A Statistical Study of Regeneration in Two Species of Crustacea by W

349

A STATISTICAL STUDY OF REGENERATION IN TWO SPECIES OF CRUSTACEA BY W. E. AGAR, Professor of Zoology, University of Melbourne.

(Received 22nd April, 1930.)

(With Three Text-figures.)

CONTENTS. PAGE I. Introduction 349 II. Methods of culture 350 III. Growth and sex 351 IV. The material available . .351 V. Structure of the antenna 352 VI. The operation . • . . . . 353 VII. General nature of the regeneration 353 VIII. Measuring and recording the amount of regeneration .... 355 IX. The influence of certain factors on regeneration ..... 357 (a) The segment through which amputation is performed . -357 (6) The level of amputation within the segment .... 357 W Age 358 (d) The simultaneous regeneration of the other antenna . . 358 (e) The previous regeneration of the other antenna . . -359 (/) Food 360 (g) General internal conditions of the animal 360 (h) General external conditions 361 X. The nature of the intrinsic factors controlling regeneration as disclosed by statistical analysis 362 XI. Discussion of results ...... 365 XII. Summary 367

I. INTRODUCTION. THIS study of regeneration in the two Cladoceran species, Simocephalus gibbosus and Daphnia carinata, grew out of a Lamarckian experiment on the possibility of improving (or otherwise altering) the power of regeneration in these animals by making them regenerate the antenna in many successive generations. Up to the present, these experiments have been completely negative in this respect, though they have furnished much data concerning the process of regeneration. The power of regeneration in a line of Simocephalus in which the antenna has been amputated a^»egenerated for seventy-four successive generations, and in a line of Daphnia for eighty generations, shows no difference from the power of regeneration of the JEB'VIliv 23 35° W. E. AGAR controls which had no ancestral experience of regeneration. This experiment in progress and will be reported in detail later. In the course of the experiment it was found necessary to analyse the factors influencing regeneration, and besides the main experiment, many subsidiary experiments directed to the elucidation of special points have been carried out. Altogether, some thousand regenerated antennae have been drawn and measured in Simocephalus, and about 950 in Daphnia, while about a thousand more have been examined and the number of setae recorded without measurement of length. One striking fact about regenerated parts in all animals is the great variation exhibited by them compared with the corresponding parts produced by embryonic development. The large mass of material which forms the basis of this investigation enables us to study this variation more closely than has been possible in previous studies of regeneration. A further great advantage possessed by this material is its genetic uniformity. All the specimens described in this paper are descended by parthenogenesis from a single female of Simocephalus or Daphnia respectively. Many experiments by myself, by Banta and others have shown that, except for occasional mutations, genetic differences do not normally arise between members of a clone. Thus whatever may be the causes of the great variation in degree of regeneration which is exhibited in this material, we can exclude genetic diversity as one of them.

II. METHODS OF CULTURE. The standard method adopted throughout these experiments (and adhered to wherever the contrary is not stated) was to rear each animal separately in a cylin- drical glass phial containing 50 c.c. of culture medium. This amount was measured wherever similarity of conditions was essential (as in the "test experiments"), but judged by the eye when this was not relevant. The culture medium used throughout the greater part of the experiment was prepared as follows. A rich culture of green Protophyta (largely flagellates) was grown in a medium prepared by boiling 3 gm. of dried fowl dung in a litre of water. The filtered fluid obtained thus, exposed to sunlight in glass jars, provides a very rich culture of Protophyta, which lasts for several months. Every few days a stock culture medium was made by adding about 100 c.c. of the thick protophyte culture (pipetted from the bottom of the jar) to 2500 c.c. of pond water, previously strained through bolting silk. In some of the experiments, an infusion of horse dung, as recommended by Banta, was added to the culture, but it was found that the protophyte culture was sufficient by itself, and less liable to fluctuation in quality. In this culture medium the animals thrive exceedingly well, and produce abundant eggs. The phials were placed in a water bath maintained at 22-5° C. by a toluol- mercury gas regulator. Statistical study of regeneration in two species of Crustacea 351

III. GROWTH AND SEX. After casting its embryonic cuticle, which happens immediately after birth, the animal moults four times (Simocephalus) or five times (Daphnia) before becoming mature—that is to say, laying its first batch of eggs. A very small percentage of animals of both genera become mature after one moult fewer than the normal. The period between the casting of the embryonic cuticle and the first ecdysis is the first instar. In the fourth (Simocephalus) or fifth (Daphnta) instar the ovaries mature. Immediately after the next moult the first batch .of eggs is laid. This constitutes the first adult instar. During the development of these eggs in the brood pouch, the second batch of eggs is maturing in the ovary, and the birth of the brood is followed in half-an-hour or so by an ecdysis and the laying of the second batch of eggs (the second adult instar) and so on. At a temperature of 22-5° C, the shortest duration of the life cycle (egg to first egg) is about seven days. Long series of measurements show that no appreciable growth takes place during the instars, but only at ecdysis. This applies to the regenerating antennae as well as to normal growth. Out of the thousands of individuals reared in this way, no sexual female occurred in the Simocephalus lines, and only three in Daphnia. A very few male broods occurred in the Daphnia lines, and a fair number in Simocephalus. These were not used for these experiments, which concern parthenogenetic females only. In confirmation of the work of Smith and of Banta, I may mention that although no single sexual Simocephalus female has occurred among the thousands produced in the main experiment (in which each individual was isolated at birth), I have never failed to obtain large numbers of these females by crowding. When a population of a hundred or so is allowed to form in a 500 c.c. jar, sexual females are always conspicuously numerous.

IV. THE MATERIAL AVAILABLE. The material dealt with in this study is' comprised as follows: (1) The "Test Experiments." At intervals comparatively large numbers of animals were tested for Lamarckian inheritance against control animals with no ancestral experience of regeneration. In these tests great care was taken to keep the conditions uniform for all the members of that particular test. The animals (with insignificant exceptions) were operated on 2-4 hours after birth, reared separately in phials containing 50 c.c. of culture medium drawn from the same, constantly stirred, stock jar, the phials all standing in a water bath maintained at 22-5° C. by a thermostat. Seven such test experiments have been carried out in the Simocephalus lines, co^kising altogether 537 individuals. In Daphnia there have been five tests, cornprising 530 antennae. 23-2 352 W. E. AGAR (2) 37° Simocephalus and 360 Daphnia reared under somewhat diA-| n conditions and only utilised for the effect of external conditions (Section IX (' (3) A large number of both genera used for special experiments on the effects of various conditions on regeneration.

V. STRUCTURE OF THE ANTENNA. In both genera the second antenna consists of a basal segment and two branches, the dorsal branch having four segments and the ventral three (Fig. 1).

Fig. 1. Second antennae of Daphnia (above) and Simocephalus. The dorsal branch is to the right in both figures. The arrows indicate the level at which amputation was performed. The lower figure is at a higher magnification than the upper one.

The dorsal branch carries four setae, and the ventral one five, in the positions shown in the figure. Each seta is jointed, approximately in the middle, into two segments, and is thickly covered or feathered with very fine hairs, which are longer in Daphnia than in Simocephalus. In the latter genus, the most dorsally situated of the terminal setae differs from all the rest in structure. The hairs are replaced by a very fine short comb, and the tip of the seta is sharply recurved to form a minute hook. It is by means of this hook that the animal suspends itself in characteristic fashion from weeds, or minute roughnesses on the glass sides of an aquarium. In Daphnia this seta is similar to the others, and therefore this genus has not the power of suspending itself by its antenna. The antennae are cylinders, the walls composed of two layers, hypodermisTITid Statistical study of regeneration in two species of Crustacea 353 Inside the cylinder are bundles of muscles, bathed in blood. These muscles run^J the tips of both branches of the antenna, and give off a twig to each seta. The setae are so jointed on to the antenna as to permit of movement through a right angle only. They can lie flat against the antenna (the point of the seta directed distally) or stand out at right angles to it. Observation of muscular twitches in freshly amputated and still living antennae show that these changes of position of the setae are carried out by muscles inserted into them. The setae can also bend at the joint in their middle, but also only through one right angle, the direction of motion from the extended to the flexed being towards the base of the antenna (downwards in the figure). Thus the setae are fairly complicated structures. They are provided with muscles (inserted into their bases, at least), they have definite joints, one at the base and one near the middle, so shaped as to allow of motion through one right angle only, and they are set with hairs.

VI. THE OPERATION. The dorsal branch of the antenna was amputated through segment III in Simocephalus, and segment II in Daphnia (I being the basal, and IV the apical, segment). These segments were chosen because (a) it was desired to amputate the branch of the antenna as near to its base as possible, and segment I is too small to work with, and (b) preliminary experiments showed that (unlike in Daphnia) very little regeneration took place in Simocephalus after amputation through segment II. In all lines of Simocephalus, and in two lines of Daphnia, the right antenna was operated upon. In a third line of Daphnia, the dorsal branch of both antennae was removed. Except when stated to the contrary, the operation was performed in the first instar. The new-born animal was placed in a 3 per cent, solution of ether for a few seconds to immobilise it. It was then placed on a microscope slide, lying on its right side with the right antenna stretched out in line with the body. The appropriate segment was then cut through with a fine needle, and the animal placed in a phial with 50 c.c. of the culture medium, where it remained till mature. The arrows in Fig. 1 show where the amputations were performed.

VII. GENERAL NATURE OF THE REGENERATION. The operation in Simocephalus removes one and a half segments of the antenna together with four setae. In Daphnia two and a half segments and the four setae are removed. Regeneration consists essentially of the formation of new setae, but the missing segments are never reproduced (Sciacchitano found the same in Ceriodaphnia). There may be a little growth of the stump of the amputated segment, but the missing joints are not formed. The regenerated branch of the antenna cc^fcts of a variable number of setae growing out in a bunch from the end of the truncated stump. Fig. 2 illustrates the variation in the number and length 354 W. E. AGAR of the new setae which may be formed in Daphnia. Figures for Simocephalu^re almost identical, except that the hairs on the setae are shorter and rather l^Rr, corresponding with the differences in the adult antennae. The average number of setae regenerated in Simocephalus is about i-8, and in Daphnia about 3-1. The regenerating antennae of a number of individuals of both genera were drawn and measured at each instar from the first to the tenth {i.e. fourth adult instar in Daphnia, fifth in Simocephalus). These measurements show that the regenerated setae continue to grow throughout life, as does the animal as a whole. The regenerated setae, however, grow relatively much more rapidly than the whole animal, though they never catch up with the setae of the unoperated antenna. There is often a considerable degree of necrosis of the regenerated setae in later life, which leads sometimes to an actual decrease in the length of individual setae.

Fig. 2. Examples of regenerated antennae in Daphnia, taken at the first adult instar. The number of setae regenerated in these examples varies from o to 5. The basal joint of the ventral branch of the antenna is shown in each case.

In Daphnia especially, new setae may grow out from the regenerating antenna after maturity is reached. Two specimens have added new setae even as late as the sixth adult instar. This matter is dealt with more fully below. In Simocephalus on the other hand additional setae are rarely formed after maturity. The condition of the antenna in the first adult instar is taken as the measure of regeneration unless otherwise stated. By whatever criterion used, the vigour of regeneration is higher in Daphnia than in Simocephalus. This is perhaps connected with the fact that the setae of the normal antenna are longer and more abundantly feathered in Daphnia (Fig. 1). The regenerated setae are usually very good replicas of the normal ones, provided with the proximal and median joints, furnished with muscles which move them on the antenna stump, and supplied with the usual hairy covering. The latter, however, is seldom as dense as in the normal setae, and occasionally the regenerated setae are naked, unjointed or definitely misshapen. The specialised hooked seta of Simocephalus never appears among the regenerated ones, which are always of the generalised type, even when four or more are regenerated (when the seta alone is cut off, leaving its base intact, it regenerates in its proper Statistical study of regeneration in two species of Crustacea 355 . was not part of the scheme of this research to investigate the histology of the regeneration process, but a few antennae of both species were stained and mounted at short intervals after operation. The immediate result of the operation is that tissues in the remaining stump of the segment form a plug, which fills up the wound and bulges beyond the chitinous cuticle. After a few hours, necrosis of this tissue takes place, which extends for a variable distance down the wounded segment. Apparently the amount of dying back of the tissue has an important influence on the amount of subsequent regeneration, as discussed below. After about six hours a cleavage appears between the living and the dead tissue, and the plug of dead material is soon after thrown off. In Simocephalus (operated through segment III)

1 7

Fig. 3. Daphnia. The regenerating antenna of the same individual drawn at successive instars. As these figures had to be drawn from the living animal, it was impossible to reproduce the feathering of the setae accurately; hence only the axes of the setae are shown. The number below each figure represents the instar in which the drawing was made, the operation having been performed in the first instar. The fifth instar is the first adult. The animal was not drawn at the sixth instar. the dying back in most cases extends to the base of the operated segment, the cuticle of which is left empty after the sloughing off of the dead tissue and is thrown off at the next ecdysis, so that regeneration appears to take place rather from the distal end of segment II than from segment III. Sometimes, however, a small stump of III is left.

VIII. MEASURING AND RECORDING THE AMOUNT OF REGENERATION. Four measures of the amount of regeneration were used. (1) The number of setae formed. (2) and (3) The length of the setae. This can be expressed either as the combined length of all the regenerated setae (total seta length), or as their average length (mean seta length). ^•k) The length of the antennar stump—i.e. segments I, II and sometimes (in l^micephalus) a portion of III. 356 W. E. AGAR While these are quantitative measures, they are indirectly qualitative Long setae are nearly always better^formed, and more thickly set with hairs, short ones. The length of the antennar stump and of the setae was measured as follows. The antenna was dissected off with a needle and mounted in dilute glycerine. A camera lucida drawing was then made of it, and the stump and setae, as magnified in the drawing, were measured. The measurement of the setae offered no difficulties, but owing to the obliquity of the bounding surfaces, the accurate measurement of the antennar stump was less easy. But the amount of regeneration in the stump is very small, involving only a slight growth of the operated segment and no formation of new segments. In any case, therefore, the stump does not afford a good criterion of regeneration, and little use was made of it. In the following account, all measurements have been converted into actual (unmagnifled) lengths in /x. The two principal measures of regeneration used were, therefore, number and length of setae. In any large group of animals, there are always some which fail to regenerate any setae at all. The mean number of setae regenerated per animal in such a group was obtained by dividing the total number of setae regenerated by the total number of individuals, including those which regenerated no setae. On the other hand, the mean total seta length was obtained by dividing the total summed lengths of setae regenerated by the number of animals which regenerated them—that is to say, omitting those which did not regenerate any setae at all. The first measure therefore is of the number of setae formed, the second is a measure of the amount of their growth. The length of the antennar stump is correlated (in Daphnia) with number of setae, total seta length and mean seta length, the coefficients of correlation being (183 antennae) + -59 ± -03, + -67 ± -03 and + -53 ± -04 respectively. In Simo- cephalus, on the other hand, there is probably no significant correlation between the dimensions, the coefficient for length of stump and total seta length being — 11 ± "o6, and for stump and number of setae in two experiments — -21 ± -09 and + -18 ± -io.

Many experiments made to test the point show that no change takes place in the number or length of the setae during one instar. Immediately after ecdysis the rudiments of any setae which have been forming grow out, or those already formed expand, and then no further measurable change takes place till the next ecdysis. Consequently all measurements made of different animals in corresponding instars are strictly comparable. Statistical study of regeneration in two species of Crustacea 357

THE INFLUENCE OF CERTAIN FACTORS ON REGENERATION. The factors which determine the amount and perfection of regeneration of the antenna can be divided into two classes. In the first class are certain obvious factors, easily controllable. Even when these factors are kept as uniform as possible however, the degree of regeneration still varies enormously in different individuals. No doubt laborious examination would reveal some of these factors, but probably others would resist experimental analysis, being of the nature of random combinations of very numerous, almost insensible, external and internal factors inter- acting together. The influence of these factors is discussed in Section X. In the present section we are dealing with the more obvious and controllable factors.

(a) The influence of the segment through which amputation is performed. As already stated, amputations were normally carried out through segment II in Daphnia and segment III in Simocephalus. Subsidiary experiments showed that regeneration would take place through segments II, III and IV in both genera. The number of setae formed is higher in Daphnia than in Simocephalus for all segments. The setae are larger and better formed the more distally the amputation is carried out—i.e. the longer the stump of the antenna left after operation.

(b) The influence of the level of amputation within the segment. The amount of regeneration is greatly influenced by the level of the operation within the segment—whether it is performed near the proximal or distal end. It is necessary to consider this otherwise unimportant point because of the bearing it may have on the variability of regeneration in the main experiments. When the amputation is carried out through segment II {Daphnia), the mean number of setae regenerated, and their length, increases the nearer the distal end of the segment the operation is performed, while in segment III the reverse is the case (both genera). This was discovered by special experiments devised for that purpose. For all other purposes, the influence on regeneration of the length of the wounded segment left was eliminated as far as possible by always making the cut as near the middle of the segment as possible. In the case of Simocephalus, this cause of variation was almost completely eliminated by this means. Had it operated, there should be a negative correlation between the length of the antennar stump, and the number and length of setae regenerated. As the coefficients of correlation in Section VIII show, this correlation is variable in sign and not significant. We can therefore be satisfied that the effect on regeneration of the level of amputation was fairly satisfactorily eliminated in Simocephalus by performing the operation always as near as possible to the middle of the segment. In Daphnia ever there is a significant positive correlation between length of stump and the er dimensions (Section VIII). And since experiments have shown that antennae 358 W. E. AGAR intentionally amputated near the distal end of the segment regenerate more strc than those amputated near the base, it is probable that this correlation is due to accidental deviation from the middle point of the segment when making the operation, resulting in longer stumps producing more and longer setae. Indeed, the length of the segment left after operation is probably the most important single factor in determining the amount of regeneration.

(c) Influence of age. Except where otherwise stated, all operations were performed in the first instar. Where uniformity of condition was specially important (as in the "test experiments" forming the bulk of the material utilised in this paper), the period of operation was further narrowed down to the interval 2-4 hours after birth. Operations were also performed on a few embryos of both genera, extracted from the brood pouch 4-8 hours before the normal stage of birth. The amount of regeneration, measured as always in the first adult instar, Was about the same as in those animals operated in the first instar. On the other hand, animals operated in the first adult instar produce significantly more setae than those operated in youth. Sixty-two Simocephalus, operated in the first adult instar, and examined in the fourth, showed a mean of 2-68 setae each, contrasted with i-8o, the mean number regenerated by 537 individuals after the usual operation in the first instar. No doubt the larger number of setae regenerated by adult animals is due to the larger surface exposed at the wound in the mature animal. In all probability the power of regeneration inherent in the tissues of the antenna does not vary significantly from the late embryo to well past maturity. It certainly shows no evidence of declining with age. This finding is contrary to the experience of workers on most other forms (Korschelt, p. 655).

(d) Influence of the simultaneous regeneration of the other antenna. Many workers have found that regeneration proceeds more quickly when a large part of the body, or a large number of organs (such as legs in Arthropods), is removed than when the lost parts are small or few (Korschelt, p. 649). This is not so in the present case, for when both antennae are removed, the average amount of regeneration of each of them separately is the same as when only one antenna is removed. This was tested as follows. Nine specimens of each of four broods (Simo- cephalus) had the dorsal branches of their right antennae amputated at birth in the usual manner; another nine from each of the same four broods had the amputation performed on both antennae. These 72 specimens were kept together in a 3-litre jar, and 70 of them preserved for examination in their first adult instar (two of those with right antenna only operated having been lost). The mean number of setae regenerated by the 34 with only the right a amputated was 1-41 ± •16. In the 36 with both antennae operated upon, the Statistical study of regeneration in two species of Crustacea 359 iae averaged 1-36 ± -16 regenerated setae, and the left antennae 1-75 ± -17, lg as the mean number of setae per antenna the figure 1*56 ± -n, which is not significantly different from the number regenerated when only the one antenna was amputated. The figures for total seta length are similar. When both antennae were operated upon, the mean total seta length regenerated per antenna was 434 ± 3O/A; when only the right antenna was operated, the figure was 450 ± 36/n. Thus the regeneration of an antenna seems to be unaffected by the fact that the other antenna is regenerating at the same time. This conclusion for Simocephalus was fully confirmed by a similar but smaller experiment on Daphnia.

(c) The influence of the previous regeneration of the other antenna. It would be important for our general knowledge of organic regulation if it were established that an organism carried out the process of regeneration better a second or subsequent time than it did the first. A number of workers have performed experiments designed to discover whether any change takes place in the rapidity or perfection of the regeneration of a lost part when the animal is compelled to regenerate it repeatedly by successive re- movals of the regenerate (reviewed by Korschelt, p. 573). No general conclusions can be drawn from these experiments. The technical difficulties in the way of solving this problem are obviously very great, if not insurmountable, owing to unavoidable changes in the physiological conditions of the animal, or the effect of purely incidental structural alterations in the neighbouring parts caused by the previous regenerations. As a contribution to this problem, I carried out the following experiment with Simocephalus. A new-born brood was divided into two equal groups. In one of them (group A) the dorsal branch of the right antenna was removed in the usual way. In the other (group B) the animals were treated in all respects like A, except that the right antenna was not operated upon. In the first adult instar the left antennae were amputated in both groups. The amount of regeneration of the left antennae was measured in the fourth adult instar. It was thus possible to find out whether the animals which had previously regenerated their right antennae regenerated their left antennae any way differently from those which had not had this previous experience. (The fact that the regenerating right antenna continues to grow throughout does not concern us, since, as we have just seen, the regeneration of one antenna is not affected by the simultaneous regeneration of the other.) This experiment was carried out with five different broods at different times. Summing the fiveexperiment s together, we have a total of 32 individuals in group A, and 30 in group B. The mean number of setae regenerated per left antenna in group A was 2-59 ± -23, and in group B, 2-77 ± -28, giving a difference of o-18 ± -36. As regards total seta length, the. figures are: group A, 721 ± 64/i, group B, 1047 ± 90ix, or a difference of 326 ± 110/x. in favour of those with no previous experience of regeneration. k small similar experiment with Daphnia involving 17 individuals showed no «lficant difference in regeneration of the left antenna between those which had 300 W. £. AGAR had previous experience of regenerating their right antennae, and those had not. It seems certain therefore that there is no "improvement with practice" as regards capacity to regenerate so far as the animal as a whole is concerned. On the contrary, there is rather uncertain evidence that previous regeneration of one antenna impairs the animal's power of regenerating the other.

(/) Influence of food. Since rapid growth is involved in regeneration, it seems reasonable to suppose that the amount of food available would affect the rapidity and extent of regeneration, especially in animals such as we are dealing with here, in which the rapidity of general bodily growth is much influenced by the amount of food available. It is well known, however, that in general the amount of nourishment available to the animal has very little influence on regeneration, starved animals often regenerating as rapidly as well fed ones (Korschelt, p. 654). In Simocephalus and Daphnia also, the amount of food, within the wide limits tested, does not influence regeneration. In an experiment directed to test this, a brood of 40 Daphnias was operated in the usual way, and divided into four groups A to D. Those of group A were each placed in 50 c.c. of very rich culture medium; B in the same medium diluted by an equal volume of filtered water; similarly the medium for group C was diluted to J, and for D to J. One effect of the different amounts of food was that the mean number of eggs produced by group A was more than double that produced by any others. Nevertheless there was no significant difference in the regeneration of the four groups, either as regards number or length of setae. The size of the animal as a whole is greatly influenced by the amount of food available, and the independence of amount of food and regeneration is indirectly shown by the independence of size of animal and amount of regeneration. The coefficient of correlation between total body length and (1) number of setae, was + -oi ± -oi {Simocephalus) and (2) total seta length, was + -io ± -09 {Simocephalus) and — -20 ± -07 {Daphnia).

is) Influence of the general internal conditions of the animal. Although all the members of a clone are genetically identical, they are of course by no means phenotypically identical. Even members of the same brood vary considerably in size at birth. There are often one or two which die in the brood pouch or shortly after birth. Even if kept in the same jar of culture medium, brood mates vary in regard to the number of eggs they produce, even to the extent of 100 per cent, or more. Regeneration seems to proceed uninfluenced by the differences in the internal physiological conditions which must correspond with these phenotypical differences. If regeneration were so influenced, this fact should be expressed in a correlation between the amount of regeneration in the right and left antennae c^fce same animal, in a population in which both antennae had been amputated. Wo Statistical study of regeneration in two species of Crustacea 361 t^^fcxperiments {Daphnia) in which both antennae had been operated upon were avJrlrable for this determination. Table I gives the coefficients of correlation between right and left antennae for the various measures of regeneration. Table I. Coefficients of correlation between right and left antennae (Daphnia). Experiment i Experiment 2 Number of setae - -03 ± 15 (44) — -oi ± -10(94) Total seta length - -oi ± -17(35) + -12 ± -10(89) Mean seta length + -43 ± -14(35) + -27 ± -10(89) Antennar stump + 22 ± -16(35) + -2O ± -10(89) The figures in brackets are the number of individuals. Another experiment, involving 85 individuals, was carried out on Simocephalus for this purpose. For some reason however conditions were very poor for regeneration in this experiment, and no less than 67 out of the 170 antennae failed to form any setae. Hence the distribution was hardly suitable for finding a correlation coefficient by the usual method. When however the antennae are divided into two classes, those which regenerated one or more setae, and those which failed to regenerate at all, the coefficient of association (Yule) between right and left antennae is found to be — -io. Thus it is evident that there is no significant correlation between degree of regeneration of right and left antennae as regards number of setae, or for total seta length. There are indications of a low positive correlation for mean seta length and length of antennar stump, but this is very doubtful, especially in view of the small number on which the largest coefficient is based.

(h) The influence of general external conditions. It has just been shown that the normal individual differences of physiological condition of the animals have no appreciable effect on regeneration (at any rate, on the number of setae formed). This is in agreement with the earlier finding that regeneration is independent of the nutritional condition of the animal, within the very wide limits tested; the fact that the regeneration of one antenna is unaffected by the simultaneous regeneration of the other is an example of the same kind. These facts combine to give an impression of a certain potency of regeneration resident in the antennar stump, which is but little influenced by the general physiological state of the animal. Nevertheless, the whole complex of external conditions does exert a considerable influence of regeneration; this is shown by the following: (1) It will be noticed in the next section, that one of the "test experiments" in Simocephalus gave a mean regeneration very much higher than any of the others. This can hardly be ascribed to any other cause than some unknown difference in the culture medium for this experiment. (2) A population of Daphnia was synthesised from 42 groups of ten specimens the members of each group being reared together in one vessel, in culture rnRTia prepared at different times. In this population a high correlation was 362 W. E. AGAR exhibited between the members of each group, which can only be attribute^k the influence of differences in the culture media of the different groups. The aninWs forming this population (which was reduced by death and accident to 360) were classified as (1) with none or one setae, (2) with two or more setae. A fourfold table was constructed by associating each animal with every other animal reared in the same vessel. The number of entries in the table was 2874, and the coefficient of association + -44. In Simocephalus, a similar population composed of 43 groups, with a total of 37c individuals, yielded a coefficient of association of + -22. (3) There is, also apparently conclusive evidence (in Daphnia) of an effect of external conditions of a remarkable kind. The above population of 360 individuals, composed of 42 groups of a maximum of 10 per group, each group living in 500 c.c. of medium, averaged 1 -76 ± -06 setae regenerated per animal; the "test experiments," on the other hand, yielding a population of 530 animals each isolated in 50 c.c. of medium, averaged 3-13 ± "04 setae. The only known difference of conditions in the two populations was the one just mentioned. I can offer no explanation for the enormous influence which it apparently exerted on the regeneration.

It cannot be doubted, therefore, that the general complex of external conditions influences regeneration. It may be recalled that temperature is not one of the conditions concerned, since this was kept uniform throughout the experiments.

X. THE NATURE OF THE INTRINSIC FACTORS CONTROLLING REGENERATION AS DISCLOSED BY STATISTICAL ANALYSIS. In this section we can fairly claim to be dealing with variation due to intrinsic factors, since we are dealing with the "test experiments" in which all animals were treated in a strictly comparable way. Amputation was performed at the age of 2-4 hours after birth, and each animal was reared from that point to maturity in a measured 50 c.c. of culture medium drawn from the same, constantly stirred, jar. The phials stood together in a water bath at 22-5° C. As we have already seen, the only discoverable intrinsic factor which exerts a significant influence on regeneration is the length of the segment left after operation; we have shown that care to perform the operation in the middle of the segment satisfactorily eliminated that source of variation in Simocephalus, and while it probably had a considerable share in causing the variation in Daphnia, it certainly was only one among many other causes, for the correlation between length of antennar stump and degree of regeneration is far from complete (Section VIII). A statistical analysis of the variation in the number of setae regenerated allows us to formulate the situation as follows: (1) The number of setae regenerated depends primarily on intrinsic factors localised in the regeneration blastem (the tissue forming below the wound surface, from which the growth of the new parts takes place). The combined operatiq^fcf these factors results in a distribution of number of setae approximately accoramg Statistical study of regeneration in two species of Crustacea 363 he normal probability distribution. . Hence we may conclude that these factors »numerous and relatively independent (such as the numerous factors determining the number and pattern of cells closing the wound and the amount of dying back of the injured tissues, etc.). We will call the physiological state resulting from the combined action of these factors the regeneration potential of the regeneration blastem. The value of this potential, which expresses itself primarily as the number of setae regenerated, follows the normal probability distribution. (2) The resulting form of the distribution of seta number (which has a standard deviation of about 1-33 setae) is modified at its left-hand end by the fact that the regeneration potential must reach a certain threshold before any seta can be formed. All antennae in which the potential fails to reach this point appear alike and have no setae regenerated. (3) After the threshold has been passed, the increment in potential corresponding to one additional seta is about the same in all parts of the scale until the number 4 (which, it will be remembered, is the number of setae removed by the operation) is reached. After this point, larger increments of potential are required to add each additional seta. (4) As in all morphogenetic processes, the intrinsic factors operate in relation to external conditions. Variations in the latter shift the whole distribution of setae to the right or left as they act in a manner favourable or adverse to regeneration (cf. groups A and B, Table II). The data on which these conclusions are founded are as follows. The seven test experiments of Simocephalus fall into two groups: Group A, containing six of the experiments, in which the mean number of setae regenerated was, respectively, 1-40, 1-57, 1-58, 1-65, 1-67 and 1-77, giving a total mean (417 individuals) of I-6I ± "04. Group B, comprising the one remaining experiment, with a mean (for 120 individuals) of 2-44 ± "08. It is clear that for some reason conditions prevailing at the time that the latter experiment was being performed were more favourable for regeneration than when the experiments of Group A were in progress. The five test experiments of Daphnia were sufficiently uniform to allow of them being treated as a single group. Table II gives the actual distributions of the number of setae in the two groups of Simocephalus and the single Daphnia group, together with the normal distributions for comparison. In calculating the latter, the medians of the actual distributions were used as the means, since the condensing of the actual distribution at the extremities obscures the true mean. Also the standard deviations of the theoretical distributions are taken as rather larger than those of the actual distributions, since the deviations of the latter are reduced by the terminal condensations. A study of Table II shows the justification for the above hypothesis regarding the variation of the regeneration potential and its expression as number of setae uced. Considering the two extremities of the distributions, it will be seen that £es o and 1 added together correspond very closely in the actual and theoretical 364 W. E. AGAR distributions, though taken separately they show considerable discrepancies. is doubtless because of the arbitrariness of putting the midway point between tne two classes as the point where the threshold is passed. Table II. Variation in number of setae regenerated; comparison of actual distribution of number of setae with the underlying theoretical normal distribution of regeneration potential. A and B the two groups of Simocepkalus; C, Daphnia.

Number of setae 0 1 2 3 4 S 6 789 Total

A. Actual distribution 84 121 114 7i 24 3 0 417 M = I-6I, a = 1-19 Normal distribution 9i "3 us 68 24 5 1 — — — 417 M = i-53, a = 133 B. Actual distribution 12 14 33 33 26 2 0 — — — 120 M = 2-44, a = i-27 Normal distribution 8 19 33 33 20 7 2 — — — 122 M = 253, a = i-33 C. Actual distribution 24 27 90 168 173 38 6 211 53° M = 3-13, a = 1-29 Normal distribution 10 39 103 156 I3S 66 19 300 531 M = 3-24, a = 130

At the right-hand end there is a tendency for the numbers in class 4 setae to be in excess, and for those beyond it to be in deficiency, indicating that after the number 4 is reached, larger increments of potential are needed to add an additional seta. Consequently there is a telescoping of the distribution of seta number from the right-hand end on to the number 4. This is most obvious in the case of Daphnia where, owing to the higher mean, the distribution is more extended to the right than in Simocephalus. It will be noted that in Simocephalus no antenna with more than five setae appear in these distributions. Two such cases have, however, appeared among about 1300 operated in the standard manner. The justification for the above interpretation is made clearer by giving the distributions in the condensed form shown in Table III. Table III. The distributions shown in Table II reproduced in condensed form.

Number of setae 0—1 2 3 4-9

A. Actual 205 114 7i 27 Normal 204 "5 68 3° B. Actual 26 33 33 28 Normal 27 33 33 29 C. Actual 5i 90 168 221 Normal 49 i°3 156 223 Statistical study of regeneration in two species of Crustacea 365

XI. DISCUSSION OF RESULTS. We have seen ftiat when the dorsal branch of the second antenna of Simocephalus or Daphnia is amputated, the missing segments are not regenerated, but new setae are formed in variable number and length. Since we may assume genetic identity in our material within both species, we must look for other than genetic factors to account for this variability. It has been shown that the causes of the variation are mainly numerous small causes localised in the regeneration blastem itself, and that regeneration is but little affected by the general condition of the animal (as evidenced, especially, by absence, or low degree, of correlation between the amount of regeneration in the right and left antennae of the same animal). These numerous intrinsic factors combine to give the regeneration potential, the value of which varies approximately according to the normal probability distribution. The action of the complex of external conditions (from which we may exclude food and temperature) is, in these experiments, comparatively small, and is confined to shifting the whole distribution to the right or left. Animals that fail to regenerate at all merely represent the left-hand end of the probability distribution of regeneration potential. They represent those blastems in which the value of the potential did not rise above the threshold value (Section X). It is plain therefore that differences :n degree of regeneration—or even whether regeneration takes place or not—depends not upon any deep-seated differences between different individuals, but simply on random combinations of small factors in the blastem itself. It has been suggested {e.g. by Schaxel) that the difference between certain classes of animals which have high and low powers of regeneration is of this character, being due to comparatively trivial happenings at the locality of the wound, and not to any important difference in the nature of the organisms them- selves. These experiments are in favour of this view.

The form of the setae regenerated follows closely that of the normal (jointing, feathering, musculature), except that they are frequently defective in some degree, and that the specialised hooked dorsal seta in Simocephalus never appears. It is plain however that the number of setae regenerated does not bear any close relation to the number (4) removed. If however we look at the form of the distributions of seta number in the preceding section, we see clear evidence that the number 4 occupies a privileged position. There is a tendency for this number to occur more often than the theoretical expectation, and for the numbers above it to be in deficiency. This was interpreted as showing that after the number 4 is reached, larger increments of regeneration potential are required for the addition of another seta. This could also be expressed as a tendency to satisfaction of the regeneration stimulus by the formation of four setae. ^ttThe fact that there is so much variation in the number of setae regenerated may b^connected with the non-regeneration of the missing segments of the antenna, JEB'Vlliv 24 366 W. E. AGAR thus making it impossible to regenerate a perfect organ. The missing part antenna is not regenerated as a whole; only setae are formed. The effect incompleteness of the regeneration of the axis of the antenna'is shown by the following experiment. Sixty-nine specimens of Daphnia were operated through the middle of the terminal segment (IV), thus removing the distal half of the segment together with the three terminal setae. When the setae are regenerated, there is therefore no incompleteness about the antenna except for a shortening of the last segment. Table IV shows that under these conditions the mean number of setae regenerated is very near the number removed (3 in this case, instead of the usual 4) and that variation from this number is slight, and mainly on the side of deficiency. Table IV. Daphnia: distribution of number of setae in 530 individuals amputated through segment II (Group A) and in 69 individuals amputated through segment IV (Group B).

Number of setae ... 012 3 4 56789 Mean a

A 24 27 90 168 173 38 6 2 1 i 129 B 0 016 45 8 00000 III 058 The setae regenerated after amputation through segment IV are not only nearer the normal in regard to number, but also in form. They are more uniform and perfectly formed than those regenerated from segment II. Thus there is evidently a much stronger tendency to reproduce the correct number (3) of setae after operation through segment IV, than there is to produce the correct number (4) removed by operation through segment II. It seems probable that the reason for this is that by the reproduction of three setae after operation through IV the antenna is practically restored to its original condition, and the regeneration stimulus is satisfied. But owing to the failure to regenerate the missing segments, the formation of four setae after operation through II by no means restores the antenna to its normal condition, and therefore the stimulus to regeneration cannot be fully satisfied.

A connection between failure to complete the axis of the antenna and in- determinacy in the number of setae regenerated is shown even more strikingly in another way. As we have seen, in Daphnia (and to a much less extent in Simo- cephalus) new setae are added to the regenerate throughout life. This applies parti- cularly to the standard operation through segment II. When, however, the operation takes place through IV (that is to say, when the axis of the antenna is left almost complete), very few new setae are added after the initial regeneration. Thus, 17 specimens of Daphnia, operated in the usual way through segment II in the first instar, were followed beyond the first adult instar, the number of setae present at each instar being recorded. One of these animals was followed to the fourth, five to the fifth, ten to the sixth, and one to the eighth adult instar. Of fe individuals, four specimens added two new setae to those already formed i Statistical study of regeneration in two species of Crustacea 367 ) adult instar, two added three, two added four, and one added seven new setae. Expressed as the mean number of setae added per instar after the first adult instar, this gives 0-362 new seta added per individual per instar. In a parallel experiment, 15 animals were operated through segment IV and followed up to the later instars (one to the fourth, two to the fifth and twelve to the sixth). Three individuals added one new seta each; none of the others formed any new setae. This gives a mean of only 0-042 seta added per individual per instar. I suggest that this striking difference is a further indication of a common organic basis of psychical and morphogenetic processes—and indeed of vital processes in general. Consider an instinctive action. This consists of a train of behaviour appropriate to the attainment of a specific goal. It is set in motion by the combination of a certain physiological state of the organism, and a certain external situation. The instinctive action ceases on the attainment of the goal. If for some reason, the proper end of the instinctive action is not, or cannot be, attained, the animal will continue for an indefinite period (depending on various factors) to perform the type of action (varied according to the principle of trial and error) which is appropriate to the attainment of the goal. In the case of morphogenesis which we are considering, the goal is the complete antenna, which is normally reached with a very high degree of precision in embryonic development. By the removal of part of the antenna, a situation is created essentially similar to that of an organism with an unsatisfied instinct and the blastem sets in operation a series of processes appropriate to the attainment of a complete antenna. In the terms of Gestalt psychology, the configuration is open, and "closure" tends to take place. For some unknown reason, however—perhaps a purely incidental, mechanical one—new segments of the antennar axis cannot be regenerated. Hence, after operation through segment II, the goal—a complete antenna—is impossible of achievement. Consequently the natural cause of cessation of the regenerative process never operates; the configuration remains open, and the formation of new setae continues throughout life. When, however, the amputation is carried out through the terminal segment, the inability to regenerate new segments is of little consequence, and when the three lost setae are regenerated,.the antenna is restored very nearly to its original form, differing from it in little else than an abnormally short terminal segment. Hence the goal is nearly attained by the formation, of three setae, and the stimulus to regeneration is thereby satisfied in most (but not quite all) cases.

XII. SUMMARY. 1. The material for this study consists of over 1900 camera lucida tracings of antennae of the two Cladoceran genera, Simocephalus and Daphnia, in measurements of the regenerated parts were made. In about a thousand more examination was confined to the number of setae regenerated. 24-2 368 W. E. AGAR 2. The standard operation consisted of the amputation of the dorsal ^^ the right antenna (in some experiments, of both antennae) through the second segment (Daphnia) or third (Stmocephalus), thus removing four setae in both genera, together with two and a half segments of the antennar axis in Daphnia, and one and a half in Simocephalus. 3. The missing segments are never regenerated, but a variable number of setae are produced in place of the four removed by the operation. The setae have a fairly complicated structure. 4. The degree of regeneration was chiefly measured by the number and length of setae regenerated. 5. The effect of varying the segment of amputation, or level of amputation within a segment, is described. 6. The vigour of regeneration shows no decline with age (at any rate, from late embryo to first adult instar). 7. The regeneration of one antenna is not affected, either favourably or un- favourably, by the simultaneous regeneration of the other. 8. No "improvement with practice" takes place. The regeneration of one antenna is either unaffected, or adversely affected, by the previous regeneration of the other. 9. Within the wide limits tested, the state of nutrition of the animal has no influence on regeneration. 10. The general internal condition of the animal, within the limits of variation dealt with, has very little influence on regeneration, as shown by the absence or low degree of correlation between the regeneration of right and left antennae of the same animal. 11. The general complex of external conditions exerts a significant influence on regeneration. 12. The main cause of variation in degree of regeneration is to be looked for in very numerous small factors localised in the regeneration blastem. The combined action of all these factors determines the regeneration potential. The value of this varies approximately according to the normal probability distribution, and expresses itself primarily in the number of setae produced. The left-hand end of the distribution of seta number resulting from the distribution of potential is cut off owing to the fact that the potential must rise beyond a certain threshold value before any setae can be produced. The distribution is condensed at the right- hand end owing to a tendency for the formation of new setae to cease when four have been formed. 13. Owing to the failure to regenerate the missing segments of the antennar axis, the formation of new setae after amputation near the base of the antenna only very imperfectly restores the original condition. When the amputation is performed through the apical segment, however, the regeneration of the setae removed by the operation practically restores the antenna to its original state. It is suggestedjhat this is connected with the fact that after amputation near the base of the an^^ia, regeneration is highly variable and indeterminate, new setae being added to the Statistical study of regeneration in two species of Crustacea 369 nerate throughout life; whereas after operation through the apical segment regeneration generally consists in the formation of the proper number of setae and then stops.

REFERENCES. Further references can be found in the bibliography to Korschelt's text-book by consulting the references to that book contained in the text. AGAR, W. E. (1914). Phil. Trans. Roy. Soc. London, B, 205, 421-489. BANTA, A. M. (1921). Carnegie Institute Publication, No. 305. BANTA, A. M. and BROWN, L. A. (1929). Physiological Zoology, 2, 80-92. KORSCHELT, E. (1927). Regeneration und Transplantation, 1, Berlin. SCHAXEL, J. (1921). Untersuchungen fiber die Formbildung der Tiere. Erster Teil, Berlin. SCIACCHITANO, J. (1925). Zool. Anz. 62, 173-177. SMITH, G. (1915). Proc. Roy. Soc. London, B, 88, 418-434.