The evolution of sex: Domains and explanatory pluralism[1]

Carla Fehr

Iowa Sate University

Abstract

The evolution of sexual reproduction is a striking case of explanatory pluralism, meaning that one needs to refer to more than one explanation in order to adequately account for it. I develop the concept a domain of phenomena in order to analysis this pluralism. Pluralism exists when a phenomenon can be included in more that one homogeneous domain or in a heterogeneous domain. I argue that in some cases domain partitioning can be used to decrease pluralism, but that in the case of sex domains are overlapping and interconnecting, or in other words bear an orthogonal relationship to one another, and hence cannot be partitioned in such a way as to eliminate pluralism.

In light of his theory of evolution by natural selection, even Darwin could find no strong reason for the predominance of sexual reproduction. He wrote that the “whole subject is as yet hidden in darkness.” In the nearly one hundred years after Darwin wrote The Descent of Man (1872), biologists consistently hypothesized that sexual reproduction functioned to maintain the viability of the sexually reproducing group or species. However, in the 1960’s and 1970’s, biologists such as G.C. Williams (1966, 1975) and John Maynard Smith (1978), forcefully argued that these previous sorts of explanations were extremely limited. Thus, biologists were left with a lacuna in their description of reproduction: there appeared to be no evolutionary explanation for sex. A hundred years after Darwin, Graham Bell (1982) called the evolution of sex “the queen of problems in evolutionary biology.” He continued to write that “perhaps no other natural phenomena has aroused as much interest; certainly none has sowed as much confusion.” I analyze the evolution of sexual reproduction as a case study in the philosophy of biology, and develop a positive argument for the existence of explanatory pluralism in the study of complex biological phenomena.

This paper has three main sections. First, I spell out why sex represents such a tantalizing phenomenon for evolutionary biologists. To illustrate the plurality of explanations for this phenomenon I will describe two, of the many, current explanations for the evolution of sex: the Red Queen explanation and the DNA Repair explanation. In the second section of the paper I describe the philosophical problem posed by this pluralism, and consider several possible causes of explanatory pluralism.

In the final section of the paper I give an account of pluralism in terms of explanatory domains. A domain is a group of phenomena, be they states of affairs, events or regularities that scientists group together as something that merits explanation. Pluralism obtains when we need a multiplicity of explanations to account for a given domain. I argue that in some cases we should expect this sort of pluralism because domains can hold an orthogonal relationship to one another, or in other words they can be overlapping and interconnected. I conclude that in the case of evolutionary phenomena, pluralism can be a virtue.

1 The evolution of sex as a case of explanatory pluralism

1.1 What is sexual reproduction?

Sexual reproduction is a complicated phenomenon. This paper is not focused on sex in terms of gender, or sexual dimorphism or particular acts of sexual reproduction. This paper focuses on why an organism would reproduce sexually as opposed to asexually.[2] At the most general level, sex can be thought of as any exchange of genetic material. More specifically sex is often characterized in terms of two processes, out crossing and meiosis. Out crossing occurs when the DNA of two different parents is combined in the formation of a single offspring. This process relates to the more general definition because it allows for the exchange of genetic material between different individuals. Meiosis is, roughly, a process associated with the formation of gametes (eggs and sperm) during which the ploidy of the daughter cells is reduced to half that of the original mother cell. Two things happen during meiosis. First, during meiosis crossing over can occur (do not confuse crossing over, which happens during meiosis, with out crossing which is what can happen to gametes that are produced via meiosis). Crossing over refers to an exchange of genetic material among an individual’s chromosomes and can result in new gene combinations and hence increase genetic variability. Second, during meiosis duplicate copies of chromosomes come into close association with one another. While they are in this position, if one of them happened to be damaged, the other can act as a template in order to repair the damaged chromosome without losing essential genetic information.

Depending on whether sex is defined in terms of out crossing or meiosis, the same organism could be characterized as sexual or asexual. The relationship between out crossing and meiosis is complicated. For example, in many flowers, such as the Common Morning Glory, all of the gametes are formed by meiosis, but sometimes a flower is fertilized by, or fertilizes, a different flower, and this is a case of out crossing, or sometimes that flower can fertilize itself, and this is called self fertilization or inbreeding.[3] If sex is defined in terms of meiosis both of these flowers are reproducing sexually. If sex is defined in terms of out crossing, it is the fate of the gamete that determines whether it is sexual or not: if it out crosses it is sexual and if it is involved in self fertilization it is asexual. Notice that these processes, or processes roughly equivalent to them, need to be intimately interconnected. If gametes are formed, they need to be joined and if gametes are joined, they need to be formed in the first place.[4] But, there is more than one way to make gametes and more than one process by which gametes can be joined together.

1.2 Why is sex an evolutionary problem?

Sex is expensive. Although we see it nearly wherever we look in the living world the costs of sexual reproduction are very high. All else being equal there is a twofold benefit that asexual reproduction has over sexual reproduction because when an organism reproduces asexually it passes one hundred percent of its DNA to its offspring while a similar sexual organism only passes fifty percent of its DNA to its offspring with the other fifty percent coming from the other parent. Furthermore, sex can break apart favorable gene combination and lead to increases in homozygosity. When one adds to these costs all the trouble of finding a mate, what we already knew becomes even more apparent: sex is expensive and time consuming.[5] Given these costs one could argue that we ought not find sex in the natural world. But, sexual reproduction evolved from asexual reproduction, and moreover, there are some organisms that can either reproduce sexually or asexually, but that reproduce sexually much of the time. The predominance of sexual reproduction and the evolutionary switch from asexuality to sexuality, in light of these costs, creates what Richard Michod (1995) calls the "paradox of sexual reproduction" and has spawned a plethora of answers to the question "why sex?"

Sexual reproduction is one of the finest cases of theoretical pluralism that exists in the biological sciences; Kondrashov (1993) created a taxonomy of over 20 different kinds of theories attempting to explain the evolution of sex. Michael Ghiselin (1988) writes that, “perhaps researchers shall have to accept a pluralistic assemblage of explanations”. He sadly accepts that at least some of this pluralism is persistent – that there are many explanations for sex that are going to have to exist at the same time. Two of these theories are the Red Queen explanation and the DNA Repair explanation.[6]

1.3 The Red Queen

The Red Queen explanation is named after the character in Alice in Wonderland who had to continually run in order to stay in the same place. Supporters of this view note that the environment is changing, and postulate that out crossing allows sexual organisms to stay adapted to this changing environment. The particular sort of environmental change considered is the continual and ubiquitous coevolution between forms of parasite virulence and host defense mechanisms. According to this explanation, sexual reproduction is maintained over evolutionary time because out crossing can produce novel gene combinations, which may allow an individual’s offspring to avoid disease and parasitism.

Consider the following scenario. Imagine a fir tree, a tree that lives for 50 years. Imagine a parasite on that tree, a parasite that produces a new generation every three months. The tree has a stable genotype for its entire life, and as a part of that genotype there is a code that enables the tree to resist the infection of a certain parasite genotype. When the tree is young, this genetic resistance to parasite infection functions well. But as the years go by, or, more accurately, as the parasite generations go by, there is lots of time for the population of parasites to evolve: the number of parasite individuals with the genotype that the tree is resistance to will decrease in the parasite population, and the number of individuals with a genotype that the tree is not resistant to will increase in the population. As the tree ages, the parasite population adapts and becomes better able to infect the tree. If that tree reproduces asexually, its offspring will be genetically identical to it. And that parasite population that adapted to the parent tree over its entire life will be adapted to the young tree right from the beginning of its life. If that parent reproduces sexually, it has a chance of coming up with a new genotype in its offspring that the parasite will not yet have had a chance to adapt to.

If there is a genetic basis for a host's ability to resist parasite infection and a parasite's ability to overcome this resistance, then an arms race will tend to develop between host and parasite in which the host is continually developing new strategies for resistance and the parasite is evolving mechanisms to overcome that resistance. According to this theory, because sexual organisms out cross, the offspring are genetically different from the parents and so are less susceptible to the parasites and disease than either parent, or a similar asexual organism, would have been exposed to. In other words, sex allows parents to produce offspring with locally rare genotypes that benefit from negative frequency dependent selection.

The lion’s share of the support for this explanation is in the form of population genetic models demonstrating that there can be cycling relationship among host and parasite populations (for example Seger and Hamilton 1988). But there is also empirical support for this explanation. Edmunds and Alstad (1981) found that Pine Leaf Scales responded to genetic variation in their host plant, the Douglas Fir. Antonovics and Ellstrand (1984) conducted an experiment in which they took grass seedlings, some of which were produced from seed (sexual) and some of which were cut from runners (asexual) and planted them at varying distances from the parent plant. They found that near the parents the sexually produced offspring set significantly more seed than the asexually produced offspring. Curtis Lively’s research group (Dybdahl, M. and Lively, C. 1995a,b; Howard, R. and Lively, C. 1994) studied a species of guppy and of snail, both of which can produce sexually or asexually. They found that in situations in which there was a higher level of parasitism there were more sexually reproducing individuals than there were clonally (asexually) reproducing individuals. These three cases show that parasites can adapt to their hosts, and that it can be beneficial to reproduce sexually when offspring will develop in close proximity to the parent or in situations where there are significant amounts of parasitism.

1.4 DNA Repair

Although, the DNA Repair theory for the evolution of sex has been recently championed by Harris Bernstein and Richard Michod, this is not a new idea. Already in the 1950’s Dougherty (1955) hypothesized that the origin of sex was due to the ability of sexuals to repair genome damage. Bernstein et. al. (1985a; 1985b; 1987) explain the evolution of sexual reproduction in terms of the function of the molecular mechanisms of meiosis. They considered the structure and biochemistry of the reactions that take place during meiosis and the regulation of those reactions, and found that meiosis is very good at, and appears to be well designed for, repairing double strand DNA damage.[7] Supporters of this view (Bernstein et. al. 1985a; Michod and Long1995; Long and Michod 1995) argue that the explanation of both the origins and the maintenance of sex is “based on a selective advantage arising from recombinational repair of genetic damage (Bernstein et. al. 1985a).” This repair process allows organisms to produce a greater number of genetically intact gametes and hence a greater number of viable offspring.

DNA damage is an alteration of the structure of a DNA molecule that prevents the molecule from being replicated and hence from being inherited. This sort of DNA damage occurs frequently and is caused by a variety of factors such as the by-products of cellular activity, x-rays and UV light. Double strand DNA damage, if not repaired, results in the loss of genetic information, which can be harmful or lethal to an individual’s gametes and offspring. During meiosis the homologous chromosomes of diploid organisms are brought into intimate association with one another other. While they are close together crossing over can occur and one chromosome can copy information from the other and use it to patch damaged areas and recover lost information. Experimentally researchers have found that cells that don’t undergo this repair process are very sensitive to agents of DNA damage and that the rate of repair activity increases in response to an increase in agents that cause DNA damage (reviewed in Bernstein 1983; Bernstein et. al. 1987). In this explanation, the force of the argument is found at a molecular level of biological organization. It comes from being able to discover the actual chemical steps that occur during the cellular process of meiosis.

1.5 Summary

Explanations of the occurrence of out crossing, often make reference to the processes of meiosis. These references are not a central part of the modeling and experimentation in this area, but, since these explanations refer to diploid systems in which gametes join, there is going to have to be some mechanism (at least in animals) for reducing the ploidy of the gametes. But, what is more interesting is that Bernstein's group (Bernstein et. al. 1987), which is in the business of explaining meiosis (in terms of crossing over in diploid systems), even though they have a wealth of data concerning the molecular functioning of meiosis, also makes reference to the ecological circumstances in which one expects to find and does find elevated rates of DNA repair. Although this work concentrates on the functioning of meiosis, it has a detailed story concerning the role of outcrossing and diploidy. Something like meiosis and something like fertilization need to occur together. Even though these authors focus their work different aspects of sexual reproduction, out crossing and meiosis, these two phenomena are both parts of a complicated functioning sexual system. These two explanations are not mutually exclusive. In some cases they could both explain the evolution of sex. In some cases one or the other or neither explanation may be necessary to account for the evolution of sexual reproduction. There is more than one explanation for the evolution of sex. The task that faces both biologists and philosophers alike is to illustrate the relationships between these different explanations.

2 Philosophy and explanatory pluralism

2.1 The Newtonian Ideal

Explanatory pluralism, a situation in which more than one explanation is needed in order to adequately account for a phenomenon or domain of phenomena, is philosophically challenging because it conflicts with what John Beatty has called the Newtonian Ideal. Newton wrote that “Nature does nothing in vain and more is in vain when less will serve, for Nature is pleased with simplicity and affects not the pomp of superfluous causes” (1686). The Newtonian Ideal is a traditional scientific ideal according to which, a good explanation will unite many phenomena. This ideal describes an aim of science, which is to explain a phenomenon, or a domain of phenomena, in terms of as few different mechanisms, or explanations, as possible and best of all with one mechanism. The Newtonian Ideal and explanatory pluralism are in conflict with one another.

2.2 Possible causes of pluralism

There are many different possible causes for explanatory pluralism. The presence of explanatory pluralism may be due to the immaturity of the investigation of a particular topic or subject area, in which case the reason why there are many explanations for a phenomenon is that scientists have yet to find the correct one. In cases like this it may be just a matter of time, or possibly of scientific progress, before researchers find the correct explanation of a phenomenon and hence decrease the number of explanations needed to account for it to a single one.

The presence of explanatory pluralism may also be due to ambiguous explanation-seeking questions, in which case when the question is clarified, the number of acceptable answers to the question would decrease. This pragmatic approach is most famously pushed by Bas van Fraassen and his example of a priest asking a bank robber, “why did you rob the bank?” and the bank robber replying, “because that was where the money was” (van Fraassen, 1980). This is an obvious case of miscommunication that can be solved by clarifying the question being asked. The bank robber was assuming that the priest was asking him why he was a bank robber as opposed to getting a job as a lawyer or as an assistant professor. And the bank robber assumed that the priest was asking him why he robbed the bank as opposed to robbing the orphanage or the public hospital. In this example, once the intent of the questioner is made clear, we end up with a question to which there is one right answer; pluralism is decreased. Pluralism in biological explanation can result from the pragmatic realization that several well-founded questions can arise concerning a single phenomenon (Sherman 1988). The question “why sex?” is ambiguous in several respects, such as the taxa that the researcher is asking about and the period of evolutionary history the researcher is asking about. For example, the question “why sex?” could mean, “why did sex originate?” or “why is sex maintained in extant populations?” This particular disambiguation of the question has resulted in at least a temporary decrease in explanatory pluralism and I will address it in more detail in section 3.3.[8]

The previous possible cause of explanatory pluralism concerns ambiguity in the explanation-seeking question. Sandra Mitchell’s (1992) work on pluralism focuses not on the explanation seeking question, but rather on classifications of what counts as acceptable answers to these questions. To this end, she argues that there can be a plurality of ideal causal models that could be explanatory, but that there will only be a single constellation of causes that explain any particular event. She argues that there will be a piecemeal integration of these various models in the explanation of particular cases.

Researchers may differ in terms of what kind of answer or kind of causal model they find explanatory, and this could result in explanatory pluralism when a particular problem is considered from the perspective of different disciplines, backgrounds, or different research traditions. This sort of explanatory pluralism would be the result of disciplinary differences, which could be characterized in terms of cultural differences among researchers working in different biological disciplines or could be characterized in terms of the multiplicity of the aims of different researchers. In this case, pluralism could be decreased by only comparing explanations within disciplines and not among disciplines. If one was to ask why a flower reproduced sexually as opposed to as asexually, a geneticist may refer to the genetic causes of flower color and morphology, while a behavioral biologist may accept an answer in terms of the behavior of the bumble bee that pollinates the flower. These two answers are acceptable and useful in the two different disciplines. In this case there need not be pluralism within a discipline or research program. Interdisciplinary studies, especially evolutionary studies, may require one to consider more than one discipline at a time and hence be able to integrate or compare these different explanations. [9]

There could be other causes for explanatory pluralism. Note that in the different causes of pluralism that I have considered so far there is an implicit assumption that explanatory pluralism ought to be decreased or minimized; that pluralism represents a problem that requires a solution. The belief is that researchers can and ought to develop their investigations such that their explanations of phenomena come to more closely coincide with the Newtonian Ideal. The Newtonian Ideal is a normative ideal and it entails an assumption that pluralism represents a problem to be solved, a problem that it is a result of confusion or of the limitations of researchers. But, this normative assumption need not hold. Just because ‘Nature’ is pleased with simplicity, it doesn’t follow that we must be, especially if we find that nature is not simple. John Beatty (1997), in paying close attention to evolutionary contingency, begins to take seriously the relationships that exist among the explanations of things in the world that aren’t simple, the relationships among the explanations of complex evolutionary phenomena.

The evolution of sexual reproduction is just such a complex phenomenon. Sex seems to turn the Newtonian Ideal on its head because in order to account for it researchers will likely need both Red Queen and DNA repair explanations. In fact, there are likely other explanations that researchers will need to engage in order to adequately understand this complex phenomenon. For example, Graham Bell (1988) has argued that we will need DNA repair explanation, the Red Queen explanations as well as a third explanation, Muller’s Ratchet, in order to adequately account for the evolution of sex. Sex is a case where of one phenomenon that is, or will likely need to be, accounted for by many explanations.[10]

2.3 Beatty on theoretical pluralism.

Beatty begins his discussion of pluralism with the, possibly controversial, claim that "multiple, alternative explanations are required for almost every domain of phenomena in biology (Beatty 1994, 1996). The sort of pluralism that he is discussing is not the result of a lack of scientific understanding, or ambiguous questions or multiplicity of aims of different researchers. Theoretical pluralism, as described by Beatty, is the idea that our best science suggests that many different theories are required to explain a domain of biological phenomena. He writes that,

theoretical pluralism obtains when the evidence suggests that a multiplicity of theories or mechanisms are needed to explain a domain of phenomena, different items in the domain requiring explanation in terms of different theories or mechanisms. There simply is no unitary account of the domain. This is not merely a matter of insufficient evidence for a single theory; rather, it is a matter of the evidence indicating that multiple accounts are required" (Beatty 1994, 37).

Beatty doesn't explain what he means by a ‘domain of phenomenon’, but it is charitable to follow Shapere's (1984) characterization of domains in relation to explanations and think of a domain as what researchers are interested in accounting for. Throughout this paper I will refer to a domain as those phenomena, which can be events, states of affairs in the world, or regularities, that scientists group together as something to be explained.

According to Beatty, domains of biological phenomena are theoretically heterogeneous because they are the result of evolution, and evolution is a highly contingent process. [11] Beatty calls this position the evolutionary contingency thesis. There are many different mechanisms that are functionally equivalent, meaning that a biological effect can be produced in many different ways and that it is possible that natural selection may not be able to distinguish among these different causal mechanisms. Roughly speaking, there could be, and likely are, many different origins and causal histories of what, from the perspective of natural selection, is the same effect. Any biological role can be accomplished in a variety of ways and, assuming a causal mechanical view of explanation, all of these different ways of fulfilling a particular role will correspond to a different explanation of the phenomena.

For example, it is advantageous for animals that swim to have a torpedo shape. But if researchers were to give a causal history, or explanation, for how organisms become torpedo shaped, the story, or explanation, would be different for various taxa--fishes and whales and penguins all have different histories and hence different stories of how they became torpedo shaped; they all came to have that morphology via different routes. Therefore there exists a plurality of explanations for the phenomenon that aquatic animals tend to have a torpedo shaped morphology.[12] In other words the domain that consists of the morphology of aquatic animals is heterogeneous. A domain is heterogeneous when the phenomena that it includes can be explained by, or could be included in the scope of, several different explanations. Different explanations may apply to different instances or one instance can be explained by more than one explanation.

Beatty’s theoretical pluralism can partially capture aspects of the case of the evolution of sex. There are many different kinds of sexual mating systems or life cycles, each of which is a different mechanism of sexual reproduction. If we consider sexual reproduction to be a single domain of phenomena, there will be different members of that domain that have different mating systems and hence require different explanations or different constellations of explanations in order to account for them. Sex is a complicated process that has many component parts, various sexual systems may include different subsets of these different sexual parts. Some organisms may undergo meiosis without out crossing or may out cross with gametes that are not produced by meiosis. The former may be best explained in terms of DNA repair and the later might best be explained in terms of the Red Queen explanation. But if we are going to consider the evolution of sex as a whole, then we will need to include both of these explanations and we will have a case of theoretical pluralism.

A problem with Beatty’s account is that it leaves us with the question of why we should consider this heterogeneous assemblage of phenomena (the ones accounted for by different causes) as a group or domain that merits explanation. Beatty argues that researchers should not expect unity of explanation in regard to biological phenomena. He writes that,

To expect a single mechanism underlying an entire domain of phenomena, researchers would have to assume that one mechanism evolved in a common ancestor of all the taxa covered by the domain, and that the mechanism has been maintained in each of those taxa ever since, and/or researchers would have to assume that the very same mechanism arose independently and has been maintained in all the taxa covered by the domain. In other words, researchers would have to assume extreme phylogenetic conservatism, and/or extremely strong and remarkably similar selection pressures resulting in extreme parallel evolution. (Beatty 1997 original italics)

Since chance plays an important role in evolution, Beatty argues that extreme phylogenetic conservatism and extreme parallel evolution are not what researchers should expect. Beatty (1994) gives the example of gene regulation, pointing out that the development of more theories of gene regulation, an increase in theoretical pluralism, is an example of impressive progress in this field.

However it is just here that Beatty's position needs to be taken one step further, because Beatty's critic should now question why gene regulation ought to be considered as a single phenomenon or single domain of phenomena in the first place. The very criteria that an investigator might use to hypothesize the grouping of phenomena into a unitary domain are extreme phylogenetic conservatism and strong selection pressures resulting in extreme parallel evolution. Why not characterize each gene regulation mechanism as a different homogeneous domain rather than lump all of these various mechanisms into a single heterogeneous domain? [13] This appears to be a rather simple way to make this case cohere with the Newtonian Ideal. If researchers cast their net broadly enough, if they try to explain a diverse enough group of phenomena, the result will always be pluralism. But what is at issue is just how broadly the net should be cast.

If we consider this problem as it relates to the evolution of sex, the question that needs to be asked is whether or not we should consider the evolution of sexual reproduction to be one phenomenon (or one domain of phenomena) or not. If some cases of sex are best explained by the Red Queen, and others are best explained by DNA Repair, then why not discard the view that sex is one thing that requires two different explanations and replace it with the view that there are two separate kinds of sex each of which has its own explanation? Pluralism entails a multiplicity of explanations for one thing, in order for it to obtain, we need to be able to think of sex as a phenomenon with this sort of unity. Beatty offers us no reason why this should be the case. To address this issue it is necessary to characterize the concept of a domain. In the next section I develop my own account of pluralism in terms of explanatory domains, which can be overlapping and interconnected and use it address this problem that Beatty faces.

3.1 Domains

Dudley Shapere has done pioneering work on the concept of a domain (Shapere 1974, 1984). Shapere, in considering the scientific characterization of electricity in the 18th century points out that,

It is by no means obvious that all the phenomena which researchers today unhesitatingly group together as forming a unified subject matter or domain under the heading 'electricity' really do form such a unity (Shapere 1974,273).

Shapere generalizes this claim by pointing out that even though researchers think of science as explaining things, it is not clear that the things that science explains are really unified or have any natural unity in themselves. The range of phenomena that an explanation can be applied to is in itself controversial. He writes that,

Nature does not happen to come once and for all divided, on the basis of anything immediately given in experience, into "areas" or "fields" for investigation...[there] are certainly observable features, but that those features should be considered as identifying or indicating the sorts of entities to be studied is not a matter of anything that could be called immediate or obvious sensory characteristics" (Shapere 1984,323).

The immediate question that comes to mind is: if explanations are not about obviously delineated chunks of nature, what is it that an explanation explains, or is about? Shapere states that explanations are of, or about, domains. Shapere characterizes a domain in the following way: A domain consists of items of information related in some way. There is a problem concerning these items. The problem is important and science has the means to deal with that problem. Shapere continues by pointing out that the relationship between the items of a domain must be well grounded and significant. Finally, Shapere writes that a claim "that a body of information constitutes a domain is itself a hypothesis that may ultimately be rejected" (Shapere 1974,281). A domain is what is "accounted for" by an explanation, and the explanation "accounts for", or gives a "deeper" understanding of, the domain which it explains.

This characterization of domains, as things that require careful and deep investigation simply to define them, accords well with Beatty’s project. For Beatty, what at one level, the level at which natural selection acts, is a unified phenomenon, may actually have many underlying mechanisms which produce that phenomenon and so there are many explanations of what, from the perspective of natural selection is one domain. A 'deeper' investigation is necessary to discover all of these different mechanisms or explanations of a domain of biological phenomenon. That these different mechanisms exist does not entail that researchers can immediately distinguish among them. Finally, researchers can change their minds as to what ought to be included in any particular domain, or in the way that phenomena ought to be grouped.

In explaining domains, Douglas Allchin (1997) employs a particularly vivid map analogy, stating that natural phenomena are like territory represented on a map and that domains delimit the territory that a particular explanation applies to. For my purposes a domain, D, is the group of phenomena P, which can be events, states of affairs in the world, or regularities, that scientists group together as something to be explained.[14] The question that must be answered is "is there an appropriate domain for explanation that remains heterogeneous even after it has been adequately characterized?" In order to answer this question I begin by considering the different ways in which a heterogeneous domain can be split into sub domains.

A domain can be homogeneous or heterogeneous. A domain is homogeneous when the phenomena that it includes are only explained, or included in the scope of, one type of explanation. If a domain is homogeneous, there will be no explanatory pluralism. A domain is heterogeneous when the phenomena that it includes can be explained by or included in the scope of, several different types of explanations. Heterogeneous domains are constitutive of explanatory pluralism. In these cases, it may be possible to decrease this pluralism by splitting up, or partitioning, a heterogeneous domain into more homogeneous sub domains. In some cases this will work and in some it won’t.

There are at least two ways that heterogeneous domains can be arranged. A domain is heterogeneous in a hierarchical sense when it includes separate and mutually exclusive sub domains, each of which is explained by a different type explanation (see figures 2 and 3). We can think of these sub domains as existing in a sort of peas in a pod arrangement of several sub domains included within a single domain. In this case there will be less pluralism when we explain a sub domain than when we explain a domain. A second way in which domains can be heterogeneous is in an orthogonal sense (see figure 4). In this case, the sub domains are overlapping and the explanations of these domains are interwoven.

3.2 Domains can change over time.

The domains into which researchers divide phenomena are changeable and dynamic. A domain could be heterogeneous because researchers have made an error in grouping phenomena or have changed their theoretical commitments and decided that phenomena should be grouped differently in terms of the sorts of explanations that should be used to account for them. For example, see figure one (1), at time T₁ researchers may have considered P₁ to be uniform, and hence part of a homogeneous domain, D_l, and therefore sought to explain it using one kind of explanation. At time T₂ researchers may find that different instances of P₁ have different explanations and hence that D_? is not homogeneous but rather is heterogeneous. D_? contains P₁ and P₂, and for a time this domain could have two explanations, and hence be a case of explanatory pluralism. In this sort of case, pluralism can be decreased by partitioning D_l into two domains: D₂ containing P₁ which could be explained with one explanation, E₁, and D₃ contained P₂ which could be explained with a second explanation, E₂. In this situation domains can change over time and heterogeneity can be decreased by this type of partitioning.

To use Allchin's map analogy, one would notice that the borders of territories are always shifting. If one only looks at the grouping of phenomena into a domain in terms of the characteristics of the phenomena themselves, one might be led to think that the changes in the structure of domains will be cumulative: a change from incorrect or indeterminate boundaries to the correct boundaries. In other words, once researchers got the domain right, they could find the single, 'right' explanation for that domain. But, scientific change is not always cumulative, and phenomena are not divided into domains solely in terms of the immediately observable characteristics of those phenomena. An important contribution that Shapere made to considerations of domains was to point out that they are merely hypotheses as to the scope of particular explanations and as such are open to rejection. The replacement of one domain with another doesn't entail that they always change from being incorrect to being correct, or even more correct.

There are many ways to consider the rejection of previously accepted domains, a change in theories being a relatively minor one. Changes in terms of revolutions or paradigm shifts are a more radical, but also a well theorized, way of showing that domain partitioning need not be permanent (Kuhn 1970). Domains change over time and this is not always a change from an incorrect or vague conception of a domain to a correct and clear description of a domain, but can be a change from a domain that is acceptable in one theoretical context to a domain that is acceptable in another theoretical context. The example of the evolution of sex reveals that this change is not always in the direction of increasing homogeneity.

3.3 Changing domains and sex

Controversy concerning the evolution of sex was decreased by the realization that there could be different explanations for different parts of the history of sex. In particular, that the series of events that led to the construction of the complex phenomena that make up sex could be very different from the evolutionary forces that maintain sexually reproducing populations. Williams (1975) and Maynard Smith (1988b) argue that it is possible that evolution by natural selection could be responsible for the origins of sex, but that its maintenance may be due to constraint, or to the fact that it is very difficult to evolve from a sexual to an asexual mode of reproduction.

In an exchange between Bernstein et. al. (1988) and Maynard Smith (1988a), the separation of the evolution of sex into two domains, origins and maintenance, led to a decrease of pluralism (a decrease in the number of explanations applied to a particular domain). Bernstein et. al. argued that the evolutionary benefit of sex is DNA repair and Maynard Smith argued that the evolutionary benefit is predominantly the creation of genetic variation. Bernstein' s theory maps more easily onto the origins of, as opposed to the maintenance of, sex. Maynard Smith applied his variation view to the maintenance of sex. Maynard Smith conceded that DNA repair is the most likely explanation for the origin, but argued that variation is the best explanation for the maintenance of sex.

In this case, domain partitioning led to the coexistence of two different explanations for sex because those explanations were applied to different parts of the evolutionary history. Although this partition is well accepted and decreases pluralism, it is not universally agreed upon. In the recent work of Long and Michod (1995), this origins/maintenance partition is contested. These researchers model a common mechanism, the DNA repair hypothesis, for the evolution of sex, looking both at issues of origins and persistence of sexual reproduction. This is an area of contention that I raise to make the point that even the most fruitful sorts of partitioning are hypothetical and subject to refutation. Here is a case in which domain partitioning has decreased pluralism with respect to the number of explanations referring to a particular domain, but has also proved to be reversible. In the next two sections of the paper I will complicate this story, arguing that there are cases in which pluralism will obtain because a phenomenon can be a part of many coexisting domains and in some cases.

3.4 Simple hierarchical domains

I previously showed that domains can change over time in a way that is related to theory change. In this section I argue that domains can also change depending on the theoretical perspective from which one considers particular phenomena. Several theories concerning the same phenomenon can be concurrently accepted as explanatory in different fields, different research programmes, or different laboratories.

Hierarchical domains are heterogeneous domains that can be partitioned or divided into sub domains the members of which are mutually exclusive. If one member of a sub domain is also a member of a second sub domain, then all of the members of the sub domain will also be member of the second sub domain. In hierarchically arranged domains, domains at the same level in the hierarchy either completely overlap or are completely separate. In the simplest case one phenomenon or each member of a group of phenomena requires more than one explanation to account for it. The phenomenon or group of phenomena can be included in two different domains at the same time, each domain pertaining to a different explanation. For example, if I were asking for an explanation for why my cat has the reproductive system that it does, I could respond with an adaptation explanation, by pointing out that its ancestors were selected because they had that particular phenotype, or I could give a developmental explanation, by pointing out the developmental steps that occurred in the formation of the reproductive system. One phenomenon, my cat’s reproductive system, can be accounted for by two different explanations, can be a member of two domains and hence represents a case of explanatory pluralism.

Figure two (2) represents a generalization of this case: Consider a domain, D_l, which is the group of phenomena P_l, which are included within the scope of explanation, E_l. A domain, D₂, is the group of phenomena P₂, which are included within the scope of a single explanation, E₂. D_l can be the same as D₂ in which case P_l will be the same as P₂, but E_l and E₂ can be different, although not conflicting. D_l and D₂ are the same domain which is heterogeneous because E_l and E₂ are different. Simple hierarchical domains can be used to characterize pluralism that is the result of ambiguous explanation-seeking questions or is the result of different researchers finding different kinds of answers explanatory.

3.5 Complex hierarchical domains

In simple hierarchical domains, every phenomenon being considered can be grouped into more than one domain and these domains are coextensive. But a more complex case can obtain in which a heterogeneous domain can be partitioned into several sub domains that don’t overlap with one another. Refer to figure three (3). Consider a similar situation in which separate domains, D₂... D_n, are completely contained within D_l, in which case each group of phenomena, P₂... P_n, will be subsets of P₁, and the explanation, E₁, will be different from the explanations, E₂...E_n. The differences between E₁ and E₂...E_n are at least twofold, E₁ explains more phenomena than anyone of E₂…E_n does and E₁ explains these phenomena at a higher level of abstraction than any explanation E₂...E_n does.[15] With this sort of a peas-in-a-pod arrangement, the domains at any particular level of the hierarchy are autonomous and mutually exclusive. If one member of a domain at a lower level of the hierarchy is a member of a domain at a higher level, then all of the members of that lower level domain are also members of the higher level domain.

Using Allchin' s map analogy, in a hierarchical arrangement of domains, there can be different maps of a territory, but in this case, the boundaries of those various maps don't cross one another. In this sort of arrangement, if one chooses to explain only the domains at the lowest level of the hierarchy, then explanatory pluralism can be decreased or even eliminated. However, domains at higher levels may also merit explanation. Depending on the level of abstraction with which these domains are considered, we can have one unified abstract domain, D_l, or we can have several separate and mutually exclusive less abstract domains, D₂... D_n. It is important to note that the different levels of this hierarchy are often described by different scientific disciplines with different standards of explanation. In this case there may be two complete explanations of a particular P_i, one for each domain that it is described as a member of.[16]

Beatty’s gene regulation is amenable to this sort of analysis. Gene regulation mechanisms in general would be included in the more general or greater domain and that domain could be partitioned such that each regulation mechanism was grouped into its own sub domain. We can make this more specific to the evolution of sex. The domain that we are asking about could include all cases of sexual reproduction, but all of these cases of sexual reproduction could be partitioned such that each taxonomic group was included in its own domain. For example, three sub domains could be sex in guppies, sex in mold, and sex in species X and these could be explained by the Red Queen, DNA Repair, and explanation X, respectively. In this situation, one could argue that it may be the case that asking for an explanation of sex in general is asking a misguided question and that we should just be looking for explanations of the various sub domains, sex in guppies, mold, and species X.

If our goal is to follow the Newtonian Ideal and to decrease pluralism as much as possible, we could argue that we should only explain the phenomena in terms of the most homogeneous sub domain possible. Sometimes this is appropriate, but not always. In response to this possibility, it is necessary to consider whether or not D₁ is a domain that merits explanation at all. Do we want there to be such a thing as E₁? In other words, why should one explain a heterogeneous domain rather than just providing explanations of homogeneous sub domains?

There could be many reasons why one would choose to explain a heterogeneous domain as opposed to a homogeneous sub domain. There is some sort of similarity among gene regulation mechanisms: they all regulate gene expression. All of the cases of sexual reproduction do involve some sort of exchange of genetic information between the two parents. But, in both of these cases, if one was bent on implementing the Newtonian Ideal, one could decrease pluralism by choosing to explain the most homogeneous domain possible. The question of the evolution of sexual reproduction, differs from the question concerning gene regulation mechanisms. Explanations of both of these cases can be pluralistic if researchers use more general or abstract or inclusive criteria for domain membership, and in both of these cases, pluralism can be decreased by more specifically identifying the domain. But in the case of sex there is another cause of pluralism that does not obtain in the case of gene regulation mechanisms. One can expect to find many explanations of sex, and sexual reproduction is a domain that has unity, because the sub domains that we can partition it into are overlapping and intricately interconnected. The domain has unity because its sub domains are not organized in a peas in a pod arrangement as they would be if we embraced Beatty’s concept of pluralism, but are rather arranged orthogonally to one another.

3.6 Orthogonal domains

When one partitions hierarchically heterogeneous domains, the relationship among sub domains is all or nothing. At a particular level in the hierarchy, either one domain contains all of the members of another domain or it contains no members of the other domain. As a result, this sort of heterogeneous domain can be partitioned into more homogeneous sub domains, and if researchers choose to explain the sub domains, pluralism can be decreased. In order to capture the details of the case of the evolution of sexual reproduction, we need to employ orthogonally heterogeneous domains. When this sort of domain is partitioned into sub domains, the sub domains are partially overlapping. Some members of one sub domain are included in a second sub domain, and some aren’t.

Figure four (4) demonstrates this relationship among domains. In this case, a domain, D_l, is the group of phenomena, P_l, that are included within the scope of explanation, E_l. A domain, D₂, is the group of phenomena, P₂, that are included within the scope of a single explanation, E₂. D_l can be similar to D₂ where this similarity is defined as a situation in which most but not all of P_l are the same as P₂, and most but not all of P₂ are the same as P_l, but E_l and E₂ can be different. In this case there can be partial overlap among domains, to say it in other words, domains can be arranged orthogonally to one another.

To return to the case of sex, sexual reproduction is the general domain that researchers are trying to explain. Sex can be partitioned into two domains, based on the different parts of a sexual mating system. One domain consists of the mating systems of those organisms that undergo meiosis and the second domain consists of the mating systems of those organisms that undergo outcrossing. The first is explained by the DNA Repair hypothesis, and the second is explained by the Red Queen hypothesis. Most organisms that undergo outcrossing also undergo meiosis because outcrossing doubles the number of chromosomes and meiosis halves it. But it is not always the case that these two processes occur together. There can be other ways of fulfilling the same sort of function. Some organisms undergo meiosis but don’t out cross, some organisms undergo meiosis and out cross, and some organisms out cross with gametes that are not produced by meiosis.

This case of orthogonally related domains cannot be explained by a simple conjunction of E_l and E₂ because there will be some P_i that will not be included in the conjunction, namely those that are explained solely by either E_l or E₂. The phenomena explained by E_l and E₂ cannot simply be separated because there is significant overlap between these two domains; the overlap itself being something that merits attention. In some cases meiosis needs to be explained in conjunction with out crossing and in some cases it doesn’t. In the cases where these two processes occur together, they are both needed to form a functioning mating system. In cases where they don’t occur together, there is often some other way of decreasing or increasing the ploidy of the gametes and offspring, such that it remains stable from generation to generation. One is left with a case of unavoidable pluralism.

A critic, who was bent on decreasing explanatory pluralism may still object that even this case of orthogonally related domains is not really a case of pluralism because we can decrease the pluralism by partitioning these domains into more homogeneous sub domains. This seems like a promising move because it was successful in the cases of hierarchically organized domains. For example for a period of time pluralism in regard to the explanation of sex was decreased by making a division between explanations for the origins of sex and explanations for the maintenance of sex. The former being explained by the DNA repair hypothesis and the later being explained by the Red queen hypothesis. In this aspect of the evolution of sex it was possible to create two mutually exclusive sub domains. This critic could go on to argue that even in the case of orthogonal domains partitioning could be used to decrease pluralism as shown in figure five (5). One could argue that this case ought to be partitioned into three domains, namely: in domain X, P_x is explained by E_l; in domain Y, P_y is explained by E_l and E₂; and in domain Z, P_z is explained by P_z. I argue that this is not a good partition to make. Although this partitioning decreases (but still does not eliminate) pluralism, it does violence to the possibility that it may be more meaningful to distinguish between P_l and P₂ than between Px, Py and Pz, even though P_l and P₂ are the subject of explanatory pluralism and Px, Py and Pz are not.

Even though domain partitioning could be used to decrease pluralism and so can be a method by which we can conform more closely to the Newtonian Ideal, it should not always be used in this way. In this case, it may mean something for a phenomenon to be P_l. In this case, P_l is meiosis, and whether it is explained in conjunction with outcrossing or not, it is a very meaningful biological category. It has a single origin and its instances are remarkably well conserved throughout various taxa. Although one could brutishly decrease pluralism by partitioning the domain, but it would be doing violence to the unity of the phenomena of meiosis. Pluralism cannot be decreased when the domain under consideration is heterogeneous in an orthogonal sense, its unity being justified by the overlapping and interconnected nature of its sub domains. According to the Newtonian Ideal, pluralism is something that should be decreased at all costs. But in the case of the evolution of sexual reproduction, pluralism is a virtue because it more accurately represents the complexity of the system that is being explained. Even though domain partitioning can be useful in terms of clarifying scientific problems, it is not a panacea for eliminating pluralism, and in fact, when considering complex evolutionary phenomena such as the evolution of sex, I argue that decreasing pluralism ought not to be an unconditionally accepted goal. To best account for sex, pluralism needs to be systematized, not eliminated.

Figure 1. Domains change over time.

At time T₁ researchers may have considered P₁ to be uniform, and hence part of a homogeneous domain, D_l, and therefore sought to explain it using one kind of explanation. At T₂ researchers may find that different instances of P₁ have different explanations and hence that D_l is not homogeneous but rather is heterogeneous. D_l contains P₁ and P₂, and for a time this domain could have two explanations, and hence was a case of explanatory pluralism. At T₃, pluralism can be decreased by partitioning D_l into two domains, D₂ containing P₁ which could be explained with one explanation, E₁, and D₃ contained P₂ which could be explained with a second explanation, E₂.

D1

T1

D?

T2

D2 D3

T3

Figure 2. Simple hierarchical domains.

A case of explanatory pluralism in which a domain, D₁, is the group of phenomena, P₁, that is included within the scope of a single explanation, E₁. D₂ is the group of phenomena, P₂, that is included within the scope of a single explanation, E₂. D₁ is the same as D₂, and P₁ are the same as P₂, but E₁ and E₂ are different.

D1 = D2 = D

Figure 3. Complex hierarchical domains.

A case of explanatory pluralism in which domains, D₂…D_n, are included within a domain, D₁, in which case, each groups of phenomena, P₂…P_n will be subsets of P₁, and the explanation E₁, is different from the explanations, E₂…E_n.

D₁

D₂

D₃

D_n

E₁ = E₂ & E₃ & E_n .

Figure 4. Orthogonal domains.

A case of explanatory pluralism in which a domain, D₁, is the group of phenomena, P₁, that is included within the scope of explanation E₁. A domain, D₂, is the group of phenomena, P₂, that is included within the scope of explanation E₂. D₁ is similar to D₂ insofar as most but not all P₁ are also P₂ and most but not all P₂ are also P₁. E₁ and E₂ are different.

E₂P₂

E₁ P₁

D₁