Introduction

The need to understand ecological processes and the dynamics of natural populations becomes more urgent as our utilisation of natural animal and plant resources becomes more intensive. During the last decade there has been reluctant recognition of the limitations inherent in the use of deterministic models to describe population functions of single species. While deterministic models have found wide acceptance in fisheries and game management, few take account of the complex relationships which any species has with its environment, and with other living components of an ecosystem. Consequently the limited success of such models in accurately predicting population changes is hardly surprising. The result has been renewed interest in ecosystem unit modelling and some of the problems we face with respect to aquatic ecosystems have been ably summarised by Mann (1972).

At first sight it might appear illogical to attempt to analyse ecosystem units before situations involving only single species can be resolved satisfactorily, but this is not so. Such analyses focus on the hierarchial structures of ecosystems, the flow of energy within them, the nature and patterns of primary productivity in a given system, the efficiency of energy transfer from one trophic level to the next, and the relative importance of horizontal links for degradation of energy within a trophic level. Rather than just considering numbers of animals or plants such studies can help us assess the nature and magnitude of the extrinsic and stochastic factors which so strongly influence the structure and relative sizes of populations in the real world. They permit us to quantify and evaluate some of the complexities of community structure, and to begin to perceive the mechanisms governing ecosystem stability. They may aid in the detection of quantitative changes in numbers and relationships of ecosystem components, and perhaps enable us to forecast the onset of those long term regular and irregular fluctuations in conditions, which so frequently confound predictions based on relatively simple deterministic population models. Acquisition of such knowledge cannot help but improve our understanding of the ecology and population dynamics of single populations.

The fundamental nature of the differences in the structures of terrestrial and marine ecosystems is still not generally appreciated, partly because experimental and simulation studies of system stability have usually involved terrestrial situations. Stability in marine environments is a concept which has received relatively little attention (Steele, 1974, p. 29), and in any case ‘stability’ is a word used with varying meaning by different authors. By some it has been used to imply equilibrium within an ecosystem, with only minor population size fluctuations among its components; for example, a tropical rain forest climax. Others use the same word but imply persistence of the ecosystem despite large fluctuations in the numerical values of components. Recently Holling (1973) applied the useful and descriptive term resilient to ecosystems which survive despite such large component fluctuations, and this is used here in preference to the term ‘stability’ coined by Smith (1972). Unless otherwise specified, discussion in this paper is concerned with resilience.

The purpose of the present paper is to draw together a number of ideas and findings from several independent lines of research, and to discuss a number of factors with possible bearing on the resilience of those pelagic marine ecosystems which characteristically have large component fluctuations. Before this is done, it is pertinent to consider briefly some modern ideas which relate to natural system stability.

Population Interactions and Ecosystem Stability

May (1972), Smith (1972), Holling (1973) and Poole (1974) have reviewed the theories which attempt to describe in quantitative terms the interactions — either through predation or competition — of two or more natural animal populations. Most ecologists will be quite familiar with the Lotka-Volterra equations (Lotka 1925. Volterra 1928), and the numerous modifications and elaborations of these paired differential equations. The validity of the Lotka-Volterra equations has long been challenged, since the constraints implied in their application are quite unrealistic, and while they generate fairly regular oscillations in purely deterministic simulations, the prey species invariably oscillates to extinction under application of realistic stochastic influences (Bartlett 1957). The model of host-parasitic interactions proposed by Nicholson and Bailey (1935) suffers from similar deficiencies (Poole 1974).

Leslie and Gower (1960) experimented with modified forms of Lotka-type equations in stochastic simulations; their results paralleled in a general way some of the earlier experimental studies by Gause (1934) with Paramecium and the protozoan predator Didinium, and Utida (1957) with the bean weevil Callosobruchus and its parasite Heterospilus prosopidis, in that unprotected prey invariably became extinct, followed by the extinction of the predator. Leslie and Gower (1960) then attempted to simulate that section of Gause's study in page 25 which he provided sediment where some of the prey population could hide from predators. The result of their simulation runs was a significant ‘damping down’ of prey population oscillations as soon as a fraction of the population was protected for any given unit of predation time, and despite additional stochastic influences in the simulation, the system no longer tended to ultimately oscillate to extinction. This simulation differed from the experimental results obtained by Gause. Once the prey not buried in the sediment of the cultures was exterminated, the predator species became extinct, permitting the culture to be repopulated by prey from the sediment.

Protection of a significant fraction of any prey population from predation at any given time, whether through heterogeneous spatial distribution (Smith 1972), defence mechanisms, superior mobility of some prey, or differing predator responses to changing prey densities as considered by Holling (1959), and Griffiths and Holling (1969), appears to be essential for the survival of systems with components with large numerical fluctuations. A second factor which may be of great importance in system persistence is the time lag implicit in predator responses to build-up of prey. The existence of such time lags is implicit in a number of studies, and the concept was explored in quantitative terms by Wangersky and Cunningham (1957). Holling and Ewing (1971) combined the features of both explicit time lags and prey protection into a theoretical model in which prey protection counteracted the destabilising effects of time lag in predator response. The idea of a fraction of the prey population being particularly susceptible to ‘contagious attack’, was developed earlier by Griffiths and Holling (1969).

One of our major problems is that there is still no common philosophical approach to formulation of such hypotheses. Many of the models include the implication that population fluctuations are taking place around some kind of mean value; others (Milne 1957a, 1957b, 1962, and Holling 1973) have suggested that such equilibria may not in fact exist but are the result of the influence of statistical theory on much of our thought. Milne concluded that the upper limit of fluctuations was controlled by the carrying capacity of the environment, and the lower level by simple extinction. There is also a growing belief that the search for models which lead to conditions of neutral stability — as with Leslie and Gower's model — should be abandoned, since such neutral stability seems to occur very rarely in nature (Holling 1973).

Bulmer (1975) investigated phase difference in predator-prey relationships and established a model for the inter-relationships of the ten-year population cycles recognisable in some boreal mammals and ground-nesting birds in Canada. He concluded that, in general, the cycle of a species which fluctuated in such a manner appeared to be driven by another cyclic species which was either the prey or predator of the first species. He also calculated that if there was no density page 26 dependence in the drive species, the predator should cycle about one quarter of a period behind the prey. Density dependence in the driven species could produce two alternative results in this model—decreasing the phase difference for the prey driving a predator cycle, but increasing it for the predator driving a prey cycle. Many have accepted the hypothesis of Lack (1954), that regulation in such systems might result from the predator switching to a second prey species as the first decline in density. Bulmer concluded that this would only bring about a progressive delay in prey cycles, and since this did not seem to occur in the field data at his disposal he concluded that the number of predators was probably much more influential in determining predation level than any switching response. Furthermore, the work by Murdoch (1969) indicated that there was no real proof that threshold switching responses by predators exist, and this viewpoint has been further supported by Steele (1974. pp. 45-56).

Implicit in all these studies is the idea that if we can satisfactorily understand which factors are most important in population interactions between two or more species, we will ultimately be able to expand this knowledge to interactions between whole trophic levels. For example, Dempster (1971, 1975) determined that starvation was the major factor in prey population control in the case of the cinnabar moth Tyria jacobaeae, with predators of the larvae exercising only an incidental effect. Limited mobility of the ‘predators’ (i.e. = herbivorous caterpillars) and spatial heterogeneity in ‘prey’ (i.e. = food plants) distribution appear to provide some degree of protection to the vegetation. Such protection of vegetation is particularly evident in forest ecosystems, where most cycling of elements takes place following leaf fall, through the detritus route (Odum 1971), and totally defoliating attacks by consumers, e.g. migratory locust, are the exception rather than the rule. Crisp (1964) in the section of his book dealing with the nature of grazing in terrestrial ecosystems, estimated that barely 15% of the total standing crop of vegetation was eaten by herbivores in a given growing season, and that most terrestrial vegetation was protected by hard tissues, underground root systems, periodic leaf-fall, and so forth.

The Structure of Pelagic Marine Ecosystems

In pelagic marine ecosystems, however, although time lag phenomena are readily recognised, for example in patterns of grazing by herbivorous zooplankton, protection of a substantial fraction of the pelagic vegetation seems to be nearly non-existent. The significance of this seems to have escaped the attention of most workers. There is ready agreement that phytoplankton seems to be consumed about as fast as it is produced, but in the literature there is relatively little appreciation of the implications of this in terms of system stability and resilience discussed in the first part of this paper.

The evidence that such protection is lacking, is formidable. page 27 Harvey et al. (1935), Wimpenney (1946) and others discussed the inverse relationship between distribution of phytoplankton and distribution of zooplankton. They concluded that this patchiness was caused primarily by intensive grazing, and was not just heterogeneous distribution resulting entirely from current movements and other physical effects. Studies by Menzel (1967), among others, have indicated that grazing by herbivorous copepods is in fact so intense that very little fallout of phytoplankton from upper waters seems to occur in most open oceanic areas. Estimates of sinking rates measured in mixed coastal waters are therefore suspect for application to general open ocean situations. Indeed, Riley (1963) doubted our ability to estimate sinking rates even to the nearest order of magnitude, and suggested that relative comparisons were the best we could hope for. These at least indicated that seasonal changes in sinking rates could occur. Animals of the lower trophic levels in most intermediate waters probably have to rely almost exclusively on detritus in the form of zooplankton faeces and colloidal aggregations if epipelagic autotrophs are eaten about as fast as they are produced. Furthermore the efficiency of energy transfer from one trophic level to the next in the sea, particularly from autotroph to herbivore, seems to be far above the 10% level generally accepted as a working value in terrestrial systems. Data for this conclusion were provided by Petipa et al. (1970), who found that transfer efficiency (measured as ratio of growth to ingestion) of herbivorous copepods in the Black Sea, was as high as 71%. The implications of the earlier discussion now become clear — if protection of a significant fraction of the epipelagic diatom vegetation of the sea does exist, it is far from obvious.

The phytoplankton populations appear to establish numerical ‘capital’ during the spring bloom, and this, together with the ‘interest’ accrued through subsequent reproduction, is steadily eroded through the season by zooplankton grazing until the populations are at a rather low level in middle to late summer. In most areas the pattern is generally repeated with a secondary autumnal bloom; after this numbers dwindle to minimal winter levels again.

Steele (1974) argued that heterogeneous spatial distribution of phytoplankton could not alone provide any kind of sufficient protection from grazing, especially since the distributions themselves seem largely to be effect rather than cause. Some have suggested that species diversity and predator switching responses may function together as protection in these systems. MacArthur (1955) argued elegantly that diversity can enhance stability in ecosystems, although he was really only considering the influence of diversity within essentially stable systems, rather than oscillating systems such as the pelagic north temperate marine environment. Smith (1972) developed an argument that both system stability and species diversity were primarily products of spatial heterogeneity, as shown in fig. 1. But more recently Goodman (1975) has concluded that no simple relationship page 28

Fig. 1: Suggested relationships among spatial heterogeneity, stability of a system, and species diversity in an ecosystem (slightly modified after Smith, 1963, and redrawn).

between diversity and stability exists in ecological systems. Furthermore, as we have seen, although predator switching mechanisms cannot be discounted, proof of their existence is far from convincing. Under such intensive predation pressures, and the considerable stochastic influences found in temperate seas, what prevents heavily grazed populations from oscillating to extinction in a relatively short time?

In his simulation of phytoplankton-zooplankton interactions in a marine system Steele (1974) was able to obtain persistence of the system for a 360-day period. He admitted, nevertheless, that this could only be done by assuming some generally not well-validated threshold responses at the herbivore level, and by incorporating some unrealistic parameters such as long-lived components and relatively low food chain efficiencies. In addition, Smith (1972) had already pointed out that translating a realistic approximation of spatial heterogeneity into a program to support such simulations is a most formidable task.

Steele concluded that unlike the situation in terrestrial ecosystems (Hairston et al. 1960), herbivores in pelagic marine systems are resource-limited. The available evidence strongly supports this view. Neglecting for the moment the influence of zooplankton, after the phytoplankton population reaches a certain level, the limiting factors will come into play. Probably the most important of these is the exhaustion of available nutrients in the immediate vicinity of the cells, or a change subsequent to this in the level of available nutrient for such processes within the cell (Laws 1975). Riley (1963) argued strongly that in his view this interplay between nutrients and phytoplankton page 29 was the only guarantee of any stability in the system, and that only the physical factors were basically casual. During the initial bloom of phytoplankton in the western North Atlantic he found a near-linear relationship between photosynthesis and incident radiation, and the subsequent reduction in primary productivity was correlated with depletion of phosphate in surface waters.

The balance of evidence suggests that grazing pressure in the pelagic marine system exerts a much more significant effect on the vegetation than in an analogous terrestrial situation. While grazing by zooplankters begins (or increases) immediately, it is unlikely that the density of overwintering zooplankters in temperate surface waters will be high (Raymont 1963). Nevertheless these populations increase rapidly after a measurable time lag. The decrease in phytoplankton concentration some weeks after the spring bloom is, at least in British waters, almost certainly associated with grazing by this increased population and not with an immediate reduction of light intensity or of available nutrients (Cushing 1958, Steele 1958). Steele found that the decrease in phytoplankton abundance occurred significantly earlier than any reduction in phosphate levels, and in a later paper (Steele 1961), he determined that quite large fluctuations in incident light levels in the North Sea had little effect on the populations of phytoplankton. It has also been suggested that there might be an adjustment by such populations to any particular light intensity, so that the rate of productivity and the nutrient supply are in balance (Ryther and Yentsch 1958). This seems somewhat less likely in view of Steele's findings concerning the lack of influence of variation in light levels. Both Steele and these workers found that production was limited by nutrient supply, but appeared to be independent of the concentration above a certain level (0.4 μg atoms P/litre according to Steele, 1958).

Riley (1963) was unable to accept the conclusion that the concentration of nutrients, and in particular phosphate, was not an important factor. If this was the case, he argued, then there would be no theoretical reason for the system to achieve a steady state; there would only be a maximum determined by the total carrying capacity. Yet the very essence of the arguments put forward by Milne (1957a, 1957b, 1962). Dempster (1975), and Holling (1973), seems to be that the idea of a ‘steady state’ is probably illusory. Nevertheless the emphatic expression of belief by Riley that nutrient concentrations are important determinative factors, is almost certainly justified. In practice the viewpoints are probably not irreconcilable, and for concentrations of phosphate below 0.4 μg, the findings of Riley and Steele are in agreement.

There remains, however, a basic conflict in the data obtained by Riley for George's Bank, and Steele, Cushing and others for European waters with respect to the relative importance of grazing by zooplankton in reducing phytoplankton numbers. Riley maintained that page 30 nutrient depletion and reduction in water transparency were of great importance, but the British workers appeared to have satisfactorily eliminated these as primary reasons for the decrease in phytoplankton in the areas studied by them. Needless to say, the oceanography of the North Sea and the George's Bank region are very dissimilar, with great differences particularly in vertical stability and mixing patterns. For the present I will adhere to the conclusion that in most regions of the temperate seas, grazing by zooplankton is primarily responsible for decrease in phytoplankton numbers.

Diatoms in the Pelagic Marine Vegetation

While the proportion of diatoms to other phytoplanktonic organisms of similar size may vary from place to place — for example, in tropical Atlantic samples dinoflagellates may predominate — they usually form by far the largest fraction of the ‘phytomass’ of 5 μ and above in northern temperate surface waters.

That diatoms possess a rather unique pectin-silica test is well known, although all functions of this structure are not understood. Streaming of cytoplasm through a complex and elaborate system of pores, for example, appears to aid in locomotion (Fritsch 1965). Porter (1976) found that colonies of the green alga Sphaerocystis schroeteri were sometimes only damaged by passage through the gut of zooplankters, and the colonies were repaired by rapid cell replacement. In fact growth appeared to be enhanced by nutrients taken up during passage of the gut. The test of diatoms might give protection against digestion under some circumstances. To the best of my knowledge this has never been tested experimentally, but if this did occur it would have to be taken into account in estimates of grazing, and might even account for certain anomalous observations in the literature which need not be pursued here.

One view of the unique mode of division by diatoms into pairs of daughter cells, one of which is smaller than its kin, is that it is imposed on them by their morphology, i.e. that it is the price they pay for having the rigid pectin-silica test. I wish to propose an entirely different view, namely that reduction in size plays an important role in the basic reproduction and survival strategy of these species.

Diatom cells vary within the size range of 2 X 10^-2 mm³ to 2 X 10^-8 mm³ (Harvey 1950). There are size differences within populations and, more often, between populations of the same species. The typical asexual reproduction of diatoms by paired daughter cell formation with consequent reduction in modal size at each stage, was amply documented by Wimpenny (1936, 1946). Harvey (1950) noted that the reduction in cell volume in Didylum after a series of divisions could be as much as thirty fold. At any given time there seems to be a recognisable modal size in any specific population of diatoms (Wimpenny 1946; Lucas and Stubbings 1948). The approximate synchrony in such populations permitted Cushing (1953) page 31 to use decline in mean size as a measure of the division rate. ‘Stocks’ of diatoms have been recognised in different areas of the North Sea (Raymont 1963). Growth rates would appear to be controlled in a rather complex way by environmental factors such as the rate of nitrate input to the cell and the value of the extinction coefficient coupled with specific light intensity in an area (Parsons and Takahashi 1973), and the minimum size seems likely to be controlled by the ability of individual cells to counter metabolically the respiration losses resulting from increasing surface to volume ratio (Laws 1975). Reduction in size does not proceed indefinitely, but is interrupted by auxospore formation, during which cell contents escape from the siliceous test and regain or exceed the original size. Sometimes auxospore formation is preceded by sexual reproduction (Chadefaud and Emberger 1960). Geitler (1932) found that auxospore formation in Navicula occurred when a critical minimum size of 8.5 μ was attained. It is fair to point out that reduction in size with division is a general rule, but that exceptions have been documented. In particular maintenance of daughter cell size in some culture strains of Nitzschia by enlargement through rapid intussusception has been observed (Fritsch 1965). This author noted that auxospores were actually quite rare in natural samples of phytoplankton, and cited other earlier workers' estimates of the relatively slow rate at which minimum size might usually be reached by diatoms; it seemed that auxospore formation would not be expected to occur more frequently than about once every two years in any particular individual lineage. Other workers (summarised by Raymont 1963), on the other hand, have found quite rapid reduction in modal size of populations within a single season.

Reproductive rates in diatom species vary considerably, but in Phaeodactylum they may be as high as one division every 24-36 hours under favourable natural conditions. Under less favourable circumstances the division rate can fall as low as one division every 18+ days (Raymont and Adams 1958). Even if grazing success can approach 100% in localised areas, it is obviously of vital importance to a diatom species that its reproductive rate be sufficient to cope with the worst possible combination of grazing pressure and unfavourable environmental conditions.

Change of Particle Size as a Possible Factor in Protection of the Diatom Fraction of the Pelagic Marine Vegetation

Since they have contagious distribution patterns, lack any significant form of habitat protection, and are subjected to progressively more intensive grazing by copepods through the summer, diatom species evolving under such feeding pressures might be expected to respond through increasing productivity. Since the supply of nutrients becomes limiting with respect to increase in biomass after a relatively page 32 short period of time, the only way productivity can be maintained or increased is by maximising for numbers rather than biomass, with reduction of mean population size at each division. This conserves resources, and would seem to be the only suitable reproductive strategy open to a diatom species in such a situation.

The rapid rate of division of most species under normal conditions will result in a significant reduction in mean particle size of the population in a relativel yshort time. The possible importance of this has not yet been fully explored, and it may well confer an important additional benefit. Brooks and Dodson (1965), and Kerr (1974), among others, have pointed out that relative size can be an important determinant in predator efficiency, and that observed size relationships are the result of size-dependent feedback between predator and prey.

Brooks and Dodson (1965) carried out a detailed study of feeding relationships between alewives (Alosa), zooplankton, and phytoplankton, in Crystal Lake, northern Connecticut. Their findings concerning the influence of size selectivity by predators can be briefly summarised as follows. When predation by alewives was moderate, both small and large zooplankters were common in their samples. When such predation was light, small plankters were competitively eliminated by large forms. The latter were presumed to be more efficient through possession of larger filtering surfaces because of a more favourable surface to volume ratio, giving reduced metabolic demands. The former were thought to have to work proportionately harder to resist sinking. When predation by alewives was intense, on the other hand, large zooplankters were selectively eliminated, and the smaller species, less attractive to fish because of their small particle size, predominated in samples.

From the energetic standpoint these authors believed that, all other things being equal, selection should tend to favour the predator with a feeding strategy that operated to take small numbers of larger particles rather than a large number of small particles. They suggested that whether or not a population was being eliminated depended on the average size of the smallest female instar which could produce viable eggs being below the particle size range exploited by the alewives. At this critical level they thought that not only particle size, but spatial distribution and escape movements might be of crucial marginal significance.

They also concluded that selection probably would not favour rigorous apportioning of food to body size, and pointed out that many congeneric zooplankters were of roughly similar size and were presumably of similar efficiency in food collecting. They suggested that all planktonic herbivores utilised small particles in the 1-15 μ range.

I believe, however, that where feeding of zooplankton on phytoplankton is concerned the specific range of particle size is far more page 33 critical in many cases than they imply. It is also worth mentioning that the majority of pelagic diatoms species appear to fall within size ranges of about 15-140 μ X 2.25 μ (Pascher 1930, Fritsch 1965), although it is difficult to obtain accurate quantitative estimates. I also have considerable reservations about accepting their findings as being applicable to a very large open ecosystem such as the temperate North Atlantic, even though valid for relatively small land-locked lakes. I suspect that the interplay of spatial and temporal heterogeneity in the temperate oceans is too great to permit such a tidy relationship to survive for long. Recruitment of fauna and flora into areas as a result of current movements, mixing, and seasonal upwelling, is probably over-riding in most regions. Elimination of food particles of a given size range in any one area would obviously be strictly temporary. Nevertheless, such temporary loss of availability of food particles in a specific range in an area will have far-eraching effects on the feeding strategies, and consequently feeding success, of whole groups of predators with similar strategies.

Kerr (1974) published a theoretical model based on trophic processes, and using the K-line concept developed by Paloheimo and Dickie (1966), concluded that prey and predator sizes are rather simply related, and that in living systems particle size was a far more important factor in predator grazing than particle density. Some important experimental studies giving insight into gain from grazing under different particle size regimes were carried out by Beamish and Dickie (1967) and Parsons and LeBrasseur (1970). When herbivorous pelagic crustaceans (copepods and euphausid furcilia) were fed with algae with individual size ranges having a modal value of 32 μ a weight gain of only 2% per day was noted; however, when the modal size of algae was in the 57-90 μ range the weight gain rose to 16-8%. It would seem, therefore, that successful utilisation of a prey species by a predator may be limited by a relatively small range of particle size. At one end of the scale the prey will be too large for the predator to easily manipulate it (Mann 1972), and at the other end of the scale be so small that not only might there be manipulation problems, but the predator's energy gain for energy cost through foraging and collection will become progressively smaller. Rapport and Turner (1975) recently published a theoretical consideration of feeding strategies; they concluded that as resources became limiting, different types of feeding strategy converged, as did respective consumption rates. Nevertheless, the available range of particles of a size which can be consumed with net energy gain must surely place constraints on just how far such convergence could go. Smith (1972) also considered the impact of relative catchability; the feeding efforts of a predator locally reducing prey numbers to a low level could well reduce the average catchability of the remaining prey, with profound effects not only on its own feeding strategy, but also on those of other predators seeking the same prey.

With respect to a specific predator or consumer, relatively rapid reduction of the individual particle size of a significant fraction of a food species population will affect first the feeding success, then shortly the feeding strategy of the consumer. A number of workers (as summarised by Steele 1974) have searched for switching responses among herbivorous zooplankton, to see if some kind of threshold was involved. The present author suggests that there may be no complex mechanism involved at all, and that simple reduction of availability of a favoured diatom species by a statistical shift of its population out of the optimal feeding particle size range characteristic of the consumer in question, will force upon the latter a change to alternative prey within that same optimal particle size range. While size is probably the most important single factor, it is clearly unlikely to be the only one involved; metabolite production by certain phytoplankton could deter zooplankters from feeding on them, and shape is probably also very important (Harvey 1937). The various larval and adult stages of zooplankters will all have different optimal particle size requirements. Marshall and Orr (1955) showed that Calanus finmarchicus will take a wide variety of diatom species if these are offered experimentally, but because of the effect of extra foraging time or handling time under limiting circumstances, a copepod is likely to optimalise its feeding strategy by taking food particles within a relatively narrow size and shape range. The successful exception to this might occur when small particles were locally dense enough that economical capture could take place. The catholic diet of Calanus finmarchicus is probably a factor in its success, since it certainly seems to be far more abundant in the boreal-temperate North Atlantic than many rather similar species. Ability to exploit a wide range of food species surely must statistically increase the chances of a predator finding food particles of optimum size.

Reduction of the modal value of particle size in the diatom population will, of course, not only function to remove it from the optimal feeding strategy zone of the first consumer, but also to expose the population sequentially to smaller consumers. Short generation time and consequent rapid reduction in size are likely to be strongly selected for, since this would reduce the time period of such exposures, and take advantage of any time lag in feeding reaction by the consumer to the newly available food source. While Hutchinson (1961) pointed out that few opportunities exist for simple physical niche diversification in turbulent open water, the concept of niche response surface, as examined by such authors as Makarewicz and Likens (1975), gives us a different view of the situation. By their definition, ‘niche’ variables may be considered as axes of an n-dimensional coordinate system defining the niche hyperspace. That part of the hyperspace occupied by or affected by a given species represents that species' niche hypervolume. Time, or duration, is certainly one of these axes, and by looking at the different particle page 35 size reqiurements of the various larval and adult stages of a zooplankter it is possible to see that several, even many, species can co-exist so that at no given instant of time need any significant interspecific competition be postulated. While I agree with Smith (1972) that the stabilising effect of spatial heterogeneity is probably very powerful, I also concur with Steele (1974) that simple spatial heterogeneity of phytoplankter distribution is not enough, in all probability, to account for protection of the primary producers from consumption. Nevertheless, it is unlikely that the populations of different species of diatom are totally sympatric. This would mean that localised rearrangements of consumer populations would have to occur before they were in a position to exploit the new food supply to the full. Concentration of a significant population of zooplankters around a phytoplankton patch must take at least several days; and the zooplankton swarm, once constituted, takes about one week to effect a measurable decrease in the density of the phytoplankton (Steele 1961). During this period of about two weeks several divisions would have occurred under normal circumstances with the result that measurable reduction in the modal diameter of the diatom population would have begun (at least in most species), with its previously discussed implications. While there have been many quantitative distributional studies of phytoplankton patches prior to grazing, and even during the process of being grazed (generally by the study of experimental cultures), the nature of the grazing process in the wild is little known. Smith's (1972) discussion on catchability, and particularly the change in catchability brought about by the effect of biased removal of prey by predation, is very relevant, and offers a promising lines of research.

Even if the auxospore sometimes functions as an over-wintering stage as some authors believe, and is often involved in the onset of sexual reproduction (Chadefaud and Emberger 1960), it also serves the purpose of a kind of ‘quantum jump’ in size to something approximating the original size mode of the diatom population. This most radical step in the life cycle not only generally occurs at the end of summer when resources are near depletion (so that relative ‘rarity’, i.e. low population density, of auxospores is not surprising), but also at a time when the population has been under sustained consumer pressure. Since zooplankter numbers are also decreasing as a result of predation, natural death of adults at the completion of life cycles, and decreasing production, such a particle size change might generate greater time lag in possible consumer response than those of smaller magnitude which occur during the spring and summer modal reductions. Escape through a sinking reaction may also become a significant factor at this time but field data are lacking. In some species microspore swarmers are formed instead of auxospores; but here again a significant change of particle size occurs.

The reproductive strategy of diatoms therefore appears to be page 36 geared to the relatively short season of optimal nutrient availability, and maximisation for numbers as these nutrients are depleted. As a result, the strategy also serves to moderate effectively the impact of high densities of consumers on any given diatom species. Since these populations persist in an environment with wide fluctuations in day-to-day and seasonal conditions, this strategy seems to be eminently successful. The very efficiency of consumers in the lower trophic levels of the pelagic marine environment (Petipa et al. 1970) would probably mitigate against the success of any other type of strategy.

Conclusion

Protection of the ‘vegetation’ of the pelagic marine ecosystem from total consumption by consumers is essentially a dynamic process, quite unlike the static protection found in terrestrial systems. Maximisation for numbers as resources become limiting, rather than size, is a logical response in a system where formation of perennial tissues is virtually unknown beyond the sublittoral zone, and this reproductive strategy confers the important benefit of repeatedly generating time lags in consumer response through particle size change of the producer species. There has also been selection for rapid generation time; this permits a diatom population to recover rapidly as soon as nutrient levels are adequate, despite previous heavy losses from grazing.

It is fair to ask why the consumer species have not undergone similar rigorous selection to place them fully in phase with the diatom populations. The answer is probably simple; stochastic influences, such as fluctuations in the availability of nutrients from one area to another in temperate seas, and variation in the strength and direction of ocean currents, so affect the distribution of diatoms and the size ranges and division rates of local populations, that consumer response lags are inevitable. Additionally, the reproductive and survival strategy of the consumer does not necessarily have to include total gearing to that of the food species. Other factors are important; increase in individual size of copepods and similar zooplankters as their life cycle progresses has its own implications for energy utilisation, but does not permit tight phasing with food species which have a decerasing modal size. Furthermore, larger body size in consumers is generally accompanied by longer reproductive period.

Nevertheless the reproductive strategy of copepods and other marine arthropods, considering their undoubted biological success, might also bear some scrutiny in respect to particle size change during the life cycle. These species also undergo significant particle size change in discrete jumps at their moult stages, albeit with an increase in size.

The only group in the lower trophic levels in which some kind of particle size change does not occur during the life cycle appears to be the flagellates, except that the mode of cell division in many page 37 dinoflagellates frequently results in the formation of chains of adhering individuals (Sverdrup et al. 1946). It may be significant that many flagellates are also known to produce toxic metabolites, and the dinoflagellates include a large number of bizarre spinous forms, especially in the tropics, which may make their handling by potential consumers difficult, and result in lowered feeding efficiency. There seems a growing realisation that particle size relationships are of crucial and fundamental importance in marine ecosystems, and that these systems function, and survive, through strategies and relation-ships which have no direct parallel in terrestrial ecosystems.

References

Bartlett, M. S., 1957: Biometrika, 44: 27-42.

Brooks, J. L., and Dodson. S. I., 1965: Science, 150: 28-35

Bulmer, M. G., 1975: J. Animal Ecol. 44: 609-622

Chadefaud, M., and Emberger, L., 1960: Traite de Botanique systematique, Tome I, Masson et Cie, Paris, 1016 pp.

Crisp, D. J., 1964: Introduction in Grazing in Terrestrial and Marine Environments, pp. xi-xvi, edited by D. J. Crisp, Blackwell Scientific Publications, Oxford.

Cashing, D. H., 1953: J. Cons. int. Explor. Mer., 19: 3-22.

——, 1958: Rapp. Cons. Explor. Mer., 144: 149-154.

Dempster, J. P., 1971: Entomol. Gaz. 22: 199-204.

——, 1975: Animal Population Ecology. Academic Press. London, 155 pp.

Fritsch, F. E., 1965: The Structure and Reproduction of the Algae, Vol. 1. Cambridge University Press, 791 pp.

Gause, G. F., 1934: The Struggle for Existence. Williams and Wilkins, Baltimore, ix + 163 pp.

Geitler, L., 1932: Arch. Protistenk. 78: 1-226.

Goodman, D., 1975: Quart. Rev. Biol. 50: 237-266.

Griffiths, K. J., and Holling. C. S., 1969: Can. Entomol. 101: 785-818.

Hairston, N. G., Smith, F. E., and Slobodkin, L. B., 1960: Amer. Natur. 94: 421-425.

Harvey, H. W., 1937: J. Mar. biol. Ass. U.K. 22: 97-100.

——, 1950: J. Mar. biol. Ass. U.K. 29: 97-137.

——, Cooper, L. H. N., LeBour, M. V., and Russell, F. S., 1935: J. Mar, biol. Ass. U.K. 20: 407-441.

Holling, C. S., 1959: Can. Entomol. 91: 293-320.

——, 1973: Ann. Rev. Ecol. Syst. 4: 1-23.

——, and Ewing. S., 1971: Proc. int. Symp. Statist. Ecol., New Haven, Conn., 2: 207-229.

Kerr. S. R., 1974: J. Fish. Res. Bd. Canada, 31: 1859-1862.

Lack. D. L., 1954: The Natural Regulation of Animal Numbers. Clarendon Press, Oxford, 343 pp.

Laws, E. A., 1975: Ecology 56: 419-426.

Leslie, P. H., and Gower, J. C., 1960: Biometrika 47: 219-234.

Lotka, A. J., 1925: Elements of Physical Biology, Williams and Wilkins, Baltimore.

Lucas, C. E., and Stubbings. H. G., 1948: Hull Bull. Marine Ecol., 2: 133-171.

MacArthur, R. H., 1955: Ecology 36: 533-536.

Makarewicz, J. C., and Likens, G. E., 1975: Science, 190: 1000-1003.

Mann, K. H., 1972: In: Essays in Hydrobiology presented to Leslie Harvey (edited by R. B. Clark and A. J. Wooton). University of Exeter Press, pp. 1-14.

Marsahll, S. M., and Orr, A. P., 1955: The Biology of a Marine Copepod, Calanus finmarchicus. Oliver and Boyd, Edinburgh.

May, R. M., 1971: Math. Biosci., 12: 59-79.

Menzel, D. W., 1967: Deep-Sea Res. 14: 229-238.

Milne, A., 1957a: Canadian Ent. 89: 193-213.

——, 1975b: Cold Spring Harbor Symposium on Quantitative Biology 22: 253-271.

——, 1962: Symp. Soc. exp, Biol. 15: 40-61.

Murdoch, W. W., 1969: Ecol. Monog. 39: 335-354.

Nicholson, A. J., and Bailey, V. A., 1935: Proc. Zool. Soc. Lond., 1935, 551-598.

Odum, E. P., 1971: Fundamentals of Ecology. Saunders, Philadelphia, 3rd edition, 574 pp.

Paloheimo, J. E., and Dickie, L. M., 1966: J. Fish. Res. Bd. Canada 23: 1209-1248.

Pascher, A., 1930: Die Süsswasser-Flora Mitteleuropas, Heft. 10. Edited by F. Hustedt, Rischer, Jena, 466 pp.

Parsons, T. R., and LeBrasseur, R. J., 1970: In: Marine Food Chains. Edited by J. H. Steele. Oliver and Boyd, Edinburgh, pp. 325-343.

Parsons, T. R., and Takahashi, M., 1973: Biological Oceanographic Processes. Pergamon Press, Oxford, 186 pp.

Petipa, T. S., Pavlova, E. V., and Mironov, G. N., 1970: In: Marine Food Chains. Edited by J. H. Steele. Oliver and Boyd, Edinburgh, pp. 143-167.

Poole, R. W., 1974: An Introduction to Quantitative Ecology. McGraw-Hill, New York, 532 pp.

Rapport, D. J., and Turner, J. E., 1975: J. Theor. Biol. 51: 169-180.

Raymont, J. E. G., 1963: Plankton and Productivity in the Oceans. Pergamon Press, Oxford, 660 pp.

——, and Adams, M. N. E.: Limnol, and Oceanogr. 3: 119-136.

Riley, G. A., 1963: In: The Sea. Edited by M. N. Hill. Wiley Interscience Publications, New York, pp. 438-463.

Ryther, J. H., and Yentsch, C. S., 1958: Limnol. and Oceanogr. 3: 327-335.

Smith, F. E., 1972: In: Growth by Intussusception. Edited by E. S. Deevey Trans. Conn. Acad. Arts Sci. 44: 309-335.

Steele, J. H., 1958: Scot. Home Dep. Mar. Res., 1958 (7), 1-36.

——, 1961: In: Oceanography. American Association for the Advancement of Science 67: 519-538.

——, 1968: Mem. 1st Ital. idrobiol. dott Marco de March 1, Pallanza, Italy, 18 (suppl. 2), 383-398.

——, 1974: The Structure of Marine Ecosystems. Blackwell Scientific Publications, Oxford, 128 pp.

Sverdup, H. U., Johnson, M. W., and Fleming, R. H., 1946: The Oceans, Their Physics, Chemistry and General Biology. Prentice-Hall, New York, 1087 pp.

Utida, A., 1957: Cold Spring Harbor Symposium on Quantitative Biology 22: 139-152.

Volterra, V., 1928: J. Cons. int. Explor, Mer. 3: 1-51.

Wangersky, P. J., and Cunningham, W. J., 1957: Ecology 38: 136-139.

Wimpenny, R. S., 1936: J. Mar. biol. Ass. U.K. 21: 29-60.

——, 1946: J. Mar. biol. Ass. U.K. 26: 271-284.

Acknowledgements

The manuscript of this paper was prepared during my sabbatical leave of December 1976 - April 1977 in the Department of Zoology, University of Canterbury. Thanks are extended to Professor George Knox and his colleagues for providing a peaceful and congenial environment. I would also like to express my thanks to Dr L. M. page 39 Dickie, Chairman, Department of Oceanography, Dalhousie University, Halifax, Nova Scotia, and Drs F. W. H. Beamish, D. M. Lavigne, D. J. Milne and G. J. D. Smith of my own department for reading and criticising drafts, and providing much-needed encouragement.