. 5. In every one of the octopuses with enlarged ovaries the lesion was found to extend beyond the limits of the vertical lobe and to include removal of the subpedunculate lobe plus a variable amount of damage to the roof of the dorsal basal lobe on at least one side, a type of lesion not found in any of the animals with normal sized ovaries.

Octopuses were blinded by bilateral section of the optic nerves (thirty-five animals) or by removal of the optic lobes after section of the optic tracts distal to the optic glands (three animals, making thirty-eight in all). Sixteen of these animals (fourteen optic nerve sections and two optic lobe removals) had ovaries greater than the maximum size found in controls (Text-fig. 4). Of the remaining twenty-two, most of which were kept for comparatively short periods after operation, only five had ovaries smaller than the mean control size, and it seems reasonable to suppose that the results with these five were due to incomplete operations. The maximum rate of enlargement of the ovary found in blinded animals is considerably lower than the maximum found in category 1, and there is a rather greater scatter in the individual rates of enlargement. At its most rapid, enlargement of the ovary following optic nerve section led to a 20-fold increase in the weight of the ovary within 5 weeks of operation; at the other end of the scale animals operated for just as long had ovaries only slightly larger than the maximum size found in controls. It is not clear why there should be this considerable variation and although reasons are given below for believing that the effect of blinding on any individual animal is ‘all-or-nothing’ in so far as ovarian enlargement is concerned, it remains possible that the scatter of rates is a reflexion of the degree of completeness of the operations. It is unfortunately almost impossible to check the completeness of optic nerve section; tests in which no visual reactions could be elicited proved on post mortem dissection to be unreliable, and dissection itself is an uncertain check because of the ease with which bundles of optic nerves remaining are overlooked or destroyed in searching amongst the flocculent ‘white body’ (Cazal & Bogoraze, 1943) that fills the orbit. This means that we cannot be absolutely certain that all the optic nerves were cut in any individual case and have no direct means of separating those animals in which the operations were complete from those in which the operations were very nearly complete. There is, however, an indirect method based on changes to the optic glands, which will be discussed later (p. 24); it is shown that blinding need only be complete on one side to be effective in causing gonad enlargement and this suggests that in the few cases where there was no increase in the size of the ovaries the optic nerves remained intact on both sides.

This category includes all the rest of the experimental animals. Forty-seven out of a total of fifty-five had enlarged ovaries (Text-fig. 6). In twenty-one of these the ovary had enlarged at a rate comparable with that found in category 1 octopuses. This is faster than the maximum found in category 2 (blinded) animals and it seems fair to assume a priori that in these twenty-one cases enlargement was due to a central lesion and not to blinding alone. One would expect to find that these animals had brain lesions affecting the subpedunculate/dorsal basal area (found to be critical in category 1 experiments), and Text-figs. 7a, b and 11 show that this was indeed the case. There is one exception, octopus HBI, which has both optic tracts cut centrally; this operation is quite different from that made to remove the optic lobes in category 2 experiments since it cuts off the optic glands (Text-fig. 3 a) from their nerve supply in the central part of the brain, a supply that is left intact by section peripheral to the optic glands. It will be shown below that this nerve supply originates in the area already found to be critically related to ovarian development in category 1 animals, so that section of the optic tracts as carried out in the case of HBI may properly be compared with lesions of the sort shown in Text-fig. 5 a in so far as any effect upon the gonads is concerned.

Since the most rapid rate of ovary development found in the forty-seven category 3 animals was not significantly greater than the maximum found in the six category 1 octopuses we can infer that the effect of blinding does not add to the effect of central lesions on gonad size.

The remaining thirty-four animals in category 3 had ovaries that were no larger than might be expected from blinding alone. In order to assess these results we must take into account the operational routine to which each individual was subjected and the nature of the lesions made to its brain. We already have a good idea of which central lesions cause gonad enlargement from the results in category 1 and from the animals showing particularly rapid enlargement in category 3, and we can subdivide these thirty-four animals into two groups on a basis of whether or not their lesions were of a type associated with gonad enlargement elsewhere. When this is done we find that out of the twenty-six animals with ovaries greater than the control maximum, twelve had lesions of the type associated with enlargement of the ovary elsewhere (Text-fig. 7 c, d). The rates of enlargement found in this group of twelve animals were all towards the upper limit of the range found in category 2 (Text-fig. 6) and four of them were subjected to two successive operations, being blinded by optic nerve section up to 3 weeks before central lesions were made. In these four the final rate of gonad enlargement is a compound of two rates, and includes a considerable period of relatively slow increase after the first operation followed by a much more rapid increase after the second. When this is taken into account it is clear that the final gonad weight indicates a rate of enlargement following the second, central operation, fully comparable with that found in category 1 animals. The rates of increase in the remaining eight animals from this group fell well within the upper third of the range found in blind animals, and although in these cases the cause of enlargement must, strictly speaking, remain uncertain, the nature of the lesions made and the rates of enlargement found together indicate ovarian increase brought about by destruction in the usual subpedunculate/dorsal basal region.

This leaves fourteen animals out of the total of forty-seven with enlarged ovaries that had central lesions of types not found to be associated with enlargement of the ovary elsewhere. Nine of these animals had central lesions (not plotted) affecting the frontal inferior and subfrontal lobes only, one had the frontal superior removed, and the remaining four lesions confined to the vertical and superior frontal lobes (Text-fig. 7 e). The rates of ovarian enlargement found in these animals all fell well within the range found in category 2 experiments as a result of optic nerve section alone, and it seems reasonable to suppose that these animals had big ovaries because they had been blinded rather than because of their central lesions. In so far as any effects of operation on ovarian development are concerned the animals in this group may properly be regarded as category 2 experiments, and their ‘time since operation ‘measured from the date on which they were blinded.

In Text-fig. 8 the category 3 experiments are reclassified, according to whether ovarian enlargement was induced (or probably induced) by central brain lesions or by blinding alone and these reclassified results are plotted together with those already given in Text-fig. 4 (results from categories 1 and 2) to define more accurately the relationship between the rates of enlargement produced by the two alternative treatments.

Eight octopuses out of the fifty-five in category 3 were kept for more than 7 days after operation without developing ovaries larger than the maximum size found in controls. Only two of these eight were kept for more than 10 days. One of these animals (C196NVB) had an ovary/body weight ratio of 0·0032—equal to the maximum size found in controls. The other (C173NVB) was kept for thirty-two days after optic nerve section; since its gonad/body weight ratio was only 0·0015 at the end of this period it must be supposed that in this case blinding was incomplete. Maps of the lesions in these animals are included in Text-fig. 7e—in neither case did the parts removed include the subpedunculate or dorsal basal lobes.

The weight of the ovary in animals that are killed within 3 days of operation does not differ significantly from that in unoperated animals. This ceases to be true after about 4 days, by which time the gonad has already begun to show signs of enlargement as a result of the treatments described above. In Text-fig. 9 the average weight of the ovary in seventy-two controls is compared with that in fifty-four animals killed 4‒7 days after operation. Comparatively few of the latter were used in training experiments and the brains of most of them have not been sectioned to check the lesions made. No attempt therefore has been made to break down this data into categories and it is likely that at least some individuals had operations that would not have produced enlargement of the ovary. Nevertheless there is a significant difference between the mean weights in this group and in controls, indicating that enlargement of the ovary begins almost immediately after operation. At this stage there is no difference in the ovaries or in their ducts visible on dissection, although the optic glands (see p. 21) may already be noticeably enlarged.

Fifteen days after suitable central brain lesions have been made, and about twice as long after optic nerve section, the ovaries of experimental animals are noticeably enlarged on postmortem dissection. Enlargement continues until the bodies of operated animals become visibly distorted by the ovary within (Text-fig. 2b). Five weeks after an operation involving removal of the subpendunculate lobe on at least one side the ovary may constitute as much as i/5th of the total body weight. On dissection of such animals it is found that the swollen ovary occupies the whole of the hind part of the body (Text-fig. 2d) considerably displacing the other organs which appear by contrast unusually small (weights taken from about thirty animals show, however, that the kidneys, ctenidia and hearts at least remain quite normal in size). The ovary itself, now pale yellow in colour, is crammed with small (1x3 mm.) eggs, all of the same size which is approximately that of the eggs when laid. As the ovary increases in size so do its ducts. These become swollen to several times their normal diameter, and the oviducal glands increase greatly in circumference, becoming almost disk-like, ribbed and brown in colour at the centre (Text-fig. 2d).

Three octopuses were kept until they laid eggs after operation. Particulars of these three animals are given in Table 2, and details of the brain lesions concerned included in Text-fig. 11. In two out of these three cases (C27 and C28) the eggs were examined at intervals and those laid by C 28 were found to be fertile, developing until the eyes of the larvae were clearly visible, at which stage a failure in the circulation system unfortunately killed both eggs and parent. We have no record of the sex of the octopuses kept in the tanks adjoining those of C27 and C28, so we do not know when the eggs from C28 were fertilized or whether the infertility of C27 was a result of isolation from male octopuses.

The behaviour of these three animals was very similar. About a week before egg-laying they began to feed irregularly, sometimes rejecting food (sardines and crabs) that they had hitherto accepted eagerly. As the time of oviposition approached the animals showed a tendency to remain in one particular well-aerated place in their tanks, in the angle between the side of the tank and the water surface. Breathing was deeper and more rapid than usual. Eggs were laid, in bunches attached to the side of the tank, at intervals over a period of about a week ; during this period and subsequently the animal did not leave the eggs except very occasionally and then only for a few seconds at a time (generally to reject food presented to it). Although in the normal way octopuses will at first attempt to eat any unfamiliar object presented to them (Wells & Wells, 1956), objects presented to brooding animals were invariably rejected without first being passed to the mouth. In many cases objects such as crabs and pieces of sardine were carried to the opposite extremity of the tank before being dumped and blown away with vigorous jets from the funnel. Brooding animals were never observed to feed, but evidently did so occasionally since sardines left in their tanks were sometimes eaten overnight. These observations agree closely with accounts of egg-laying and brooding by normal O. vulgaris (Lo Bianco, 1908; Monticelli, 1921 ; Heldt, 1948; Vevers, 1959) and with Batham’s (1957) detailed account of O. maorum which appears to behave like O. vulgaris. Taken together with the fact that these operated animals can evidently lay viable eggs, these behavioural similarities strongly support the view that the experimental treatments discussed in this account cause the gonad to mature in a manner essentially normal although precocious.

The optic glands are small (1 mm.) pale yellow bodies lying on the optic stalks (Text-fig. 3 a). These bodies have in the past been variously identified as glands or as nervous tissue, the spherical glands which are readily visible on dissection being confused with the olfactory and peduncle ganglia seen in sections (see review by Boycott & Young, 1956). Enlargement of the ovary following optic nerve section or central brain lesions is always accompanied by changes to the optic glands, which become bright orange in colour and may increase to more than ten times their normal size.

Most of the optic glands studied were fixed in neutral formalin in seawater and all measurements of glandular and nuclear volumes quoted are taken from such material embedded in paraffin and sectioned at 8μ. A small number of normal glands were fixed in Susa, Bouin and Flemming; stains used were Heidenhein’s iron haematoxylin, Azan, the Feulgen stain for DNA and Unna’s methyl green-pyronin, the last two being used only on formalin-fixed material. A modification of Cajal’s silver method (modification B) given in Sereni & Young (1932) was used to show the nerve supply.

The cells of the optic gland are arranged in solid masses and do not form vesicles; there are two types, chief cells and smaller supporting cells (Text-fig. 10a) (Boycott & Young, 1956). The chief cells have large nuclei, prominent nucleoli in finely granular nucleoplasm, and cytoplasm of irregular shape. With Azan, nucleoli and nuclear granules stain dark blue, the cytoplasm a paler cloudy mauve. The supporting cells have smaller, more regularly oval nuclei without prominent nucleoli; the nuclear granules which are larger than those of the chief cells stain bright red with Azan, while the rather scanty fibrous-looking cytoplasm stains a clear pale blue. The nuclei of the supporting cells, particularly the nuclear granules, stain intensely with Schiff’s reagent following 15 min. N/HC1 digestion at 60° C., the colour being a bright reddish purple, in contrast to undigested controls which remain unstained. In such preparations the nucleoli of the chief cell nuclei stain very faint pink, while the rest of the nucleus remains colourless. With methyl green-pyronin the nuclei of the supporting cells stain intensely, while the cytoplasm remains relatively pale, reactions that are not affected by ribonuclease digestion. Both nuclei and cytoplasm of the chief cells stain intensely with this stain and, in contrast to the results with the supporting cells, the uptake to stain is almost entirely eliminated by digestion with ribonuclease for 3 hr. at 60° C., only the nucleoli staining faintly thereafter.

The results with Schiff’s reagent and methyl green-pyronin together indicate a relatively low concentration of DNA in the nuclei of the chief cells (where it appears to be limited to the nucleoli) and a much higher concentration of RNA in both nucleus and cytoplasm than in the supporting cells. It is likely, since high RNA content is characteristically associated with protein synthesis (see Brachet, 1957), that the chief cells secrete protein and that the supporting cells do not

Measurements were made from the glands of six control animals weighing between 390 and 1140 g. and it was found that the ratio of glandular volume to body weight remains approximately constant over this range. It was also found that the mean volume of the chief cell nuclei rises with increase in body size so that the ratio of individual nuclear to glandular volume remains more or less constant (Table 3); this implies that the actual number of chief cell nuclei does not increase as the animal grows, a finding that is consistent with our failure to discover mitotic figures among the chief cell nuclei in any of the glands examined.

The nerve supply of the optic glands, which can be shown by degeneration experiments to originate in the subpedunculate/dorsal basal lobe area (p. 24), enters the gland by way of the olfactory lobe and immediately breaks up into a large number of fibres ramifying in the gland mainly in the fibrous tissue. An abundant blood supply to the gland can be demonstrated by injecting methylene blue into the dorsal aorta; the optic glands become more intensely blue than any of the surrounding tissues (Boycott & Young, 1956).

When the optic glands enlarge there appear to be no changes in the size, number or staining properties of the supporting cells and it is concluded that these play no part in secretion, their function being limited to the provision of a connective tissue framework, as in any case seems likely from their staining properties in the normal gland.

The individual chief cells, on the other hand, increase very considerably in volume as the gland enlarges (see below); their nuclei, nucleoli and nuclear granules all enlarge while the cytoplasm increases in volume and becomes vacuolated. These changes are progressive and occur both after blinding by optic nerve section and after central lesions of the type already described. Changes after blinding are relatively slow (Table 3) but appear to be qualitatively identical with those brought about by central lesions.

In no instance were mitotic figures observed in enlarged glands and, as there was no accumulation of material in the spaces between the cells, increases in the size of the glands must be attributed to increase in the size of their individual chief cells. As Table 3 shows, this increase may be very considerable since the volume of an optic gland can rise to more than ten times its normal value within 5 weeks of operation. The bulk of this increase is evidently cytoplasmic since the rise in volume of the individual chief cell nuclei does not keep pace with the rise in volume of the gland as a whole (Table 3), an observation that is in keeping with histological evidence indicating accumulation of material in vacuoles in the cytoplasm of the enlarged chief cells (Text-fig. 10b, c, d). It would appear that under these conditions the gland cells secrete material faster than it can be carried away in the bloodstream, and that the disproportionate increase of cytoplasm is due in the main to accumulation of secretory product.

Where a considerable time has elapsed since operation (as in the case of C51, see Table 3) occasional nuclei can be found in which the nucleolus has more or less broken down and the nuclear granules, further increased in number and size, have become arranged in a dense ring round the periphery of the nucleus. In extreme cases the nucleus appears to be surrounded by a vacuole (Text-fig. 10d). Such nuclei are always relatively small and may represent a final stage in the cycle of changes beginning with enlargement of the nucleus, nucleoli and nuclear granules, and ending with collapse of the nucleus and disintegration of the nucleoli.

When the gonad is enlarged it is invariably found that one or both of the optic glands is enlarged as well. Lesions that cause the optic glands to enlarge invariably also cause enlargement of the gonad. This could be coincidental, although it is difficult to imagine why cuts severing the nerves innervating the glands should lead to enlargement of the sex organs unless the optic glands are in some way responsible. To establish the causal relation between secretion by the optic glands and maturation of the sex organs a further series of experiments was carried out in which the optic glands were removed before treatments known to cause gonad enlargement.

The optic glands were removed from five female octopuses in operations that included removal of the optic lobes by cuts peripheral to the site of the optic glands and removal of the greater part of the central supraoesophageal mass, leaving only parts of the frontal inferior, subfrontal and buccal lobes.^* In a further two females the optic lobes alone were removed. These were all treatments that in the normal way would certainly have produced a noticeable enlargement of the gonad within two weeks, yet in no case was the gonad weight at death outside the control range even though some of the animals had been kept for as long as 8 weeks after operation (see Table 4). Four out of the seven females actually had ovaries smaller than the mean size found in controls.

Removal of the optic glands does not lead to any changes in the behaviour or condition of the experimental animals that we have been able to observe, a finding that agrees with Sereni’s (1930) and Callan’s (unpublished, quoted Boycott & Young, 1956) observations, nor does it lead to a significant regression in the size of the gonad, at any rate in female octopuses. In Table 4 are listed the ovary/body weight ratios from 11 animals kept for varying periods after removal of the optic glands ; the mean ovary/body weight ratio from these animals is almost exactly the same as in controls (0·0022 against 0·0020).

The situation is less clear in the case of males, where only five experiments were made. In four of these, kept for a period of from 4 to 7 weeks after operation, the testis was markedly smaller than the mean for controls of similar size ; the remaining animal, with a testis larger than the mean control size, was kept for only 8 days after operation.

The findings outlined above would appear to be exactly what one might expect from animals in the size range (200-1000 g.) studied. Optic gland removal produces no regression of the gonad in females because the ovary is in any case undeveloped. In males where the gonad is normally ripe it seems that some decrease in size is caused, although because of the small number of experiments it is impossible to be certain of this. This implies that the optic glands secrete to some extent all the time (an implication that is borne out by the histological data (p. 19)) and that a lower level of secretion is required to ripen the testis than the ovary.

As has been pointed out above, rapid enlargement of the gonad over and above the rate found in blinded animals is always associated with lesions including the subpedunculate lobe and a variable amount of damage to the dorsal basal lobe on at least one side, a type of lesion that is also invariably accompanied by enlargement of the optic glands (or gland if the lesion is unilateral) (Text-figs. 7 and 11).

From the present series of experiments we cannot be sure whether it is removal of the subpedunculate or damage to the dorsal basal lobe that is responsible for enlargement of gland and gonad. The subpedunculate lobe is a well-defined tissue and it is possible to be reasonably certain when it is absent and when parts remain. Unfortunately its removal inevitably damages also the dorsal basal lobe and it is much more difficult to assess whether or not parts of this are damaged or missing. The dorsal basal lobe has a thick roof with an ill-defined upper margin that wraps round the forward-reaching lateral extensions of the subpedunculate lobe above, and breaks up into islands of cells mixed with tracts below (Pl. 1 b). Since lesions are always followed by a certain amount of distortion of the remaining structures it is exceedingly difficult to map the area of tissue destroyed in this region at all accurately. This is particularly true in animals kept for a considerable period after operation, as were those used for the present series of experiments, and it means that although some damage to the dorsal basal lobe seems always to be found in animals having enlarged ovaries, the exact extent of this damage and whether it is itself the cause of ovarian enlargement must remain at present uncertain.

Lesions of the type associated with gonad enlargement cause degeneration of the nerves ramifying among the cells of the optic glands, and tracts, originating in the subpedunculate/dorsal basal area, can be traced along either side of the brain until they run out along the optic stalks to the optic glands. Section of the tract on either side is followed by the appearance of degeneration granules in the optic gland on that side (and only on that side) within 24-48 hr. (Pl. 26 and Text-fig. 11) and later by enlargement of the chief cells in the gland. Since removal of the optic gland is not followed by gonad enlargement the effect of severing this nerve supply cannot be attributed to the failure to release secretion by the glands. The nerve supply must therefore be inhibitory, limiting secretion in the intact animal.

The optic glands also enlarge when the optic nerves are cut or the optic lobes removed by cuts peripheral to the optic glands. The histological changes to the glands, though rather slower, appear to be identical with those following removal of the tissues overlying the dorsal basal lobe (Text-fig. 10) although neither method of blinding interferes directly with the innervation of the optic glands. As with central lesions, the effect of blinding can be unilateral, the optic gland enlarging only on the side where the optic nerves are cut. In no instance did careful dissection reveal optic nerves remaining on a side where the optic gland enlarged, while several cases in which the optic gland remained normal in size and colour on one side were found to be attributable to incomplete section of the optic nerves on that side. Since unilateral section of the optic nerves leaves the unoperated side normal in size and colour the effect of blinding on the glands must be mediated through the nervous system and not by means of any intermediate hormonal mechanism.

It seems likely that this nervous control involves reduction of the inhibitory effect of the central subpedunculate/dorsal basal area rather than excitation of the glands via innervation from another source. The optic glands are placed on the optic stalks so that innervation can come only from the optic lobes or the central brain mass. Section of the optic stalks centrally to the optic gland causes very rapid enlargement of gland and ovary (animal HBI, p. 12) and therefore presumably does not cut off an excitatory innervation as well as the inhibitory supply from the subpedunculate/dorsal basal lobe area. Removal of the optic lobes by cuts peripheral to the optic glands produces enlargement of the gonad comparable to that produced by section of the optic nerves (p. 10) and excludes the possibility of excitatory innervation from the other direction.

Section of the optic nerves or removal of the dorsal basal lobe from the brain of male octopuses causes the optic glands and testes to enlarge. The changes in the optic glands are as in females but the increase in size of the gonad is relatively much smaller, presumably because the testes are already ripe at the start of the experiment in all but the smallest of the animals used. The testis in any case never attains a size comparable with that of the ripe ovary and even in the largest animals with ripe spermatophores forms only about 1/100th of the total body weight (Table 5). Because gonad enlargement is not as spectacular as in females it is more difficult to recognize when it has occurred and this, together with the considerable range of testis size in controls (Text-fig. 1 a) makes it impossible in most individual cases to be sure whether or not the testis is larger than usual. It is found, however, that the average size of the testis in males subjected to treatments known to cause enlargement of the ovary in females is significantly greater than that in controls. In Text-fig. 12 a comparison is made between the testis weight from 84 controls and 100 experimental animals blinded by optic nerve section or subjected to operations including removal of the central brain mass as well as optic nerve section. The two experimental groups respectively correspond to categories 2 and 3 from the experiments with females. The difference between experimentáis and controls is clearly significant. There seems, on the other hand, to be no difference in the testis weights from animals in categories 2 and 3, and on this showing it could be argued that central lesions found to cause enlargement in females are without effect upon males. Enlarged optic glands occur, however, in unblinded males after central operations of the type associated with gonad and optic gland enlargement in females (see, for example, octopuses JPH and JRA in Text-fig. 11) and unless we suppose that the optic glands have a quite different function in males and females (which seems improbable) this must mean that central lesions can produce enlargement of the testis. The situation in the two sexes is therefore essentially similar, the state of the gonad being controlled by a central region in the brain acting on the optic glands.

So far in this account only results from animals weighing between 200 and 1000 g. have been considered. This range was determined by the training experiments for which most of the animals were used, animals below 200 g. being inconveniently small for brain operations, and above 1000 g. being rather too large for the available tanks. Nevertheless a number of experimental animals, weighing less than 1 kg. when brought into the laboratory, grew to weights of between 1 and 2 kg. in the course of the experiments. Most of these animals were males. Table 5 is a list of these animals, together with a few records of gonad weights from large octopuses that died in the Naples public aquarium, including one record from a female of 2 kg. (octopus ‘Z’) having an ovary weighing 45·0 g. There were visible, though not full sized, eggs in this ovary—the only record we have of a female in which the ovary appeared to be ripening without operational assistance. There are thirty-five records in all, twenty-three from males and twelve from females, most of them animals with central lesions as well as having the optic nerves cut (category 3 experiments); a few had only the optic nerves cut or the optic lobes removed (category 2). These records have not been included in the main body of results because of the scarcity of controls above 1000 g. ; they do, however, clearly agree in all important respects with those from the 200–1000 g. range, confirming that lesions including the subpedunculate lobe cause enlargement of the optic glands and gonad and that the rate of this enlargement is greater than the rate of enlargement following blinding alone.

Evidence has been presented to show that maturation of the gonad in Octopus is controlled by a secretion produced in the optic glands and that these glands are in turn controlled by an inhibitory nerve supply. Interference with this nerve supply in various ways (Text-fig. 13) causes the optic glands (or gland, since all treatments can be carried out unilaterally, affecting only the gland on the operated side) to become swollen with secretion and this is followed by enlargement of the gonad; females may lay viable eggs which they brood in an entirely normal manner. Blinding by optic nerve section or optic lobe removal, while not interfering directly with the innervation of the optic glands (Text-fig. 13), also induces enlargement of the glands and gonads, and again the treatment can be shown to be effective if carried out unilaterally. None of these means of inducing gonad enlargement is effective if the optic glands have first been removed.

The optic glands are innervated by nerves arising in the subpedunculate/dorsal basal region on either side of the posterior part of the supraoesophageal lobes of the brain. This innervation inhibits secretion and does not appear to be paralleled (as might be expected, see below) by a balancing excitatory innervation, so that blinding must have its effect via the supraoesophageal centres by reducing the restraint on secretion.

Although there is at present no direct experimental evidence available, the effect of blinding upon the state of the gonad implies that the optic gland system will untimately prove to be governed by changes in photoperiod, in a manner analogous to the regulation of sexual maturity by the pituitary system in vertebrates (Harris, 1955 ; for a more general review of the effects of changes in photoperiod on animals see Hendricks, 1956).

Optic glands have been found in all cephalopods (except Nautilus^*) so far examined, so that there is reason to believe that the mechanism here described from Octopus will prove to be general for the group. It has obvious analogies with hormonal systems controlling maturation of the gonads in other animals. The vertebrate pituitary, the insect corpus allatum, and the X-organ—sinus gland complex in the eyestalks of crustaceans all have a similar anatomical relation to the nervous system (Hanström, 1947), and while this anatomical comparison cannot be carried very far because the relevant structures in the higher centres of vertebrates, arthropods and cephalopods have evolved independently within the groups concerned and cannot be homologized, there are close functional parallels between the neuro-glandular systems regulating sexual maturity in the three groups.

All the glands concerned are regulated directly by the highest centres of the nervous system and all serve in one way or another to delay the onset of sexual maturity. The anterior pituitary of vertebrates produces a gonadotrophic hormone under control of a centre in the hypothalamus. The ultimate link in this control appears to be hormonal via the portal blood system of the pituitary stalk; if this is interrupted the gonads atrophy, as they do following lesions to the hypothalamus itself. This however is evidently not the whole story since limited lesions to certain regions of the hypothalamus, produced by tumours, may lead (at any rate in man) to precocious production of hormone and an early onset of sexual maturity, presumably by ‘damaging structures that normally inhibit the release of pituitary secretion’ (Harris, 1955). The situation in insects appears essentially similar, once metamorphosis has taken place. The corpus allatum, which in larval insects produces a ‘juvenile hormone’ (Wigglesworth, 1940) inhibiting production of adult characters, becomes inactive shortly before the last moult and later emerges as a source of a gonadotrophic hormone under control of an inhibitory centre in the supraoesophageal ganglion and an opposing excitatory centre in the suboesophageal ganglion (Englemann, 1957).

It is less easy to compare the above three systems with that found in the decapod crustaceans; operations so far carried out in the latter have tended to produce widespread metabolic effects making it difficult to ascertain whether induced changes to the gonads are incidental to, or the direct result of, specific interference with gonad-regulating hormones. It does, however, seem clear that some sort of double control exists; removal of the eyestalks causes the gonads to ripen and it appears that this can be attributed to removal of an inhibitory factor produced in the X-organ-sinus gland system (Knowles & Carlisle, 1956), while removal of the Y-organ from the antennary or maxillary segments^* is followed by atrophy of the gonad (Arvy, Echalier & Gabe, 1956). This would imply that the state of the gonad in crustaceans is regulated by a balance between two hormonal products, rather than by changes in the concentration of a single product as would appear to be the case in vertebrates, insects and cephalopods.

Notwithstanding these evident differences the crustacean system clearly falls into the same general category as the others in as much as it too is closely linked with the highest centres of the central nervous system. It seems improbable that this association has occurred independently in the four groups (three of them widely separated) by chance. A possible explanation can be derived from the observation that all these forms in which these control systems have been found are relatively advanced members of the groups to which they belong, being characterized by final forms having a considerable complexity of structure. This is particularly true of the structure of their nervous systems. Vertebrates and cephalopods are remarkable for the extent to which the higher parts of the C.N.s. are developed and both depend to a considerable extent upon learned rather than innate responses ; it seems reasonable to suppose that the flexibility of behaviour that this allows is largely responsible for the success of the two groups. Systems that involve learning from experience, however, imply a relatively long post-embryonic development if the animal is to gain advantage from a full development of its potentialities (Wells, 1958a), and this necessarily involves a delayed sexual maturity, particularly if, as in cephalopods, the animal starts its free life with those parts of its nervous system most intimately concerned with learning relatively undeveloped (Wells, 1958b).

The situation in arthropods, though similar, is complicated by their mode of development. In the decapod crustaceans and insects, where regulating devices have been found these are always associated with development by stages specialized for distribution or feeding; delayed sexual maturity is here desirable because it permits greater, and to a considerable extent independent specialization of both larva and adult (for a discussion of some of these specializations see Hardy, 1956). Once again, however, the final adult form is characterised by a considerable nervous complexity that gives it properties not found in the developmental stages; properties that again would be of no selective advantage to an animal breeding before they became fully developed.

We have now, therefore, four groups in which systems preventing early development of the gonad are closely associated with the presence of final stages having relatively complex structure and behaviour. Comparable systems have not been reported from the less advanced members of the same groups and, while it would be unwise to assert on this basis that such systems are lacking, it is tempting to speculate that this is because special mechanisms for delaying the onset of sexual maturity only become important where the gonad must be held back until the higher centres of the C.N.S. develop their full potentialities. If this is true it is not altogether surprising to find that the control of the glands responsible comes from these higher centres themselves, and the extraordinary parallelism of the mechanisms for controlling sexual maturity in three such widely separated phyla as the molluscs, arthropods and chordates becomes comprehensible and, indeed, inherently the most probable result of a parallel evolution of complex from more simple forms.

We would like to thank Mr B. B. Boycott and Prof. J. Z. Young, F.R.S., for permission to use data collected from their experimental animals and for placing at our disposal sections of the brains of these animals ; the photographs for Pl. 1 a, b, and 2b were taken by Mr J. Armstrong from their material. Prof. Young and Prof. V. B. Wigglesworth, F.R.S., have very kindly read and criticized this work in manuscript. The authors are very grateful to the Director and staff of the Stazione Zoologica di Napoli for their hospitality during this work, which was carried out during the period 1955-6, while M. J. W. was holding an EH Lilly Fellowship at the Stazione Zoologica in Naples, and during a visit to the Stazione in the summer of 1957, while holding a Fellowship at Trinity College, Cambridge.

Plate 1

(a) Longitudinal vertical section through the supraoesophageal lobes of the brain of Octopus, (b) Transverse section at the level indicated on a. The lobes of the brain are indicated as follows: b, buccal; i.f., inferior frontal; s.f., superior frontal; sub.f., subfrontal; v., vertical; s.v., subvertical; s.p., subpedunculate; d.b.a. and d.b.p., dorsal basal anterior and posterior respectively, o. optic commissure. Cajal silver preparations.

Plate 2

(a) Copulation in Octopus vulgaris Lamarck. Male on the left, with his hectocotylized third right arm in the mantle cavity of the female on the right. (b) Degenerating nerves in the LHS optic gland of octopus JQC 48 hr. after a central lesion (Text-fig. 11) was made. Cajal silver preparation.