Blute, M. (2002). The Evolutionary Ecology of  Science.
Journal of Memetics - Evolutionary Models of Information Transmission, 7.
http://cfpm.org/jom-emit/2003/vol7/blute_m.html

---

Abstract

1 - Introduction: Sociocultural/Memetic Evolution, Science and Evolutionary Ecology

2 - Density Dependence: Primary Versus Secondary Research/Teaching

3 - Density Dependence: Primary Research/Teaching Versus Moving or Innovating

4 - Frequency Dependence: Avoiding Competition

5 - Scale Dependence: Fast Specialists Versus Longer, Slower Generalists

6 - Heterogeneous Environments: Flexible Strategies

7 - Conclusion

Notes

Appendix: r and K Selection Revisited

Acknowledgements

References

Abstract

In the study of sociocultural/memetic evolution, approaches modelled on population genetics (e.g. Cavalli-Sforza and Feldman 1981) and on systematics/taxonomy (e.g. Hull 1988) have been prominent but the influence of evolutionary ecology has been slight. In the tradition of Toulmin (1972) and Hull (1988) this paper is about cultural evolution in science. In particular, it applies some principles of evolutionary ecology to the scientific process. The effects of density, scale, frequency and heterogeneity on various strategies of research and teaching in science are considered. In the future, the analysis should be extended to the sociobiology of science including the relationship between supervisors and their graduate and postdoctoral students and among peers e.g. publishing and other forms of social collaboration.

Keywords: sociocultural evolution, cultural evolution, memetics, sociology of science, philosophy of science, science studies, evolutionary ecology

1 Introduction: Sociocultural/Memetic Evolution, Science and Evolutionary Ecology

Social scientists and even humanists in a variety of fields have been getting on with it. These include not only epistemology (Campbell 1988: Part V, Plotkin 1994, Buskes 1998), anthropology (Sperber 1996, Cullen 2000, Ehrlich 2000) and sociology (Runciman 1989, Lenski et. al. 1994, Fog 1999) generally, but also more specific fields such as institutional economics (Boulding 1981, Nelson and Winter 1982), linguistics (Ruhlen 1994, Croft 2000), organizations (Aldrich 1999) and technology studies (Basalla 1988, Ziman 2000) in addition to memetics of course (Brodie 1996, Lynch 1996, Blackmore 1999 and Aunger 2000) to mention only some books.

On the topic of sociocultural evolution in science in particular which is the subject of this paper, Darwinian-style evolutionary approaches emerged more from the philosophy than from the sociology of science. Although Popper's (1963) falsificationism was evolutionary at least in the negative sense of selection against, in retrospect, it seems almost inevitable that once Kuhn (1962) started his revolution making philosophers of science more historically and sociologically aware, that a full-blown Darwinian-style evolutionary approach to science would emerge. The pioneer was Stephen Toulmin (1972) but eventually, more than anyone else, it was developed by David Hull. Hull's sophisticated Science as a Process (1988) is at once a treatise on the philosophy of science, the sociology of science, and evolutionary theory (whether instantiated biologically or socioculturally). Not immodestly, but accurately in my opinion, Hull views himself as the heir of Kuhn and Toulmin. The major elements of his evolutionary theory of science are "curiosity" (mutation), "credit" (descent) and "checking" (selection). While this paper is in the tradition of Hull's theory, there are some differences in emphasis. It places an equal emphasis on the acquisition of knowledge in science and its dissemination, it deals almost exclusively with dissemination through graduate and postdoctoral students rather than through peers by means of paper publishing etc. (the sociobiology of science is beyond the scope of this paper), and emphasizes more selection over descent (the reason for which follows).

In any evolutionary process, the currently existing array of forms is explained by constraints (most obviously the laws of physics and chemistry but there may be others), chance (mutation and sampling error in finite populations), history and selection. The three most important research programmes in evolutionary biology are population genetics (which deals with some of these formally), taxonomy which emphasizes history by building trees, and evolutionary ecology which emphasizes the forces of, and responses to, selection. In previous work on sociocultural evolution, approaches modelled on population genetics (particularly Cavalli-Sforza and Feldman 1981) and on taxonomy (particularly Hull, not surprisingly given that the science he studied was taxonomy) have been prominent but the influence of evolutionary ecology has been slight. While this paper is about cultural evolution in science then, it is particularly about evolution in science under different ecological conditions.

Darwin's theory of evolution was a theory 'that' natural selection, rather than 'of' natural selection. The Origin of Species contains no general theory of the circumstances under which selection favours this or that property of organisms. Evolutionary ecology is the discipline that attempts to fill this gap (Cockburn 1991, Bulmer 1994, Fox, Roff and Fairbairn 2001) see also the relevant parts of general ecology texts including (Begon, Harper and Townsend 1996, Pianka 1999, and Ricklefs and Miller 2000). Evolutionary ecology seeks a theoretical halfway house between the near-universal tautology of the fitness-selection nexus and the near-complete historical specificity of the myriad details of what is adaptive in locally prevailing circumstances. Its research programme is in turn divided into a number of subbranches, each of which combines evolution and ecology with some other traditional field of biological study: physiology (Sibly and Calow 1986), development i.e. life histories (Roff 1992, Stearns 1992, Charnov 1993), behaviour (Krebs and Davies 1993,  Krebs and Davies 1997) including foraging (Stephens and Krebs 1986), and extends even beyond populations to metapopulations (Rhodes, Chesser and Smith 1996, Hanski and Gilpin 1997, Hanski 1999). The appropriate terminology becomes long as in evolutionary physiological (or developmental or behavioural or metapopulational) ecology. In addition, an intersecting set of divisions pertain to the group studied microbial (Andrews 1991), plant (Bazzaz 1996), animal, and human (Smith and Winterhalder 1992). Despite this complexity, it is often recognized that similar principles should apply to all divisions (Andrews 1991) and at all scales (Bazzaz 1996).

The most important ecological conditions which control evolution are density (low versus high discussed in sections two and three) and scale (small versus large discussed in section five). These reflect the optimization approach (in the narrow sense) to evolutionary ecology. In both cases, alternatively or in addition, negative frequency dependence may obtain (it being advantageous to do the opposite of what others are doing). This is the game theory approach discussed with respect to density dependence in section four and with respect to scale dependence included in section five. In both cases as well, heterogeneity of conditions may obtain as discussed in section six. In all cases, the evolutionary-ecological theoretical approach is to ask, and try to answer logically, "Under alternative ecological conditions, what adaptations, or 'strategies' as they are called, would selection favour?" Given that the influence of evolutionary ecology on the study of sociocultural evolution and memetics has been slight as noted, the concepts employed may be largely unfamiliar to readers of this paper. If so, it might be helpful to follow the summaries in Table 1a (on density dependence and its complications) and in Table 1b (on scale dependence and its complications) as you go. For more interested or knowledgeable readers, a discussion of how the evolutionary ecology employed here differs from others where it does is included in the appendix.

Table 1. Some Principles Of The Evolutionary Ecology Of Science [note 2]

2 Density Dependence: Primary Versus Secondary Research/Teaching

There are many kinds of scientific research. New empirical research can yield knowledge, new theories, understanding, new combinations of these, explanations and predictions. New methods can alter the ways in which any of these kinds of research are conducted. Whatever classification of kinds of research one prefers, the analysis here is intended to apply to all of them but for convenience, the word "knowledge" will be used to apply to all. It has been a fundamental principle of the sociology of science since Merton (1973) virtually invented the subject, that scientists compete for status rather than wealth, income or power. They compete to have their work incorporated into that of others (through their direct cultural heirs -- graduate and postdoctoral students as well as indirectly through peers via publications with their influence or ancestry acknowledged in citations). In Hull's terminology, they compete for "credit". Of course just as organisms must grow and develop in order to reproduce, scientists must acquire knowledge i.e. do research in order to have something to communicate and receive credit for having done so. Imagine then members of a population of scientists competing to acquire knowledge from nature by means of research on a new topic or in a new field -- a low density or uncrowded environment. Because the topic or field is new, none of the competitors know very much and a whole world lies waiting to be discovered. Given the objective of maximizing personal knowledge, when under such benign conditions it pays to engage in as much primary research as possible, data is collected empirically, and axioms or postulates (or in multi-paradigm sciences theories themselves) are added. Complex inputs must additionally be analysed (broken down). In empirical research outliers must be identified, statistically significant results distinguished from the insignificant, repeatable results distinguished from the non-repeatable, and so on. Similarly in theoretical research analysis is necessary. Not only are axioms or postulates required, but theorems must be proven. As the topic or field matures however, competitors already possess a lot of knowledge and there is not much new left to be discovered. That is not the end of the task however as has sometimes been thought (Horgan 1996). Knowledge must be maintained by what has sometimes been called "secondary research". The task of synthesis is primarily one of assimilating the new into the context of old ("adding new bricks to the wall" as the expression has it) but the positively misleading must be weeded out as well. At least three studies have shown that even articles which have been retracted continue to be cited (Campanario 2000), an extreme example of the necessity of such work. Other kinds of "repairs" are required -- for example errors corrected and findings rearranged in relation to one another. Much of the work of secondary research is performed by referees and writers of review articles, meta-analyses, textbooks, and other "secondary" literature with the goal of building durable knowledge. In science, as in life, low densities favour spending on growth particularly by acquisition and analysis; high densities favour investing in maintenance particularly by synthesis and rejection.

What applies individually in evolutionary ecology applies similarly demographically. When population density (measured now in numbers) is low relative to resources [note 6], selection favours spending on maximizing r (the growth rate of the population measured in numbers) by producing as many offspring as possible and "breaking down" by assessing quality. When density is high, selection favours investing in maintaining the population at or near the carrying capacity of the environment, K, by means of intermediary (biosynthetic) metabolism -- "building up" fewer offspring with parental care and "excreting" by practising active discrimination in favour of those assessed to be of high rather than low quality. In short, low densities favour growth by quantity and assessment; high densities favour maintenance by quality and active discrimination. Again imagine members of a population of scientists competing in a low density environment, measured now in numbers. Competitors working on the new topic or field are few and a whole world lies waiting to be populated with researchers. Given the objective of maximizing public knowledge (Ziman 1968), under such benign conditions it pays to spend on what might be called "primary" teaching -- training as many graduate and post-doctoral students as possible and assessing quality. As the topic or field matures however, competitors are numerous and there is little room for more. Under such harsh conditions, it pays to invest in what might be called "secondary" teaching. Demographic success is better served by mentoring fewer students and pruning the "brood" to improve quality.

In summary, when densities are low, selection favours spending on maximizing r by acquiring knowledge and analysing it (primary research) and by teaching and assessing large numbers of graduate and postdoctoral students (primary teaching). When densities are high, it favours investing in maintaining K by synthesizing knowledge and weeding out inaccuracies (secondary research) and mentoring fewer students along with active discrimination (secondary teaching). Note that neither kind of strategy is intrinsically better than the other, rather, both should be similarly successful in their appropriate circumstances. A population that begins small and grows in size or numbers to become large and one which, already large, manages to remain so rather than declining are, at equal costs, equally successful. While often not so considered, scientists who add and those who maintain personal or public knowledge are contributing equally.

3 Density Dependence: Primary Research/Teaching Versus Moving or Innovating

In their research and teaching, scientists often have opportunities to cope with diminishing returns to primary research in a topic or field by escaping to a different, more promising one: from molecular genetics to molecular neurobiology, from the war on cancer to the war on aids, from youth in the 60's to the ageing population of the 90's, etc. Individual scientists prepared to translocate knowledge acquired in one context to another can be successful. Hull's own work applying concepts and theories acquired in evolutionary biology to the sociology of science is a case in point. Not only do scientists engage in such strategies, they sometimes prepare their students to do so. For example, many mathematicians, physicists and computer scientists moved into finance a decade or so ago (Chance and Peterson 1999) and some of the former kinds of graduate programmes specifically trained at least some of their students to do so.

In addition to maintenance and motility, high densities should also favour mutability. In 1982, Walter Fitch a pioneer of the study of molecular evolution clearly described and predicted the existence of the phenomena that have come to be known as "adaptive" and "directed" mutation respectively. "The organism might be better off if it could vary its mutation rate upward in stressful times and downward in favourable times ... If the organism needs to change only a few of its genes, one would prefer to increase the mutation rate in those genes specifically ... I predict that a mechanism for just this will be found" (quoted in Hall 1998). In 1988 in Nature and in 1991 in Genetics Cairns et. al. and Cairns and Foster created something of a sensation by reporting just such results for an experimental system in E. coli. Eventually the dust settled and many of the mechanisms even of adaptive although not necessarily directed mutation in bacteria and yeast are understood (for some reviews see Hall 1998, Foster 1999 and Rosenberg 2001). At least two mutator genes (which code for DNA polymerases with high error rates) have homologues in eubacteria, archaea and eukaryotes.

The concept of investing in innovation by means of research and development under conditions of scarcity is well known to economists many of whom see it as the solution to environmental problems. Indeed, it is well known to all of us. In recent decades, rather than acquiring more fossil fuels, we have to some degree squeezed more mileage for our vehicles and heat for our homes per unit of fuel input by increasing the efficiency of cars and furnaces (a response to the energy crisis of the 1970's suggested by Lovins in 1976). Historically, for heating, wood became scarce and expensive. We moved to coal and successively to oil and to natural gas. In the human economic world, normally innovations which increase efficiency or change niches require some investment in research and development. Depending upon how K, the carrying capacity of the environment is defined, "R&D" can be thought of as a strategy for raising carrying capacity or for realizing capacity unutilised for what evolutionists call historical reasons. The necessary innovations were simply not previously available in the population. R&D is not prescient but neither is it entirely random -- it is normally focused on the relevant technology. Similarly, its success is not guaranteed; much of the investment goes for nought, but successes often make it worthwhile. That does not mean however that economic change is not an evolutionary process. For example, Nelson and Winter (1982) and Nelson (1996) include search processes, including for information, in their evolutionary model of economic change. It is not surprising therefore that although risky, natural selection sometimes favours genes or the expression of existing inducible genes that search for genetic solutions to difficult problems or take advantage of opportunities -- genes that, in effect, invest in genetic R&D. It hardly needs to be said that there is nothing "Lamarckian" about the phenomenon of adaptive mutation. Even if somewhat , although obviously never perfectly directed (e.g. toward genes coding for enzymes in a biochemical pathway whose substrate is in short supply or toward those making organisms capable of utilizing a recently unutilised resource) mutator genes themselves must evolve by normal Darwinian processes. In eukaryotes, the logic of the evolution of such genes would be similar to that of genes for recombination i.e. genes which have their effect indirectly through other genes on which they then hitchhike. In the broadest sense, all of science is a form of R&D. In a narrower sense however, it seems likely that, as well as favouring investing in the first "2M's", stalled progress should favour a 3^rd M -- mutability, i.e. individuals with heretical tendencies or the expression of such, disposed towards breaking out of the mould of existing theories, methods, etc., in their research/teaching searching for ways to reach new goals or to break through historical obstacles to achieving existing ones.

4 Frequency Dependence: Avoiding Competition

Given that growth strategies do normally have direct ecological consequences in the form of depletion and degradation, it is not surprising that members of some populations, instead of evolving in response to density, evolve in direct response to the strategies present in the population, i.e. to what the others are doing. Rather than ecological optimization in the narrow sense, the principle is the game-theory one of negative frequency dependence. If some individuals (or all of them some of the time -- either at random or with some fixed probability) are engaging in primary research/teaching, then it pays others to do secondary research/teaching, to change topics or fields, or to innovate and vice-versa. This negative frequency-dependent principle is to avoid competition. Much science involves not so much the kind of intensely competitive race made famous by Watson (1969), as it does precisely avoiding such races: with A studying X, B studying Y, C studying Z etc. However, negative frequency dependence can be added to, rather than simply substituted for, density dependence. Under the simplest set of assumptions, at equal costs, pure density dependence predicts strategies should be present in proportion to density. These range from all growth at one extreme, through equal frequencies at intermediate densities, to all one or more of the 3M's at the other extreme. At equal costs, pure frequency dependence predicts the two should always be present at equal frequencies. At equal costs, combined density and frequency dependence still predicts equal frequencies at intermediate densities, but other equilibria are possible. For example at equal costs, with resources depleted by 20%, growth and maintenance/motility/mutability strategies should be present in a ratio of 2:1 [note 7]. In short, in science as in life, selection should be sensitive not only to the ecological environment but also to the social environment in the sense of what strategies others are pursuing.

5 Scale Dependence: Fast Specialists Versus Longer, Slower Generalists

Resources may be devoted to doing research rapidly but for a shorter period of time in a field, yielding a higher rate (mass per unit time, so to speak) of knowledge. Alternatively, resources may be devoted to doing research at a slower rate but for longer (yielding knowledge over a longer time period). Biologically (but not sociologically), trade-offs between growth rate and time are well known. Mammals that develop more rapidly by many criteria lead faster, but shorter lives. "Live fast, die young" is the rule (Promislow and Harvey 1990). More familiar sociologically is the spatial equivalent. Resources may be devoted to doing research more intensively but in a smaller area of a field, (yielding a higher mass per unit area of knowledge so to speak) i.e. as a "specialist". Alternatively, resources may be devoted to doing research less intensively but spanning a larger area i.e. as a "generalist". Similar distinctions apply to teaching.  Students with graduate degrees may be produced at short "interbirth" intervals or alternatively, at a slower rate but for longer. They may be produced in a single topic or alternatively, teaching effort may be spread more thinly but broadly across a wider range of topics. It was once thought that spatiotemporal properties of organisms including lifespan and size could be attributed to density. Now it seems obvious that they should instead be attributed to environmental scale. Organisms (and scientists) should evolve to be adapted to the spatiotemporal scale of environmental opportunities, whether small (narrow/short) or large (broad/slow but long).

Moreover there is a correlation between spatial and temporal strategies so that smaller organisms (e.g. measured in length) and more specialized tend to have fast life cycles, while the larger (measured similarly) and more generalized tend to have longer, slower life cycles. The reason for such correlations is not obvious. To attribute them to ecology simply moves the need for an explanation one step back. They may be attributable to cytoplasmic and other nongenetic forms of inheritance or to genetic correlations as follows.

Working rapidly (and by analogy) intensively is expensive in energy, i.e. in ATP or the mitochondria that generate it -- in theoretical economics, roughly labour power. Working more slowly, for longer may be more expensive in enzymes (tools), building blocks (inventories of supplies), ribosomes (protein factories), and mRNA (managerial knowledge and expertise) -- all in theoretical economics, roughly capital. Or to choose a technological analogy, Formula 1 racing cars are basically portable gas tanks (and small in area or volume) while vehicles that compete in long distance rallies carry tools, spare parts, repair manuals and so on (and are larger). In short, it may not just take a long time to grow large as has often been thought. It may also require a large area or volume to store all the things required to keep going for a long time. Hence organisms that cytoplasmically inherit more ATP or mitochondria themselves may have a material comparative advantage for both rapid growth and intense specialization. Those that inherit more "capital" may have such an advantage for both longer, slower growth and generalization. Alternatively, the preadaptations may be genetic in which genes for the former complement and mutually favour each other while genes for the latter do the same. As is well known, selection is not just ecologically-dependent, but also own condition-dependent (including, but not exclusively, genetic condition).

Examples of such contrasts and their correlations are easy to come by in the history of science. Contrast contemporary examples of competitions with their narrow objectives to find the "gene for" whatever  (which tend to develop and be concluded rapidly), with Darwin's work on evolution or even Hull's own work. The broad scope of the former is well known. So too is the length of time it took to complete. The scope of Hull's book is also truly vast in it are woven together a treatment of most major issues in the philosophy of science, the sociology of science, and evolutionary theory. According to the preface, his decision to study the systematics community sociologically was made in 1973 (although he probably became convinced of the relevance of the kin-selection concept to the sociology of science even earlier). While much of the material in his book appeared previously in journal articles, the entire package was not published until 15 years later in 1988. If the reasons for such correlations are endogenous, it means that scientists need to assess not only the scale of opportunities available in the field in which they are working but also their own suitability for different strategies. Are they personally more suited for fast, specialized races or for longer, slower work that is broader in scope?

There exists some sentiment that contemporary science is missing out on opportunities for the acquisition of knowledge in the form of too much support for specialized, short-term projects and not enough for broader, longer-term projects. However, in the past decade, there have been some moves in that direction. For example, Nature editorialized on the desirability of support for longer projects (Maddox 1993) and published a commentary decrying the loss of breadth in scientific education and practice (e.g. Greene 1997). The U.S. National Science Foundation moved toward longer, bigger grants around the same time (Mervis 1993). The applied message here is the importance of adapting science and its funding to the opportunities available in a particular field and to the different kinds of comparative advantage possessed by researchers rather than the intrinsic superiority of one strategy over the other. And of course evolutionary and ecological interactions and frequency dependence may obtain with scale-dependent strategies just as they can with density-dependent strategies. Fast specialists deplete small-scale resources but may give large-scale ones an opportunity to recover and vice-versa. If some are or are mainly, fast specialists, then it can pay others to be longer, slower generalists and vice-versa.

It is also worthwhile to point out that given similar energy content in the environment, rate/density and time/space strategies may have similar payoffs, that is only on average. Concentrating as they do in small spatiotemporal ranges, r/d strategies represent high risk, high potential return alternatives (like market timing and stockpicking) while t/s strategies represent low risk, low potential return alternatives (like index funds). Scientists who are highly specialized and work rapidly in their research and teaching may hit the jackpot, but more commonly they come up empty-handed -- the variance in outcomes should be high. Not surprisingly given the emphasis on specialized, short-term research, there is some evidence that error rates in the published literature have been increasing since the early 1970's (Hawkins 1999). Choosing exactly what to specialize in and when to exhaust resources in working at high speed is a risky business. Generalists who work more slowly but for longer may be less likely to make a revolutionary discovery, but they should also be less likely to come up empty-handed. The variance in outcomes should be lower.

6 Heterogeneous Environments: Flexible Strategies

It has long been known that development is a process not only of growth but also of morphological, physiological and behavioural change and differentiation. In individual development, life cycles pass through a sequence of stages and different kinds of cells, tissues and organs emerge, becoming different from each other -- processes largely achieved at the molecular genetic level through change and variation in control of gene expression. A genotype can give rise to individuals, each of which is internally heterogeneous in time/space. At the opposite extreme, on an evolutionary rather than a developmental scale, populations, species and higher taxa change and diverge from each other -- processes largely achieved through genetic change and variation. Between these, cyclical change can take place between generations in dual or even multi-generational life cycles (common in the complex life cycles of many parasites), and different kinds of individuals can become different from each other (morphs, genders, morphs within genders etc.). This between as opposed to within generations/individuals but still intrapopulational change and variation is sometimes achieved genetically -- gender differences in birds and mammals are genetic. Sometimes it is achieved through change and variation in control of gene activity. In many turtles and reptiles gender is phenotypically plastic. A genotype can give rise to heterogeneous generations/kinds of individuals (the former presumably by means of epigenetic inheritance mechanisms). In either the within or between generations/individuals case, heterogeneous environments favour flexibility -- which type being dependent upon patch size. If small, morphological, physiological or behavioural change/variation tends to occur within generations/individuals (change/differentiation) -- creating phenotypic checkerboards. If large, they can obtain between generations/individuals (phenotypic plasticity) -- creating a checkerboard of phenotypes. If uncertainty prevails but reliable signals are available, change/variation of either kind may be condition-dependent. Gender differences in many turtles and lizards are dependent upon environmental temperature. Investing in flexibility is favoured in heterogeneous environments because normally specialist stages/structures are more efficient in the range of a niche they specialize in than are generalists in that range. Presumably the cost of flexibility must initially be traded off against growth, i.e. there are fixed or set-up costs of switching/varying in order to adapt to a heterogeneous environment. Just as with growth versus the 3M's, under the right conditions, the advantages are more than sufficient to compensate for these costs.

Flexibility in scale that takes place both within (differentiation) and between (plasticity) individuals is illustrated by organisms with a branching architecture. Such architecture has evolved independently in organisms many times among the unicellular (some algae), the multicellular (some algae as well as fungi and plants), and the colonial. Colonies of many marine invertebrates in which the individual elements are multicellular animals develop such an architecture. It is found even among colonies of eusocial insect colonies as in polygyne (multiple queen) fire ants (Solenopsis invicta). Supercolonies of these insects spread in the American south in vast branching networks in which individual colonies, connected by underground tunnels, arise by a budding process (Mann 1994). There is a large literature on possible evolutionary explanations of the growth habit and other features of branching organisms much of it on whether the function of branching relates to resources, antagonists or competitive interactions (for a discussion and references see Buss and Blackstone 1991). Here, for convenience, we shall assume resources are at issue.

The simplest of such organisms may be viewed as being composed of what, for simplicity's sake, we shall call and "laterals" and "leaders". A branching architecture is correlated with sessility and as mentioned in Section 3, where only laterals consume and hence the internal flow of resources is unidirectional, the function of leaders is to extend the organism's reach. Leaders can function as a form of motility in sessile organisms (Andrews, 1991). Often however, both laterals and leaders consume as in a fungal hypha; hence the internal flow of resources is bi-directional. Assuming equal profitability at equilibrium, the normally shorter laterals must be consuming a higher mass per unit area or volume (and presumably more rapidly), i.e. are more r/d selected. The normally longer leaders must be consuming resources through a larger area or volume of space (and presumably for longer), i.e. are more t/s selected. The entire organism is a mixed r/d and t/s strategist, adapted to more than one ecological scale. Despite the mixed nature of their strategy, individuals and populations do not necessarily devote equal resources to the two. If laterals arise frequently, which in turn, functioning as leaders, also give rise to their own laterals frequently the overall effect is a short but dense, bushy structure, a pattern long called "phalanx" like. This is still a mixed strategy, but one tilted towards the r/d end of the spectrum. Conversely, if leader(s) only give rise to laterals infrequently, which in turn, functioning as leaders, also only give rise to their own laterals infrequently, the overall effect is a long but sparse, tree-like structure, a pattern long called "guerrilla" like. This is still a mixed strategy, but one tilted towards the t/s end of the spectrum (for diagrams and examples see Andrews 1991, Buss and Blackstone 1991). Hence the same character can exhibit both kinds of flexibility -- differentiation within and plasticity between individuals.

Branching organisms are often constructed modularly whether somatically, reproductively, or both (whether severally or jointly). Somatic modules commonly arise from buds in the upper axes between laterals and leaders. These are a physical manifestation of resources invested in differentiation. They give rise to both leaders and laterals; they permit environmental exploitation to take place on more than one scale.

The parallel in science is clear. In heterogeneous environments, scientists can be called upon to not confine themselves to a single strategy. They may be required to pay the set-up costs of varying and changing among strategies -- working on one project in one way and another in another way. At various times and places they may be required to engage in both primary and secondary research, in primary research and changing topics or fields, in being conformist and heretical, in working as a fast specialist and as a longer slower generalist -- even if there is likely to be an over-all emphasis on one or the other. Not only are such mixed strategies possible, in fields in which opportunities are available on more than one scale, such flexibility lies at the very heart of the scientific process. Scientists commonly describe what they do in their research as reasoning from the specific to the general and from the general to the specific and importantly, believe that each complements the other. Philosophers are more inclined to talk of induction and deduction, or in more modern terminology observation and theory, with explanations emerging from their interaction (Kosso 1992). Such a formulation is consistent with what is observed in nature. Resources spent on maximizing the acquisition of specialized knowledge rapidly may yield a return which can be invested in the acquisition of general knowledge, over a long period of time. Science can proceed from the specific to the general and from the short to the long run. Equally, resources spent on the acquisition of general knowledge over a long period of time may yield a return that can be invested in the acquisition of specific knowledge, rapidly. Science can proceed from the general to the specific, and from the long to the short run.

Commonly philosophers argue that for circularity to be avoided in the scientific process, observations that nourish a particular theory should be independent of those used to confirm it. Hull builds his evolutionary sociology of science from a series of narratives about the history of biology and then tests it against evidence from the contemporary systematics community. Because observation is "theory laden" the requirement may be put differently: the theory used to support particular observations should be independent of the theory for which the observations serve as evidence (Kosso 1992, 155-8). However it is put, branching organisms proceed in such a manner. Hence laterals that nourish a particular leader are distinct from the laterals that the leader subsequently supports. The part of a leader that supports particular laterals is distinct from the part of the leader that the laterals subsequently nourish. Science, like branching growth, is an iterative process. In highly developed sciences, developing new subtopics or specialties (modules) capable of giving rise to both is one of the most valuable activities of all.

For heuristic purposes I slipped from scientists' "reasoning from the particular to the general and from the general to the particular" to philosophers' induction and deduction or observation and theory. I emphasize however that the original terminology more adequately expresses the evolutionary-ecological strategy distinction drawn here. That distinction is not the traditional philosophical one rather it is between specialized research performed rapidly, and more generalized research performed slowly but for longer. Empirical, theoretical, or methodological research can be carried out in either style. The scientific literature is replete with theoretical research for example, mathematical models, of narrow scope and short lifespan.

At the same time, some caution is in order. There is something of a cult of variation and change in science today which from a bench perspective often appears to have more to do with the careers of administrators and the whims of granting agencies than with the real needs of research and teaching. Scientists and departments are told to change and distinguish themselves from one another. It sometimes appears to scientists that some days and places administrators want more research/teaching and others better research/teaching; some staying put and becoming established in a field and others pursuing the latest fad in topics, methods and theories; some being conservative and others radically innovative; some narrow specialists and others broad generalists. As long as the demand for such flexibility is imposed by real heterogeneity in the environment of research and education it should be profitable, but such change/variation does impose a fixed or set-up cost which, if unjustified, will detract from, rather than add to, the acquisition and dissemination of knowledge.

7 Conclusion

The general principles of evolutionary ecology applied to science here are not the kind which are normally directly tested empirically in evolutionary ecology. Rather, they are general principles which logically must be true if the general theoretical framework, that culture, including that of science evolves adaptively, that it does so under ecological control, and that energy content, time and space (including niche time and space) are fundamental dimensions of ecological control. However, they do lead into a multitude of byways in which additional assumptions can be added, among which there are choices to be made, which therefore lead to different testable predictions. While concentrating on highways rather than byways then, this paper has covered enough ground to make the case that introducing more evolutionary-ecological thinking into sociocultural evolutionism/memetics in general, and science studies in particular, is liable to be productive. Evolutionary-ecological concepts including density, scale, frequency and heterogeneity; appropriate strategies; and correlations among strategies including the role of comparative advantage suggest the direction in which explanations should be sought for such basic phenomena in science as primary and secondary research/teaching, changing topics or fields, and innovating; fast specialized and longer, slower more generalized research/teaching; avoiding competition; and the need for flexibility.

Notes

2. Strategies included in these tables apply similarly to the individual and the demographic. In addition to density/scale conditions relative to resources summarized in these tables, strategies can be favoured if at low frequency or if their user possesses a material/genetic comparative advantage (preadaptation) for them.

3. Fog's 1997, 1999 cultural r and K selection are different, mechanism-free concepts as he indicates, see (1999), Chapter 4.

4. How resources in the denominator should be measured depends upon the details. For predators, numbers of prey might matter most while for parasites, the size of hosts might matter most for example.

5. With simple inputs for which the decision to accept or reject can be made as a whole, it might be possible to acquire and excrete at low densities (likely using a single entry and exit) and to break down and build up at high densities. Commonly however resources are more complex and need to be processed for the decision to be made. In these circumstances low densities favour growth by acquiring and breaking down; high densities favour maintenance by building up and excreting.

6. See note 4.

7. This follows from a simple (linear) equilibrium model of density and frequency dependence. Assume that strategies of growth and maintenance/motility are present in frequencies of N and (1-N) respectively at costs of C and (1-C) respectively. Let E be the amount of resources which remain freely available in the environment and (1-E) be the amount which has been absorbed into the population. For pure density dependence we set the product of the frequency and cost of one strategy divided by its total benefits equal to that of the other, i.e. at equilibrium:

(1) 

For pure frequency dependence the total benefits of one strategy are proportional to the frequency and cost of the other so that at equilibrium:

(2) 

For combined density and frequency dependence then:

(3) 

Appendix: r And K Selection Revisited

In essence, the original r versus K selection theory of what was to be expected under low and high density conditions did not distinguish between uncircumscribed and circumscribed environments and tried to explain the spatial and temporal characteristics of organisms by paying attention only to the energy content of the environment and ignoring its spatial and temporal features. In section 5, the problem of explaining why size and life span are commonly positively correlated was addressed beginning with the concepts of small versus large environmental scale and rate/density versus time/space selection.

Stably low densities were thought to be unlikely because populations were expected to normally eat and breed their way up to the carrying capacity of the environment. Hence the original theory of r selection proposed that low or a history of unstable densities favour growth strategies. A population with a history of exponential growth punctuated by catastrophes was expected to be dominated by such strategies, which is reasonable. However, it also needs to be added, whether with reference to strategies of individual or demographic growth or both, that catastrophes of similar magnitude may occur with high frequency relative to period putting a premium on the rate of individual/demographic growth. Alternatively, a long period relative to frequency puts a premium on the time through which growth is sustained. Hence the distinction was drawn between r and t selection, attributable not to density but to (temporal) environmental scale. Similar distinctions apply spatially betweend (density, i.e. in the sense of mass or numbers per unit area or volume) versus s (space, i.e. area or volume) selection. Ecological correlations between temporal and spatial scales, or alternatively, correlated comparative advantages (material or genetic) are what must explain positive correlations between life span and size -- whether physical or metaphoric, i.e. niche span and size. Scientists in their research and teaching, no less than organisms in their consumption and production of offspring, should evolve to conform to the spatio-temporal scale small versus large of resources available in their environment.

The original theory of K selection proposed that maintenance strategies were favoured at high densities or with a history of stably high densities. However, just as stably low densities are unlikely because populations are liable to eat and breed their way to the ceiling, stably high densities are unlikely because populations are liable to be eroded to the floor by their parasites and predators. On the other hand, just as catastrophes can strike growing populations, bonanzas can give a boost to declining ones. Hence it is reasonable to conclude that high densities or a history of instability in the form of bonanzas should favour maintenance strategies. Bonanzas of similar magnitude, like catastrophes, can take place with high frequency relative to period/distance, putting a premium on speed of motility/density of maintenance. Or they can take place with a long period/distance relative to frequency, putting a premium on the time/space through which such are sustained.

In addition to claiming that low densities favour growth strategies while high densities favour maintenance strategies (a form of which has been embraced here for uncircumscribed environments, although motility and in some cases mutability are also options), and that low density selected life cycles should be small and fast while the high density selected should be large and slow but long (rejected here explaining organismic scale requires concepts of environmental scale), the original theory of rversus K selection claimed that the low density (r) selected devote more resources to demographic growth (e.g. in science, teaching) while the high density (K) selected devote more to individual growth (e.g. in science, research). This question which can be viewed as belonging properly to the sociobiology of science is beyond the scope of this paper.

Finally on classical evolutionary ecology, r and K selection theory had nothing to say about the other half of the life history story beyond growth/some of the 3M's, i.e. about flexibility. Levins (1968) introduced the subject of evolution in changing/varying environments. Densities are not necessarily low because of catastrophes or high because of bonanzas -- they may be both at different times/places. The scale of environmental resources is not necessarily small or large -- resources may be distributed and utilized on more than one scale. Such heterogeneity calls for effort to be devoted to flexibility which itself may be fine (change/differentiation) or coarse-grained (phenotypic plasticity).

Acknowledgements

References

Andrews, J.H. 1991. Comparative Ecology of Microorganisms and Macroorganisms. SpringerVerlag, New York.

Aunger, R. (ed.) 2000. Darwinizing Culture: The Status of Memetics as a Science. Oxford University Press, New York.

Basalla, G. 1988. The Evolution of Technology. Cambridge University Press, Cambridge.

Bazzaz, F.A. 1996. Plants in Changing Environments: Linking Physiological, Population, and Community Ecology. Cambridge University Press, Cambridge.

Begon, M., Harper, J.L. and Townsend, C.R. 1996. Ecology: Individuals, Populations and Communities. Third Edition, Blackwell Science Ltd., Oxford.

Blackmore, S. 1999. The Meme Machine. Oxford University Press, Oxford.

Boulding, K.E. 1981. Evolutionary Economics. Sage Publications, Thousand Oaks, CA.

Brodie, R. 1996. Virus of the Mind: The New Science of the Meme. Integral Press, Seattle.

Bulmer, M. 1994. Theoretical Evolutionary Ecology. Sinauer Associates, Inc. Sunderland, MA.

Buskes, C. 1998. The Genealogy of Knowledge: A Darwinian Approach to Epistemology and Philosophy of Science. Tilburg University Press.

Buss, L.W. and Blackstone, N.J. 1991. An Experimental Exploration of Waddington's Epigenesis Landscape, Philosophical Transactions of the Royal Society, London B 332:49-58.

Cairns, N.J., Overbought, J. and Miller, S. 1988. The Origin of Mutations, Nature 335:142-145.

Cairns, J. and Foster, P.L. 1991. Adaptive reversion of a frameshift mutation in Escherichia coli, Genetics 128:695-701.

Campanario, J.M. 2000. Fraud: Retracted Articles Are Still Being Cited, Nature 408: 288.

Campbell, D.T. 1988. Methodology and Epistemology for Social Science: Selected Papers. University of Chicago Press, Chicago.

Carneiro, R.L. 1970. A Theory of the Origin of the State, Science 169:733-738.

Cavalli-Sforza, L.L. and Feldman, M.W. 1981. Cultural Transmission and Evolution: A Quantitative Approach. Princeton University Press, Princeton.

Chance, D.M. and Peterson, P.P. 1999. The New Science of Finance, American Scientist 87:256-263.

Charnov, E.L. 1993. Life History Invariants: Some Explorations of Symmetry in Evolutionary Ecology. Oxford University Press, Oxford.

Cockburn, A. 1991. An Introduction to Evolutionary Ecology. Blackwell Scientific Publications, Oxford.

Croft, W. 2000. Explaining Language Change: An Evolutionary Approach. Pearson Education Limited, Harlow, Essex.

Cullen, B.S. 2000. Contagious Ideas: On Evolution, Culture, Archaeology, and Cultural Virus Theory. Oxbow Books, Oxford.

Dawkins, R. 1976. The Selfish Gene. Oxford University Press, Oxford.

Degler, C.N. 1991. In Search of Human Nature: The Decline and Revival of Darwinism in American Social Thought. Oxford University Press, New York.

Ehrlich, P.R. 2000. Human Natures: Genes, Cultures and the Human Prospect. Island Press, Washington, D.C.

Fog, A. 1997. Cultural r/k Selection, Journal of Memetics - Evolutionary Models of Information Transmission, 1(1). http://cfpm.org/jom-emit/1997/vol1/fog_a.html

Fog, A. 1999. Cultural Selection. Kluwer Academic Publishers, Dordrecht.

Foster, P.L. 1999. Mechanisms of Stationary Phase Mutation: A Decade of Adaptive Mutation, Annual Review of Genetics 33:57-88.

Fox, C.W., Roff, D.A. and Fairbairn, D.J. 2001.Evolutionary Ecology: Concepts and Case Studies. Oxford University Press, Oxford.

Gill, D.E. 1974. Intrinsic Rate of Increase, Saturation Density, and Competitive Ability II. The Evolution of Competitive Ability, American Naturalist 108:103-116.

Greene, M.T. 1997. What Cannot Be Said In Science, Nature 388:619-620.

Hall, B.G. 1998. Adaptive Mutagenesis: A Process That Generates Almost Exclusively Beneficial Mutations, Genetica 102/103:109-125.

Hanski, I.A. 1999. Metapopulation Ecology. Oxford University Press, New York.

Hanski, I.A. and Gilpin, M.E. (eds.) 1997. Metapopulation Biology: Ecology, Genetics and Evolution. Academic Press, London.

Hawkins, B.A. 1999. More Haste, Less Science, Nature 400:498.

Horgan, J. 1996. The End of Science: Facing the Limits of Knowledge in the Twilight of the Scientific Age. Addison-Wesley Publishing Company, Inc. Reading, Mass.

Hull, D.L. 1988. Science as a Process: An Evolutionary Account of the Social and Conceptual Development of Science. The University of Chicago Press, Chicago.

Hull, D.L., Langman, R.E. and Glenn, S.S. 2001. A General Account of Selection: Biology, Immunology, and Behavior, Behavioral and Brain Sciences 24:511-573.

Kosso, P. 1992. Reading the Book of Nature: An Introduction to the Philosophy of Science. Cambridge University Press, Cambridge.

Krebs, J.R. and Davies, N.B. 1993. An Introduction to Behavioural Ecology, Third Edition. Blackwell Scientific Publications, Oxford.

Krebs, J.R. and Davies, N.B. (eds.) 1997. Behvioural Ecology: An Evolutionary Approach, Fourth Edition. Blackwell Science Ltd., Oxford.

Kuhn, T.S. 1962. The Structure of Scientific Revolutions. University of Chicago Press, Chicago.

Lenski, G., Lenski, J. and Nolan, P. 1994. Human Societies: An Introduction to Macrosociology. McGraw Hill, Whitby, On.

Levins, R. 1968. Evolution in Changing Environments. Princeton University Press, Princeton.

Lovins, A. 1976. Energy Strategy: The Road Not Taken, Foreign Affairs 55:65-96.

Lynch, A. 1996. Thought Contagion: How Belief Spreads Through Society. Basic Books, New York, N.Y.

MacArthur, R.H. 1962. Some Generalized Theorems of Natural Selection, Proceedings of the National Academy of Sciences U.S.A. 48:1893-1897.

MacArthur, R.H. and Wilson, E.O. 1967. The Theory of Island Biogeography. Princeton University Press, Princeton.

Maddox, J. 1993. Competition and the Death of Science, Nature 363:667.

Mann, C.C. 1994. Fire Ants Parlay Their Queens Into a Threat to Biodiversity, Science 263:1560-1561.

Merton, R.K. 1973. The Sociology of Science. The University of Chicago Press, Chicago.

Mervis, J. 1993. At NSF: Fewer, Longer Grants, Science 262:l636-1638.

Nelson, R.R. 1996. The Sources of Economic Growth. Harvard University Press, Cambridge.

Nelson, R.R. and Winter, S.G. 1982. An Evolutionary Theory of Economic Change. Harvard University Press, Cambridge, Mass.

Pianka, E.R. 1970. On r and K selection, American Naturalist 104:592-597.

Pianka, E.R. 1999. Evolutionary Ecology, Sixth Edition. Addison-Wesley Educ., New York.

Plotkin, H. 1994. Darwin Machines and the Nature of Knowledge. Harvard University Press, Cambridge, MA.

Popper, K.R. 1963. Conjectures and Refutations. Basic Books, New York.

Promislow, D.E.L. and Harvey, P.H. 1990. Living Fast and Dying Young: A Comparative Analysis of Life History Variation Among Mammals, Journal of Zoology 220:417-437.

Rhodes, O.E., Chesser, R.K. and Smith, M.H. 1996. Population Dynamics in Ecological Space and Time. The University of Chicago Press, Chicago.

Ricklefs, R.E. and Miller, G.L. 2000. Ecology, Fourth Edition. W. H. Freeman and Company, New York.

Roff, D.A. 1992. The Evolution of Life Histories: Theory and Analysis. Chapman & Hall, London.

Rosenberg, S.M. 2001. Evolving Responsively: Adaptive Mutation, Nature Reviews Genetics 2:504-15.

Ruhlen, M. 1994. The Origin of Language: Tracing the Evolution of the Mother Tongue. John Wiley & Sons, Inc., New York.

Runciman, W. G. 1989. A Treatise on Social Theory II, Substantive Social Theory. Cambridge University Press, Cambridge.

Sibly, R.M.. and Calow, P. 1986. Physiological Ecology of Animals: An Evolutionary Approach. Blackwell Scientific Publications, Oxford.

Smith, E. A. and Winterhalder, B. (eds.) 1992. Evolutionary Ecology and Human Behavior. Aldine De Gruyter, New York.

Sperber, D. 1996. Explaining Culture: A Naturalistic Approach. Blackwell Publishers Ltd. Oxford.

Stearns, S. C. 1992. The Evolution of Life Histories. Oxford University Press, Oxford.

Stephens, D.W. and Krebs, J.R. 1986. Foraging Theory. Princeton University Press, Princeton.

Toulmin, S. 1972. Human Understanding. Princeton University Press, Princeton, N.J.

Watson, J.D. 1969. The Double Helix. New American Library, New York.

Whiten, A. 2001. Imitation and Cultural Transmission in Apes and Cetaceans, Behavioral and Brain Sciences 24:359-360.

Ziman, J.M. 1968. Public Knowledge: An Essay Concerning the Social Dimension of Science. Cambridge University Press, Cambridge.

Ziman, J.M. (ed.) 2000. Technological Innovation as an Evolutionary Process. Cambridge University Press, Cambridge.