Territorial Behaviour in Barnacle Settlement

When a heavy settlement of the larvae of certain sedentary marine organisms has taken place on a uniform surface, the resulting pattern may quite clearly appear to be non-random, the individual larvae having each settled at a more or less uniform distance from one another. Such a pattern is formed during the settlement of barnacles and of the tubeworm Spirorbis borealis (Wisely, 1960). On the other hand, some marine larvae, notably those of many Polyzoa, settle apparently at random on a uniform surface, individuals sometimes touching each other. When the settlement is less dense, the tendency for barnacles and Spirorbis to be spaced out may not be so obvious to the eye, presumably because the distance over which the larvae regularly avoid each other has become small in relation to the total distance separating them. Nevertheless, the avoidance of the territory around settled individuals can still be revealed by a suitable analysis of the pattern.

The simplest analysis can be made by counting the number of individuals in each of a large number of equal quadrats ; if settlement is random these will fit the Poisson distribution; if the larvae show a territorial tendency the counts will be more uniform; if they are gregarious the counts will be more divergent. The significance of their deviation from the Poisson distribution may be obtained by means of the χ² test. A typical example of such a treatment applied to a fairly heavy settlement of cyprids of the barnacle Elminius modestus is shown in Table 1. There are certain difficulties in applying this treatment. It is, in practice, so difficult to provide a surface uniform in respect of all environmental factors, that settlement is rarely uniform at the outset. Moreover, the gregarious tendency shown by some larvae, notably by barnacles, leads to an increase in settlement rate in the presence of settled individuals (Knight-Jones & Crisp, 1953) so that a high rate of settlement tends to be maintained in the neighbourhood of the first groups of individuals to become established, further settlement spreading thence outwards in course of time. Consequently, any initial unevenness in the early phase of settlement is accentuated. In making a test for territorial behaviour it is clearly necessary (a) to restrict counts to an area of comparatively uniform settlement, and (b) to use a quadrat small in comparison with the size of any patterns formed by the inhomogeneities in settlement density. With the use of quadrats of increasing size the effect of uneven or gregarious settlement is exaggerated so that the frequency dispersion increases and may eventually exceed the value expected from the Poisson distribution.

Apart from establishing a significant degree of over-dispersion, this form of analysis gives no information on the nature of the pattern which develops, and in particular fails to indicate the distance over which the repulsion between previously settled and exploring larvae operates. A far more revealing analysis can be applied by measuring the average population density as a function of the distance from any one individual. This analysis will be described both for the settlement of larvae in one dimension, along a line, and for the more usual case of settlement over a uniform area.

In organisms whose larvae are attracted to grooves (rugophilic) settlement may, at least at first, be entirely constrained to a predetermined line (Crisp & Barnes, X954). If the position of each individual along such a line were independent of that of the others and of any other factor, the arrangement would be a random linear series. If the mean linear density were ρ individuals per unit length, the chance of finding an individual in any short length dl would be ρ dl. If we now take any one individual as a reference point, we may consider how far along the line we are likely to go before reaching the next individual. The chance of finding the next individual at a point lying somewhere between l and l+dl from the reference individual will be called P₀, the subscript referring to the condition that no individuals are present between the zero point and l. This probability, P₀, will be the product of two independent probabilities, namely that of not finding an individual between 0 and l, and that of finding one between l and l+dl. The former probability will be taken as a function of l, f(l), and the latter will be ρ dl\ hence P₀= f(l) ρ dl. The probability of not finding an individual between I and l + dl is clearly (1 —ρ dl), and therefore the probability of not finding one between o and l+dl will be the product f(l)(1 — ρ dl). Hence

The analysis can be made more informative if we consider not simply the distribution of distances separating successive individuals, but the density of settlement as a function of distance from the reference individual.

This equation therefore gives the distribution of distances between any two pairs of individuals on a plane surface. It will be seen that the relationship differs from that of the linear series in that the most probable distance between successive individuals is not zero, but finite. The modal value of P₀ will occur where dP₀/dr = 0, i.e. at ⁠.

Text-fig. 3 illustrates by means of a broken line the distribution of distances expected for a settlement of density 10·4/cm.². The points and the full line connecting them show the actual distribution for a settlement of Balanus crenatus of population density 10·4/cm.². It can be seen that there are few spacings within 1 mm. of a settled individual, while at a distance of 2−3 mm. the frequency of spacings is abnormally high, so that the modal distance is greater than ⁠, and there is also a much more pronounced peak.

In order to illustrate more precisely the effect of an established individual on settlement, essentially the same treatment was applied as for the linear series. A large number of individuals were taken as reference points, the number present in successive annuli counted, the results pooled and averaged, and finally expressed as the number per unit area by dividing by the area of the appropriate annulus. In practice this operation was carried out on an enlarged photograph of the settlement pattern, not on the settlement itself. A transparent grid, on which successive annuli had been tinted in order to allow easy identification of the annulus, was placed over the photograph for the purpose of counting. The results of a number of such observations on different species are described below.

As the available surface becomes more densely populated, subsequent arrivals will either have to settle closer to established individuals or else will have to search out the few unoccupied territories still available, or may do both. The influence of the density of the population on the settlement pattern is shown in Text-fig. 4 for the immigrant species Elminius modestus, which settles at extremely high population densities in the Essex rivers during the summer months. The curves all show a general similarity. The territory situated 1 mm. or slightly less from the reference individual is avoided, especially in the more sparse settlements. The population density then rises very sharply and at distances varying from 1 to 4 mm. reaches a value in excess of the normal level.

Since the population density curve rises so steeply at a certain distance from the reference individual, we may reasonably use the distance between the reference individual and the point at which the steep part of the curve cuts the average value of population density as a measure of the avoidance of the territory of established spat by later colonizers. This distance, which will be referred to as the territorial separation, becomes reduced as the population density increases. This is seen more clearly in the inset graph in which the same data are presented, but the absolute population density is replaced by the population density expressed as a fraction of the normal value found at some considerable distance away from the reference individual.

The fact that population density influences the territorial separation between individuals complicates any comparison of the behaviour of different species. Textfig. 5 gives results for two species, Balanus crenatus and B. balanoides, whose cypris larvae and metamorphosed spat are larger than those of Elminius modestus, while Table 2 summarizes the values of territorial separation for all three species, including those of linear settlements of Balanus balanoides shown in Text-figs. 1 and 2. It will be seen that the species with larger larvae and spat have larger territorial separations. The two curves for B. crenatus were from the same settlement, using, as reference individuals, just-metamorphosed cyprids (curved) and the largest spat (curveB). The results show that the settlement around larger spat (mean diameter 1·7 mm.) has a territorial separation which exceeds that around cyprids and just-metamorphosed individuals of diameter 0·4 mm. by a distance equal to the difference in the radius of the central individual. Evidently territorial behaviour depends on physical contact in barnacles, as it does in Spirorbis (Wisely, 1960), the newly settled individual having spaced itself from the edge of the previously established one. Moreover, if the mean radius of the existing spat (d₂, column 4) is added to the mean length of the cypris larva (d₃, column 5) and the total subtracted from the value of the territorial separation, the difference for all species is a little more than half a millimetre in sparse populations, and approaches zero in dense ones. It is known that during the early exploratory phase of settlement the larva explores a wide area, while towards the end of this phase, and just prior to fixation, it moves to and fro repeatedly over a small space (Knight-Jones & Crisp, 1953), moving scarcely more than its own length from its chosen position and making occasional swivelling movements (Text-fig. 6). These final movements would cause it to keep at a distance of at least its own length from the edge of any outstanding object, such as its own spat, unless it deliberately settled in the angle between the object and the surface. This pre-settlement behaviour can therefore adequately account for the values of territorial separation given in Table 2.

If there were sufficient space available few individuals would chance to settle in the position shown in Text-fig. 6 with the caudal setae as close as possible to another individual. If the amount of space had become very restricted, as in a dense settlement, such a site might be the only suitable one to be found, and so an increasing proportion of individuals, even after a long search, would lie within this distance of each other.

When a settling cyprid encounters a small barnacle it may receive two kinds of stimuli : a simple physical stimulus, which would be invoked by touching any similar object, and a more specific stimulus, possibly chemical, through which it might recognize the object as another barnacle. When a barnacle is removed from a surface it usually leaves behind its base, which remains firmly cemented in position. Knight-Jones (1953) demonstrated that barnacle cyprids could recognize these bases, as well as the living barnacle, and if they belonged to the same or related species settlement was encouraged (Knight-Jones, 1955). Hence we may assume that if the base were flat it could be specifically recognized, though the physical stimulus it exerted would be small.

A number of plates of inert plastic, on which small pits had been drilled in order to localize settlement, were placed in the sea and one barnacle was allowed to settle and grow in each pit (see Crisp, 1960a). After the barnacles had grown to about 1 cm. in diameter the pits were usually to be found near the centres of the bases. The basal membrane covered the pit, but did not fill it up, and the membrane itself replicated the concavity. The barnacles were then cleaned off the panels, leaving behind the bases, together with a few strands of muscle tissue. These were then exposed in the sea to settlement during the following season. Pl. 1 shows the appearance, both of the original settlement of adult barnacles and of the later spat in identical positions on two of the panels. It can be seen that the new settlement was very dense in the membrane-covered pits situated roughly at the centre of each base, and another dense zone of settlement took place at the periphery of the base. The remainder of the base was generally sparsely colonized.

Text-fig. 7 shows population density as a function of distance from a reference point situated in the position of the pit where the adult first settled, after averaging a series of counts over some twenty-five bases. The dense settlement in the original pit is shown by the peak on the vertical axis : then follows a region of low settlement density; finally the curve rises to a further peak at a distance of 4 mm., approximating to the radius of the base of the adult shell. The low value of population density just outside the pit may be caused in part by the territorial influence of the spat settled in the pit, but this cannot account for settlement densities less than the average at distances of up to 4 mm. from the pit, because the territorial influence does not normally extend beyond 2 mm. (Table 2). Moreover, a few adult barnacles did not settle over a pit, and their bases therefore did not include a concavity (Pl. 1, X, Y). These also showed sparse subsequent settlement within the basis, though they lacked the dense central colony. These experiments indicate that the centre of the base tends to be avoided, but the periphery to be heavily colonized. While this behaviour suggests that there may be a specific response to the integument causing the barnacle to avoid the base of a detached individual or one of its own species, there remains the possibility that the centre of the base or the outside of a small individual may have a physical contour that is unfavourable for settlement in comparison with other surfaces nearby.

Concavities in the surface are preferentially colonized by cyprids (Crisp & Barnes, 1954). When more than one cyprid enters a small pit, it is forced to settle very close to the original occupant. Hence within such a pit one would expect territorial behaviour to be reduced or eliminated. In some instances cyprids can be seen to have settled so close together that some are unavoidably in contact with small spat of their own species. However, it is possible that the behaviour leading to the recognition of and settlement in a pit may be separate from the recognition and avoidance of a previously settled individual. If this were so, the two elements of behaviour might play opposing roles during the colonizing of a small pit.

An experiment was devised to investigate whether there was any tendency to ‘space out’ when settlement occurred within a small pit. Into a sheet of inert plastic were drilled out three groups of hemispherical pits having different radii of curvature. Twelve of each of the different sized pits were randomized into groups of 36, and each of these groups were replicated three times. After drilling, the pits were polished with fine carborundum followed by jewellers’ rouge, then scrubbed with a detergent solution, washed in water, and exposed to a settlement of Balanus balanoides.

The areas within the pits were so limited in extent and curved in shape that it was impossible to make counts at various distances from each individual. Instead the distances between neighbouring individuals were measured, and their frequencies plotted to find whether the modal value differed from that calculated from random expectation. Text-fig. 8 shows the series of curves obtained for a plane surface and for each of the series of pits of increasing curvature. The shift of the curves towards smaller values of separating distances as the curvature of the pits was increased indicates the concentration of spat in the smaller pits under identical conditions of settlement. The modal value of each frequency curve is shown in Table 3, together with the value expected for a random arrangement within the pit. It will be seen that, even at the highest population density, the cyprids appear to be spaced out more than would be expected from chance. However, at the higher population densities, the expected modal distance is so close to, and in one case less than, the sum of the radii of two metamorphosed spat (0·6−0·7 mm.) that the apparent spacing-out effect might be due only to the fact that two adjacent small barnacles cannot overlap each other. An alternative method was therefore employed to confirm that territorial behaviour was displayed during settlement in pits. A number of pits of different sizes were drilled out on a plastic sheet to form a pattern of squares containing 36 pits, each pit being 1 cm. from the next. The pits were so arranged that each occurred once in each row and once in each column, forming a 6 × 6 Latin square design. Three such square patterns of pits were repeatedly exposed to settlement by B. balanoides, and the number of cyprids and spat counted. The experiment was continued until an average of 2 to 3 spat were present in the smallest pits and 6 to 7 in the largest; the pits were then cleaned out and the experiment repeated. In this way it was possible to obtain for each type of pit a series of curves showing how the proportions of pits containing zero, one, two, three, etc., individuals varied as the average number per pit increased. If settlement were random, the curves would follow the Poisson series of distributions, and the variance between counts would equal the mean number per pit. If the territorial effect were infinitely strong, all the pits would be filled by one individual each before any were entered by a second, and so on. The value of the variance would then always lie between o (when the average number per pit was an integer) and 0· 25 (when the average number contained half an integer). A simple presentation of the results is shown in Table 4 where the proportions of pits with 0, 1, 2, 3, etc., individuals are given for a mean settlement of 1·0 and 2·0 individuals per pit respectively. At these relatively low values of settlement density there was adequate space in each pit, and particularly in the larger pits, for further individuals to settle. This is clear from the columns giving the dimensions of the various pits and the maximum number observed in each pit at high settlement density. The figures show that the distribution is very different from the Poisson; the central frequencies are much higher. Even in the largest pits there was only a slight approach towards the values predicted for chance settlement. The variance increased in the larger pits, but was everywhere well below the Poisson value. Evidently the existence of an individual settled in a pit greatly reduced the chance of a second entering the same pit.

Cyprids of B. balanoides were obtained from plankton hauls during April and observed in the laboratory under a low-power binocular microscope as they explored and settled on various surfaces presented to them. The cyprids were presented with the following surfaces: barnacled stones, both smooth and rough; barnacled stones with various densities of young spat already settled on them ; flat plastic sheets covered in small hemispherical pits spaced 0·5 mm. apart of radius of curvature of approximately i mm.; and flat plastic sheets with grooves approximately 1 mm. wide and $12$ mm. deep containing a few spat. In one experiment on the pitted surface, each pit contained an average of two spat, in the other experiment the pits had been cleared, leaving only the bases of the spat behind. Each cyprid was carefully watched and timed to find out how long it spent in exploring, what proportion of this time was spent within pits and grooves, and whether the exploration terminated in settlement. The cyprids showed two types of movement on the antennules, a simple translatory sten which took them on a more or less straight course through a distance of about 0·5 0·75 nun., and a less deliberate movement, characterized by pivoting and oscillating on the antennules, often accompanied by a change of direction, which will be called a ‘halt’. Translatory steps generally predominated during the early phase of settlement, and halts towards the end. Table 5 records the length of time of the exploratory phase and shows very clearly the influence of previously settled spat or their bases. These provide the necessary stimulus which causes the cyprids to remain on the surface and to continue to search for suitable sites on which to settle. However, although this response was common to all surfaces which were barnacled, the best settlement occurred where a low density of settled spat (0−3/cm.²) was present on the surface. Where the density was more than 10/cm.² a clear reduction in the tendency to settle was evident.

The time that was spent exploring the surface was not greatly affected by the presence of dense spat. In fact, the exploration time on panels containing pits was greater when spat had already occupied the pits than when the pits were vacant. The cyprids could be seen entering the occupied pits one after another, exploring them and walking on to the next, apparently seeking an empty one. Indeed the only occasion when a cyprid was observed to settle on this type of panel took place when it encountered a pit which happened to be unoccupied. The number of times that cyprids entered pits, walked out and entered others is shown in Table 6.

The exploration time of the cyprids which were presented with the grooved panel containing only a very few spat in one corner was much less. This could be attributed to the small number of spat, which resulted in few of the larvae encountering them, the majority making short and fruitless searches which terminated in their swimming away from the surface. Those which did encounter spat remained for about the same length of time as those on more heavily barnacled surfaces, as can be judged from column 5 which gives the maximum times.

Unlike the territorial behaviour of active animals, in which the holder of a territory usually displays aggression, that of sedentary animals must be attributed solely to the behaviour of the invading individual, through which the territory of the first is respected. It can hardly be doubted that the observed separation is brought about by the characteristic short to and fro excursions and the twisting movements that take place repeatedly just before fixation.

The territory avoided by later arrivals has been shown to be quite limited in range. It does not usually extend beyond a circle of radius 2 mm. from the previously settled individual. Of this distance 0·3−0 ·5 mm., according to the species, must be allowed for the radius of the central individual and a similar distance for the radius of the new arrival because the shells of the two individuals obviously cannot interpenetrate. Only the observation of a regular separation of individuals beyond 0·6−1·0 mm. is therefore of consequence. Such separations, though small, are nevertheless real and may be retained even at population densities as high as 20 to 30 individuals per square centimetre.

In spite of its very short range, territorial behaviour is particularly interesting and significant because it seems contradictory to the thigmotropic tendency displayed by cyprids when they settle preferentially in grooves and pits. One might expect from their reaction to concavities that they would settle in close contact with other individuals, nestling in the angle between them and the substratum. However, it is not certain that cyprids settle in depressions simply in response to increased physical contact. There are a number of reasons for believing that they may be capable of recognition of surface contour. First, they do not settle into a pit as soon as they come to it, but explore it, entering it or similar pits many times. Secondly, pits and grooves with very large radii of curvature attract cyprids although the increased contact compared with a plane surface must be very small indeed. Thirdly, cyprids avoid sharp angles and prominences, just as they seek out depressions.

The junction between a small barnacle and the substratum offers an internal angle in the vertical plane, but it is strongly convex in the horizontal plane. The angle does not therefore provide a greatly increased degree of physical contact since the cyprid can only make contact tangentially with the small barnacle. The cyprid might instead respond as it would to a convex rather than to a concave surface, avoiding it and settling a little way off. The base of a large barnacle, on the other hand, is much less convex; indeed a species such as Elminius modestus with a sinuous octaradiate outline has eight concavities which are depressed in both planes, and so are highly attractive, often being occupied by a small barnacle in each.

If the reaction of cyprids to contact with other individuals were clearly different from their reaction to an inert surface, the mechanism of territorial behaviour would involve, in addition to the physical response to an obstacle, the specific recognition of other barnacles as their own species and a change of behaviour in consequence. Recognition of their own species has been clearly shown by Knight-Jones (1953) to exist in barnacles. Knight-Jones & Moyse (1960) also find that, while the cyprids of certain barnacles avoid contact with individuals of their own species, they settle directly in contact with barnacles of other species. This clearly implies that specific recognition of the surface of their own species is the basis not only of gregarious but also of territorial behaviour. Some of the observations reported in this paper lend support to the view that the three species studied avoid to some extent settling on the surface of their own species. The sparse colonization of the membranous base suggests a specific response. It is difficult to explain why cyprids do not settle readily into a pit containing young spat without assuming that they recognize them as barnacles. Such a pit, even though it contains a small barnacle, remains essentially a concavity. At the bottom of the pit, where the small barnacle generally lies, a groove is formed between it and the pit wall. When a pit is drilled with a blunt twist drill a similar configuration is produced at the bottom of the pit, but in this case it is of inert material and the cyprids settle readily all round the groove. The similar groove formed by the sides of a barnacle already settled at the bottom of the pit is avoided, however, and later cyprids entering usually settle high up on the walls of the pit. However, pits retaining the basis of a small barnacle, but otherwise empty, seem very attractive. Perhaps, therefore, both a convex surface contour and the presence of a barnacle surface produce the strongest avoiding reaction.

Not all barnacles avoid settling on adults of their own species. Cyprids of Balanus hameri settle predominantly on the sides of the adults, forming in time groups of barnacles growing out in a tree-like form from an older basal individual (Moore, 1935). In the season of settlement of this species the adults are covered with young spat to a greater degree than the Modiolus shells on which they commonly grow. Chelonobia patula, an epizoic barnacle found in America predominantly on Callinectes sapidus also settles heavily on its own adults, not only on the sides but also just in the angles of the opercular aperture. They seem commoner on their own species than on the cuticle of the crab itself. Young individuals of Lepas anatifera and Pollicipes cornucopias settle very close to or even on the stalk of other individuals, though they avoid the valves. In another pedunculate genus, Scalpellum, the complemental male must settle on the valves of hermaphrodite specimens. All these examples are species which are obviously gregarious; it follows that territorial behaviour during settlement is entirely independent of gregariousness. These species also resemble each other in having an isolated or restricted habitat, and in not spreading out over it to the same extent as rock-living balanoids.

The three species investigated in this paper behaved very similarly at settlement, land seemed to possess the same basic mechanism. If soon after alighting the cyprid came in contact with its own species it would continue to explore the surface for a long time. If it did not encounter its own species it was likely to swim off. Towards the end of its exploration, when it was about to settle, contact with its own species caused a renewal of exploring and testing behaviour until the cyprid obtained a more acceptable position nearby, such as an unoccupied depression.

It is necessary to postulate only that contact with its own species stimulates further walking to account both for gregariousness over a wide area, and territorial behaviour at short range. Early in the pre-settlement phase contact with another barnacle would cause the cyprid to walk and explore widely, so that it eventually settled somewhere in the vicinity. Towards the end of settlement, when the activity of the cyprid was mainly confined to halting and pivoting, contact stimulation by its own species would similarly cause it to explore further; but, because of its slower progress, it would not move far before attempting to settle again (Fig. 6).

The only obvious adaptive advantage in spacing out during settlement is to allow the individuals sufficient space for growth. It seems clear that the behaviour of the cyprid during the immediate presettlement phase, when it meticulously explores by random steps, and tests with the antennules an area within 3−4 mm. of its final settling position, is an adaptation to ensure that the immediate vicinity is free of all objects that might interfere with settling or with its later growth. From the point of view of the growth of the individual, the cyprids could with advantage space themselves much farther apart, since they are capable of growing to a diameter of over 1 cm. during the first season. However, if the cyprid were too exacting in its requirements in comparison with cyprids of other species, it would be likely to leave enough space for the latter to grow in between itself and its neighbours. There are three serious disadvantages to any species of barnacle that allows a second species to interpenetrate. The second species would reduce the available surface for the first species to colonize. It would expose the first species to the danger of being crushed or overgrown (Connell, 1959). Finally, if interpenetration were considerable and the first species were crossfertilizing, it could result in a proportion of individuals being isolated and consequently sterile. It is thus clear that the cyprids cannot afford the risk of spacing out sufficiently far apart to give themselves enough space to grow to the maximum size. The optimal distance apart is that which is as near as possible to each other consistent with retaining a reasonable chance of survival to maturity. After a heavy spatfall, therefore, many small barnacles will be too close to survive. However, there are two mechanisms which accommodate to some extent the varying numbers of individuals that survive the spatfall. During growth, an individual which is close-packed can modify its shape from that of a flat cone to that of a tall narrow column, without apparent harm. Perhaps more important, once established, an individual or group of individuals under pressure can slide slowly over the substratum to fill up odd spaces, or to make room for other growing individuals (Crisp, 1960b). Hence it seems likely that the most critical phase for survival occurs during or just after metamorphosis, while the shell plates are not fully formed, hardened, and cemented to the substratum. It is at this time that the impact of another growing individual might be disastrous, since the very small specimen might be crushed rather than pushed along. The interval of time from fixation of the cyprid to the completion of the shell plates and the beginning of growth in girth is in B. balanoides about 1 to $112$ weeks. The rate of growth of fast-growing young barnacles is about 1 mm. per week (Barnes & Powell, 1953; Crisp, 1960a). Hence a spacing out of about 1 mm. from established individuals would allow just enough time for subsequent spat to establish themselves, but would probably make very insecure any individual that attempted to settle in between. Territorial behaviour is therefore to be expected in those dominant forms which spread out flat during growth, covering the substratum to the exclusion of other species. Forms which exist as small isolated groups in unusual habitats are relatively free from competition ; they need not therefore become adapted to monopolize the available surface and so may not need to protect themselves against being squeezed off by spacing out during settlement.

I am indebted to Prof. E. W. Knight-Jones and Mr J. Moyse for allowing me to read their manuscript on intraspecific competition in sedentary marine invertebrates while this paper was still in the press.

The upper pair of photographs show a settlement mainly of Balanus balanoides. These were removed, leaving their bases behind. These can be seen in identical positions in the lower pair of photographs. The newly settled spat tend to avoid the bases (except for the depression in the middle of each base) and settle at the periphery of the base or outside it X and Y are bases without central pits. The left-hand photos include specimens of Elminius modestus in pairs at the top centre and bottom right.