1 Russell Powell Duke University IS CONVERGENCE MORE THAN AN ANALOGY? APPEALING TO THE FREQUENCY OF HOMOPLASY AS EVIDENCE FOR A NON-CONTINGENTLY CONSTRAINED ADAPTATIONAL DESIGN SPACE ABSTRACT It is widely accepted among contemporary biologists and philosophers of biology that replaying the proverbial ‘tape of life’ would result in wildly unpredictable and radically different evolutionary outcomes. This position arises primarily from the empirical assumption that the invariant laws of nature carve out exceedingly broad channels that only loosely constrain the evolution of organismic design. A number of authors, however, have touted examples of ‘convergent evolution’ as evidence par excellence for the central role of adaptation in shaping predictable trajectories of macroevolution. Of these individuals, Simon Conway Morris (“SCM”) is arguably the most prominent champion of homoplasy and its purported implications for a non-contingent, counter- factually stable view of life. However, there are numerous conceptual and empirical difficulties that arise in appealing to the frequency of homoplasy as evidence for a non- contingently constrained adaptational design space. These problems are best illustrated by critically evaluating SCM’s use of the homoplasy literature in an attempt to bolster a strong adaptationist view of life, and thus his two major and most recent works dedicated to this end will be the focal point of my critique. Nonetheless, this paper is not intended to be a negative excoriation of SCM’s work; rather, it is designed to transform a less systematized approach to the philosophical implications of homoplasy into a positive and rigorous research program in counterfactual biology within which the relative significances of contingency and selection in shaping the history of life can be assessed. Specifically, my critique will be based primarily on five grounds: (1) SCM’s analytically and methodologically inadequate treatment of the role of homoplasy in phylogenetic reconstruction, including the fact that homoplastic pervasiveness confounds the very process by which it is identified; (2) SCM’s express failure to make the crucial distinction between convergent evolution (due to external, non-contingent constraint) and parallel evolution (due to internal, contingent constraint), and to consider how the respective frequencies of these importantly different sources of homoplasy affect his proffered view of life; (3) the presence of deep homology and its implications for macroevolutionary / exo-biological extrapolation; (4) the general failure to specify hierarchical levels of analysis in assessing homology and homoplasy, which only exacerbates an already confused and cross-talking discourse on evolutionary convergence, and (5) difficulties in quantifying or otherwise operationalizing the degree of homoplasy so as to assess its relative significance in and between given lineages. Of these five major points of criticism, the second looms most philosophically large and will thus form the heart of the critique.
46
Embed
Is Convergence More Than an Analogy - Final First Draftand stochastic processes play a de minimus role in shaping macroevolution over deep geological time, stating that such mechanisms
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
Russell Powell Duke University
IS CONVERGENCE MORE THAN AN ANALOGY? APPEALING TO THE FREQUENCY OF HOMOPLASY AS EVIDENCE FOR A NON-CONTINGENTLY CONSTRAINED ADAPTATIONAL DESIGN SPACE
ABSTRACT
It is widely accepted among contemporary biologists and philosophers of biology that replaying the proverbial ‘tape of life’ would result in wildly unpredictable and radically different evolutionary outcomes. This position arises primarily from the empirical assumption that the invariant laws of nature carve out exceedingly broad channels that only loosely constrain the evolution of organismic design. A number of authors, however, have touted examples of ‘convergent evolution’ as evidence par excellence for the central role of adaptation in shaping predictable trajectories of macroevolution. Of these individuals, Simon Conway Morris (“SCM”) is arguably the most prominent champion of homoplasy and its purported implications for a non-contingent, counter-factually stable view of life. However, there are numerous conceptual and empirical difficulties that arise in appealing to the frequency of homoplasy as evidence for a non-contingently constrained adaptational design space. These problems are best illustrated by critically evaluating SCM’s use of the homoplasy literature in an attempt to bolster a strong adaptationist view of life, and thus his two major and most recent works dedicated to this end will be the focal point of my critique. Nonetheless, this paper is not intended to be a negative excoriation of SCM’s work; rather, it is designed to transform a less systematized approach to the philosophical implications of homoplasy into a positive and rigorous research program in counterfactual biology within which the relative significances of contingency and selection in shaping the history of life can be assessed. Specifically, my critique will be based primarily on five grounds: (1) SCM’s analytically and methodologically inadequate treatment of the role of homoplasy in phylogenetic reconstruction, including the fact that homoplastic pervasiveness confounds the very process by which it is identified; (2) SCM’s express failure to make the crucial distinction between convergent evolution (due to external, non-contingent constraint) and parallel evolution (due to internal, contingent constraint), and to consider how the respective frequencies of these importantly different sources of homoplasy affect his proffered view of life; (3) the presence of deep homology and its implications for macroevolutionary / exo-biological extrapolation; (4) the general failure to specify hierarchical levels of analysis in assessing homology and homoplasy, which only exacerbates an already confused and cross-talking discourse on evolutionary convergence, and (5) difficulties in quantifying or otherwise operationalizing the degree of homoplasy so as to assess its relative significance in and between given lineages. Of these five major points of criticism, the second looms most philosophically large and will thus form the heart of the critique.
2
INTRODUCTION
It is widely accepted among contemporary biologists and philosophers of biology that
replaying the proverbial ‘tape of life’ would result in wildly unpredictable and radically different
evolutionary outcomes. Some biologists even espouse the more radical notion that virtually
every interesting event in the history of life, including the evolution of the most robust complex
adaptive solutions, falls into the realm of historical contingency. This position arises primarily
from the empirical assumption that the invariant laws of nature carve out exceedingly broad
channels that only loosely constrain the evolution of organismic design. On this view, observed
evolutionary products are sensitively dependent on stochastic initial conditions, a feature of
macroevolutionary processes which undermines attempts to formulate robust generalizations
regarding the evolution of organismic form. Because even the most successful biological
generalizations are undermined by multiply realizable solutions to contingent design problems
(Beatty, 1995) that are destined for obsolescence in the unrelenting arms race of natural selection
(Rosenberg, 2001; Van Valen, 1973), they tend to lack the nomic necessity, counterfactual
stability, and qualitative predicates canonically characteristic of natural laws.1
With this nomological vacuum as the backdrop, historicists tend to explain the
inhomogenous distribution of organismic form in morphospace (at all levels of the genealogical
hierarchy) as the result of internal developmental constraints that restrict the set of possible
variants on which selection can operate. Strong adaptationists, on the other hand, tend to view
the clumping of actualized morphology not as a contingent consequence of developmental
canalization, but rather as an ecologically optimal set of solutions to relatively stable functional
1 There are two notable exceptions to this rule: the Principle of Natural Selection (Rosenberg, 2001) and the Principle of Drift (Brandon, 2005, in press), neither of which, however, allow for macroevolutionary predictions regarding the evolution of form. In fact, both selection and drift may be seen as undermining macroevolutionary extrapolation.
3
problems. In defending this view, a number of authors have touted examples of ‘convergent
evolution’ as evidence par excellence for the central role of adaptation in shaping predictable
trajectories of macroevolution.2 Of these individuals, Simon Conway Morris (“SCM”) is
arguably the most prominent champion of homoplasy and its purported implications for a non-
contingent, counter-factually stable view of life. In essence, his view is this: The ubiquity of
‘convergence,’ or the independent origination of similar forms in distantly related organisms, is
prima facie evidence for an adaptational design space that is so severely constrained by the
invariant chemico-physical laws that organismic form will, through the optimizing forces of
natural selection, repeatedly and inevitably converge on certain identifiable functional attractors
or biological properties. Thus, according to SCM, the program of universal biology should be
primarily aimed at identifying these elusive ‘laws of convergence’ which operate beneath the
surface of a superficially stochastic history.
However, there are numerous conceptual and empirical problems that arise in appealing
to the frequency of homoplasy as evidence for a non-contingently constrained adaptational
design space. These problems are best illustrated by critically evaluating SCM’s use of the
homoplasy literature in an attempt to bolster a strong adaptationist view of life, and thus his two
major and most recent works dedicated to this end (2003 / 1998) will be the focal point of my
critique. Nonetheless, this paper is not intended to be a negative excoriation of SCM’s work;
rather, it is designed to transform a less systematized approach to the philosophical implications
of homoplasy into a positive and rigorous research program in counterfactual biology within
which the relative significances of contingency and selection in shaping the tree of life can be
assessed.
2 See e.g. Conway Morris (2003); Foley (1999); Dennett (1995); Patterson (1988).
4
Specifically, my critique is based primarily on five grounds: (1) SCM’s analytically and
methodologically inadequate treatment of the role of homoplasy in phylogenetic reconstruction,
including the fact that homoplastic pervasiveness confounds the very process by which it is
identified; (2) SCM’s express failure to make the crucial distinction between convergent
evolution (due to external, non-contingent constraint) and parallel evolution (due to internal,
contingent constraint), and to consider how the respective frequencies of these importantly
different sources of homoplasy affect his proffered view of life; (3) the presence of deep
homology and its implications for macroevolutionary / exo-biological extrapolation; (4) the
general failure to specify hierarchical levels of analysis in assessing homology and homoplasy,
which only exacerbates an already confused and cross-talking discourse on evolutionary
convergence, and (5) difficulties in quantifying or otherwise operationalizing the degree of
homoplasy so as to assess its relative significance in and between given lineages. Of these five
major points of criticism, the second looms most philosophically large and will thus form the
heart of the critique.
SIMON CONWAY MORRIS’ CORE CONTENTION: PERVASIVE HOMOPLASY IMPLIES THE INEVITABILITY OF EVOLUTIONARY ENDPOINTS
Biologists and philosophers have described the evolutionary phenomena of convergence
as nature’s way of re-winding the proverbial tape of life (Dennett, 1995), biology’s closest
analog to independent experimental replication (Gould, 1976; but see Gould, 2002). As one
preeminent comparative physiologist suggests, the project of identifying convergence offers
more than just evidence for adaptation (Endler, 1986), for it enables biologists to distinguish
functional complexes that strongly determine structural form from those that are less important
in design, increasing the predictive power of generalizations regarding the characters that matter
5
most (Vogel, 1996 / 1998). In The Crucible of Creation: The Burgess Shale and the Rise of
Animals (1998) and more recently in Life’s Solution: Inevitable Humans in a Lonely universe
(2003), SCM sketches a view of life that is somewhat unusual from a modern philosophical and
biological standpoint—one that in several unique respects breaks rank with traditional
interpretations of macroevolutionary patterns. Pointing to the putative pervasiveness of
‘convergence,’ SCM defends the strong adaptationist view that life inexorably navigates toward
certain pre-ordained “endpoints,” to which “the routes are many, but the destinations few, and
the landscapes across which all organisms must travel are adaptive” (2003, p. 297). Thus, in
contrast to the contingency-flavored philosophical paradigms of biological lawlessness (e.g.
Beatty, 1995), SCM holds that if only biologists could identify these elusive ‘laws of
convergence’ operating beneath the surface of stochastic chaos, they would be able to divine
robust, counterfactually stable predictions regarding the evolution of life on Earth and
throughout the Universe.3
Pit against such a radical functionalist weltbild is the ‘historicist’ (sensu Gould, 2002)
view of the history of life, which views the inhomogeneous distribution or ‘clumping’ of
organisms across morphospace (noted by Dobzhansky, 1951) not as an optimal set of solutions
(courtesy of natural selection and the invariant physical laws) to functional problems, but as the
result of internal, contingent constraints restricting the realm of the possible.4 Rather than a
3 Ironically, while Conway Morris holds that the emergence of sentience is essentially inevitable once life has begun, he argues that there are extremely few places in the universe with conditions suitable for life (hence his titular construct, ‘Inevitable Humans in a Lonely Universe.’ There are, of course, numerous critiques of this unusual view, including a direct challenge by Gould (1998). Additionally, many structuralists believe that the universe is ‘pregnant’ with life (Kaufmann, 1995) and quasi-life (Salthe, 1994) which emerge from complexity dynamics ‘for free’ without the operation of natural selection. However, these structuralist contentions apply more to simpler forms of life, and they may have less (or nothing) to say about the inevitability of robust sentience or self-awareness. 4 A third macroevolutionary world view is ‘structuralism’ (see note above) which like adaptationism is enamored of homoplastic phenomena, but attributes its ubiquity not to the supremacy of selection but to non-functional (indeed, non-biological) laws of complexity (see infra note 15). Of course, these views are not mutually exclusive, as the
6
predictable march of organismic form toward identifiable optimality or equilibrium, the
historicist’s history of life is (as Henry Ford was fond of saying) simply ‘one damn thing after
another’—an accumulation of accidents, perhaps intelligible in hindsight but wildly
unpredictable in prospect, fluctuating in rhythm with a stochastic ecology and exhibiting no bias
toward any particular functional solution.5 Of course, historicists agree with adaptationists that
natural selection is the only known mechanism for producing function, although they downplay
the role of selection in directing macroevolutionary patterns.6
While biologists and philosophers of biology quibble over the respective significance of
their favored evolutionary phenomena and attempt to relegate their opponent’s preferred
processes to a relevant but less significant role, SCM goes even further to declare that contingent
and stochastic processes play a de minimus role in shaping macroevolution over deep geological
time, stating that such mechanisms are completely “irrelevant…in so far as the history of life is
concerned” (1998, p. 205). This is a strong and iconoclastic claim, but in fact his assertions are
even more radical, as he refers to the notion that contingent forces drive evolutionary processes
as a biological fundamentalist “myth” (2003, p. 322). These views alone provide sufficient
grounds for labeling SCM as a hard adaptationist (sensu Amundson, 1994), as they entail that
geometrical, developmental, and selective constraints on bivalve / brachiopod accretionary growth make clear—instead, the debate centers on their respective significance in the history of life. 5 Thus, the historicist maintains, the vertebrate body plan is the result of a frozen accident—namely, a series of quirky stochastic events that permitted the competitively weak vertebrate ancestral lineage Pikaia to survive the early Paleozoic extinctions and ultimately radiate into 45,000 extant species. This is of course the upshot of the popular bible of contemporary historicism, Gould’s Wonderful Life (1989), which attributes the dearth of law-like generalizations in functional biology to the fact that evolutionary endpoints, rather than inevitable, are sensitively dependent on initial conditions and thus counterfactually instable. That is, negligible stochastic differences in early conditions result in substantially different macroevolutionary trajectories through deep geological time. According to Gould, these trajectories are prospectively unpredictable, either in principle or due to insurmountable epistemic limitations, but can be adequately re-constructed and fully appreciated in hindsight. 6 For instance, historicists tend to attribute phylogenetic trajectories to stochastic models rather than competitive interactions (see e.g. Gould & Calloway, 1980). Nevertheless, the crux of the historicist dispute with SCM is not the relative significance of selection per se but rather that the former attributes all macroevolutionary change—whether functionally or stochastically driven—to contingent rather than ineluctable events.
7
(virtually) all scientifically interesting features of life are of adaptive provenance, with selection
overpowering any developmental constraints or stochastic tendencies such as genetic drift.7
Hard adaptationism assumes that adaptational design space is highly constrained externally, that
is by directional and subsequent stabilizing selection, the latter containing genetic drift once a
(local) optimality has been achieved. However, it also presupposes that morphospace is virtually
unconstrained internally, namely by postulating the inexorable tendency of natural selection to
overcome developmental constraints that would otherwise lead to the ubiquity of adaptive sub-
optimality. For the purposes of argument, I will simply assume the existence of strong external
(ecological) constraints on the evolution of form. It is primarily the latter assumption regarding
the flexibility of developmental channels that will be the basis of my challenge. Before
launching into the substantive critique, however, I will briefly address a methodological concern
regarding phylogenetic reconstruction in the context of concerted homoplasy.
1. THE ROLE OF HOMOPLASY IN PHYLOGENETIC RECONSTRUCTION: WHY HOMOPLASTIC PERVASIVENESS CONFOUNDS THE VERY PROCESS BY WHICH IT IS IDENTIFIED
Since all organisms presumably share a common ancestor, and the differences between
species are the result (in part) of descent with modification, biologists endeavor to reconstruct a
branching tree of life, first envisioned by Darwin (1859), which reflects the recency of common
ancestry. Generally, the ability to infer past states from present states depends on the present
7 According to Amundson’s (1994) terminology, ‘hard adaptationists’ may be distinguished from ‘soft adaptationists’ who agree that (virtually) all scientifically interesting features of organismic form are subject to natural selection, but disagree with the notion that selection can break through any and all developmental constraints to achieve (non-local) optimality. ‘Neutralists,’ on the other hand, contend that many interesting traits are not operated on by natural selection (such as the mammalian ear bone shape), and thus not all constraints on form are constraints on adaptation.
8
states’ degree of ‘contingency,’ or sensitive dependence on initial conditions. If the present state
is highly multiply realizable, that is, could have been effected by an unmanageably large set of
distinct initial (past) conditions, historical scientists will be unable to map the present state onto
relevant features of a discrete and identifiable past state. Sober (1988) makes the following
helpful analogy: consider a ball released on the rim of an even bowl; as it moves either down or
up the bowl’s side, we may infer from those various states something about its starting position.
However, once the ball comes to rest (‘equilibrates’) in the middle of the bowl, the information
regarding its initial conditions is effectively destroyed, and nothing can be inferred about its past
states. This example is useful because it demonstrates that difficulties in extrapolating from
present to past states need not necessarily be due to epistemological limitations in handling the
complexity of information.
With respect to non-accidental macroevolutionary generalizations, there is an asymmetry
or inverse correlation between our ability to reconstruct the past and our capacity to project into
the future, assuming the relative significances of adaptation and contingent processes are
unchanging through time. This is because the recognition of a dominant role for contingency, or
counterfactual instability, undermines the potential for robust future extrapolation. At the same
time, however, contingent (non-equilibrial) phenomena tend to preserve historical signatures and
thereby make past-states inferable from present ones. The greater the sensitivity to initial
conditions—that is, the more intricately grooved the bowl of morphospace—the greater our
precision in pinpointing previous evolutionary states. In fact, it is the quirky trace of history that
forms the epistemological base of Darwin’s central homological argument—namely, that nested
sets of shared of traits (identified by Richard Owen in the pre-Darwinian era) reflect genealogical
relationships. To the contrary, homoplastic events tend to wipe out evolutionary memory by
9
erasing the signature of history. Although radical contingency need not necessarily preserve
historical information, an intricately carved bowl is more likely to retain evolutionary memory
than one with sleek contours and a strong attractor at the center of the bowl base.
Thus, natural selection is not only the principle engine of evolution, but also the major
source of phylogenetic distortion (Vogel, 1996; Endler, 1986). If SCM is correct, and many or
most of the observed evolutionary ‘endpoints’ are highly multiply realizable, the degree of
certainty with which we can abduct propinquity of descent declines dramatically.8 That is, if
homoplasy is not only robust and concerted on occasion, but in fact ubiquitous at all hierarchical
levels as SCM would have us believe, then this would call into question core methodological
assumptions underlying modern phylogenetic reconstruction.9 Darwin found such a view
implausible, stating (1860) that
[I]t is incredible that the descendents of two organisms, which had originally differed in a marked manner, should ever afterwards converge so closely as to lead to a near approach to identity throughout their whole organization. If this had occurred, we should meet with the same form, independently of genetic connection, recurring in widely separated geological formations; and the balance of evidence is opposed to any such admission.
The abductive precept of ‘parsimony’ holds that entities should not be postulated without
necessity. In the context of phylogenetic reconstruction, parsimony generally dictates that the
8 It is true that the signature of history can be maintained by continuity (or lack there of) in genetic sequences that are orthogonal to morphological evolution or upon which organism-level homoplastic events multiply supervene. Most genetic mutations are neutral, and organismic change need not track genetic alteration (Vogel, 1996). Nevertheless, phylogenetic reconstructions based solely on molecular-genetic comparisons run into difficulties of their own, and inferring evolutionary relationships from shared derived macroscopic character traits (synapomorphies) remains the most effective and widespread means of cladistic reconstruction. In fact, SCM (1994) is himself somewhat skeptical of molecular methods of phylogenetic reconstruction, and thus relies heavily on morphological-level analysis to identify putative instances of homoplasy. Furthermore, he dedicates portions of his 2003 treatise to arguing for molecular-level convergence (e.g. pgs. 295-98), a phenomenon if as widespread as he claims would further undermine our ability to reconstruct the past and consequently impede the detection of homoplastic events. 9 There are, however, other methods of phylogenetic reconstruction which do not depend on parsimony, such as ‘maximum likelihood estimation’ (Fisher, 1912), which involves statistically based inferences about the probability distribution of a given data set.
10
best phylogenetic hypothesis, consistent with a relevant data set, is the one that assumes the
fewest homoplasious (convergent or parallel) events.10 This follows from the notion that
postulating a single cause (e.g. homology) is more empirically plausible than relying on the
existence of multiple causes (e.g. convergences) for the presence of an identified similarity, or
shared derived trait (Hennig, 1966).11 Thus, the crux of cladistic reconstruction hinges on
identifying synapomorphies as derived homologies, thereby ruling out homoplasy (independent
origin). This methodological preference for derived characteristics is based on the evolutionary
assumption that derived character traits are usually produced only once and are not lost (Sober,
1988). Even if evolution adheres to the above two assumptions, however, we must reserve a
certain probability for homoplasy and thus we cannot deduce phylogenetic relationships based on
any set of similarities.12 As such, faulty assumptions regarding the frequency and degree of
homoplasy could profoundly affect the systematist’s ability to construct an accurate phylogeny.
For these reasons, “the ubiquity of homoplasy is a major concern in phylogenetic
analysis, which ironically is its primary means of detection” (Wake, 1991, p. 544). For as Wake
(Ibid) notes, “if homoplasy is rampant, existing cladistic methods fail.” This puts SCM into a 10 In order to determine the relatedness of various species within a given monophyletic group, cladists determine the distribution of relevant ancestral (plesiomorphic) and derived (apomorphic) characteristics. They then identify symplesiomorphic and synapomorphic traits, or those ancestral and derived character states shared by the various species within the monophyletic group (comprised of a single species and all of its descendent species). Because of its single origin, the entire tree of life forms the largest conceivable monophyletic grouping—as a result, no traits are ancestral or derived in the absolute (non-relational) sense, but only with respect to a specified level of genealogical hierarchy. 11 This intuition may be rooted in the precept of likelihood similar to that offered by Reichenbach (1956) and Salmon (1984) for the postulation of common causes, the operating supposition of which is that the existence of separate causes for similar phenomena is less likely than the existence of a single cause (Sober, 1988). Nevertheless, the less parsimonious reconstruction cannot be deductively ruled out. However, in the context of phylogenetic reconstruction, the preference for homology over homoplasy seems to be grounded more in the likelihoods attached to multiple origins rather than causes, for in theory there could be unified law-like causes of homoplasy that produce similarity more often than inter-generational transmission (and by implication propinquity of descent). 12 The potential for reversions and parallelisms may be far more common between groups that share certain developmental pathways—a crucial fact that will be addressed at length in the following Section.
11
serious methodological bind, for his claims about the inevitability of certain evolutionary
endpoints hinge on our ability to identify homoplasy, which in turn depends on our capacity to
construct an accurate phylogeny (see e.g. Vogel, 1996). As noted above, however, our ability to
recognize homology and distinguish it from homoplasy turns in large part on our ability to detect
the contingent signature of history, the very phenomena that SCM endeavors to minimize. Thus,
his wholesale denial of the sensitive dependence of present evolutionary states on initial
conditions calls into question some of the most fundamental conclusions of evolutionary biology,
including the very notion of common ancestry itself.
Nevertheless, even if homoplasy is far more ubiquitous than most cladists would prefer
(an assertion made by Wake, 1991), the precept of cladistic parsimony can addresses this
problem by placing a lower but not upper bound on the amount of homoplasy recognized by a
given reconstruction—in other words, the most parsimonious tree is one that minimizes
assumptions (homoplasies), although it does not assume that such assumptions will be minimal
(Sober, 1988). I submit, however, that there is a threshold frequency of homoplasy after which
central cladistic assumptions regarding the evolutionary process are no longer justified.
Alternatively, distance measures which determine the overall genetic dissimilarity of
compared taxa may be employed in cladistic reconstruction, and could perhaps allow for
concerted macroscopic homoplasy while preserving the trace of evolutionary history through
molecular homology. Because natural selection tends to act at the organismic level, homoplasy
will be most frequently encountered in morphology, rather than sub-cellular biochemical
pathways, the vast majority of which represent ancestral similarities (Vogel, 1996 / 1998;
Sterelny, 2006). For similar reasons, the degenerate nature of the genetic code makes gene-level
convergences (where the same base sequences generate the same product) extraordinarily
12
improbable.13 Nevertheless, if there is in fact widespread molecular convergence with respect to
either the genetic code or the message—such as at the level of protein configurations as SCM
contends (2003, p. 287-298), then these alternative molecular-based methods of phylogenetic
reconstruction might be vitiated as well.14 In sum, SCM recognizes (p. 297) that homoplasy is
“viewed by most cladists as [] profoundly disruptive to their neat schemes,” but he does not seem
to realize that he cannot have his cake of ubiquitous convergence and eat it too.
2. CONVERGENT VS. PARALLEL EVOLUTION:
HOW THE RESPECTIVE FREQUENCIES OF THESE DIFFERENT SOURCES OF HOMOPLASTIC CONSTRAINT AFFECT HISTORICIST AND ADAPTATIONIST VIEWS OF LIFE
The overarching if unarticulated assumption of hard adaptationist views of life is that the
channels of adaptational design space are highly constrained. It is this constraint which restricts
the universe of potential solutions to putatively pervasive design problems to a manageable
handful that admits of prediction without a burdensome litany of ceteris paribus qualifications.
In pointing to instances of homoplasy as evidence for counterfactually stable limitations on
organismic form, however, it is imperative to recognize that there are two major classes of
constraint that could alone or jointly be responsible for the observed morphological regularities.
The first class is external to the organism and non-contingent—namely, the constraints imposed
on formal design solutions by the chemico-physical laws and its interaction with the optimizing
agency of natural selection. For instance, because of the laws of optics and the properties of
light, there may be only a handful of ways to macroscopically arrange an image-forming organ 13 As Mayr noted (1963), because functional demands supervene on a diverse molecular and biochemical base, functional constraints only limit the end-product which can be realized via many genetic and biochemical routes. 14 Specifically, Conway Morris suggests homoplastic evolution in numerous occupied areas of protein ‘hyperspace’, including (inter alia) proteases, peptidases, cytokinases, aminoacyl-tRNA synthetases, light-harvesting proteins, elastic proteins, chitin-binding proteins, and steroid signaling proteins (2002, p. 295). But see Szathmary’s review (2005) of Conway Morris (2003), in which he argues for a highly contingent peptide space.
13
of given acuity—hence its ‘convergent’ evolution up to 15 times within and between distant
phyla (Land, 1992). Similar biomechanical constraints are probably responsible for the
‘convergent’ evolution of the fusiform shape in various marine-vertebrates, and numerous other
iteratively evolved forms whose viable morphospace is constrained by and particularly sensitive
to the invariant physical laws.
A second class of constraints, well known to the structuralists and historicists but often
overlooked by post-synthesis adaptationists, are those that are internal to the organism and
contingent—namely, developmental strictures on the potential universe of formal solutions due
to the canalization or entrenchment of developmental pathways and subsequent stabilizing
selection.15 Gene regulatory networks are hierarchical, with earlier linkages having more
pleiotropic effects than the more distal fine-grained terminal processes. The former upstream
sub-circuits, which Davidson and Erwin (2006) have termed ‘kernels,’ specify the more general
domain of the developing organism. Because kernels are ‘recursively’ wired, interference with
any single kernel gene “will destroy kernel function altogether and is likely to produce the
catastrophic phenotype of lack of [a] body part” (Ibid, p. 796). The result—massive
conservation in upstream regulatory sub-circuits—best explains why so little change has
occurred at the phylum level since the establishment of body plans in the base of the Cambrian.
Adaptive change has been largely relegated to lower levels in the network, offering an internal-
constraints perspective on the decimation-diversification model championed by Gould (1989).
Moreover, the progressive entrenchment of gene regulatory networks maps well onto the clumpy
15 In discussing explanations of the robust conservation of homologues, Sterelny & Griffiths (1999, p. 228) note that stabilizing selection need not be the driving causal force, since the persistence of such traits could equally result from chance, assuming change is less frequent than stasis. Dan McShea (2005) has shown, however, that modifications (and by implication complexity) accumulate rapidly in the absence of stabilizing selection. Viewed in this way, it is (ironically) the phylogenetic preservation of organismic structure—rather than its transformation—that lies at the heart of evolutionary explanation.
14
distribution of form at all levels of the taxonomic hierarchy. In fact, it seems reasonable to
assume that the lability of homologous downstream sub-circuits probably underwrites the lion’s
share of parallel evolutionary events.
The ‘consensus paper’ of Maynard Smith et al. (1985, p. 281) suggests that development
can “serve as a directing force, accounting in part for oriented features of various trends and
patterns” (see also Alberch, 1982). While some external constraints may represent the pressures
of the physical laws on formal solutions to relatively universal design problems, internal
(developmental) constraints probably reflect not invariance but frozen contingency, or the radical
conservation of kernels which arose in response to local, stochastically fluctuating ecological
pressures in the pre-Cambrian. Unfortunately, SCM systematically conflates internal
(developmental) and external (abiotic, ecological) constraints on design space in arguing for a
fundamentally non-contingent view of life.
The most effective way of categorizing independently evolved similarities so as to reflect
the above distinction is by recognizing two causally differentiated sub-categories within the
larger rubric of homoplasy. These are convergent and parallel evolution. SCM’s only mention
of this “vital” (Simpson, 1961) distinction in his singular work on the importance of convergence
is in a brief footnote, in which he states: “Now is the time to avoid that old chestnut of whether it
is convergent evolution as against parallel evolution” (2003, p. 435 n. 1). One of the main
purposes of this paper, however, is to convince the reader that the distinction between convergent
and parallel evolution is no small chestnut. Expressly failing to acknowledge the importance of
this distinction, SCM reasons that the difference between convergence and parallel evolution is
“obviously” one of degree rather than kind (Ibid). But this conclusion is far from obvious,
particularly given the importance to many preeminent evolutionists of distinguishing these two
Patterson, 1988; Saether, 1983; Simpson, 1961; Haas, 1946). In fact, SCM’s failure to address
the implications of this distinction for his hard adaptationist view of life is perhaps the most
severe weakness in his appeal to the literature on homoplasy in order to bolster the project of a
universal biology.
To clear up perceived conceptual confusion in the literature regarding the contrast
between convergent and parallel evolution, Haas and Simpson (1946) published a seminal
review of the topic. They defined homology as similarity based on common ancestry, in
contradistinction to homoplasy, which simply refers to similarity that is not due to common
descent. The latter concept was introduced by E.R. Lankester (1870) in recognition of the
epistemic ambiguities surrounding causal origins, and thus does not make any positive claims
about common selective pressures or functional origins of the identified similarity. Use of the
word analogy, however, often used interchangeably with convergence, has come in the literature
to imply a common selected function as the underlying basis for a perceived similarity (Ibid).
Where analogy ultimately falls in the evolutionary lexicon is not terribly relevant for the purpose
of this paper, which rather is primarily concerned with whether convergence is more than an
analogy in the philosophical (rather than definitional) sense.16 Of greatest relevance here will be
the contraposition of convergence with parallel evolution, each a type of homoplasy but with
importantly different causal origins.
16 There are several possibilities concerning the definition, use, and status of the term ‘analogy.’ First, one might view analogy as a subset of parallelism or convergence, making positive claims about the selective causes of origin in either case. Alternatively, analogy may be considered a subset only of convergence, in which it makes selective-origin claims in the absence of or by factoring out underlying developmental homology. Finally, analogy may be equivalent to and synonymous with the notion of convergence, remaining neutral with regard to its underling causal mechanism.
16
Haas preferred to distinguish convergence and parallelism geometrically (rather than
causally)—with the former entailing that two lineages resemble one another more than their
ancestors did, and the latter referring to cases in which two lineages evolve in the same direction
without resembling one another any more than their ancestors did. Simpson, however, advocated
a causal differentiation of the two concepts, with convergence resulting solely from common
selective pressures, and parallelism tied to underlying homologous developmental pathways, or
what he termed a “community of common ancestry” (1961, p. 103). Thus, while Haas might
view concurrent increasing body size in Orthoptera (grasshoppers, locusts and crickets) and
Pinnipeds (seals, sea lions and walruses) as an example of parallel evolution, Simpson would
presumably decline to use the label of parallelism—or analogy for that matter—since the causal
factors involve neither shared developmental generators nor common selective pressures.
Whereas the distinction between homology and homoplasy is relatively crisp, the concept
of parallelism occupies a ‘gray zone’ between definitional homology (retention by common
descent) and convergent homoplasy (similar design and function with entirely different
too much ‘homology-ness’ to be considered solely the result of similar ecological (extra-
organismic) pressures. As Gould (2002, p. 1078) remarks, “parallelism contains aspects of both
constraint and individual selection—not as a wishy-washy mixture in one grand pluralistic glop
of all-things-for-all-people, but in rigorously different parsings for different levels of
consideration” (emphasis in the original). The distinction between parallel and convergent
evolution is in some sense one of degree (Diogo, 2005), although the decision to categorize a
homoplastic event as one or the other is not arbitrary. It is true that all organisms share
important homologues (such as the general structure of the DNA code) as they descend from a
17
single common ancestor, and thus at certain grains of analysis the distinction between parallelism
and convergence disappears (Wilmer, 2003). However, it is the underlying homology with
respect to the generators directly causally responsible for the homoplastic event that defines
parallelism and non-arbitrarily distinguishes it from convergence. Nevertheless, as Diogo (2005
p. 739) adequately points out, there are in fact significant epistemic difficulties in differentiating
the two types of homoplasy in the practical phylogenetic context.
Theoretically, parallelism exists because ‘organismic branching’ occurs before ‘genetic
branching’ in cladistic reconstructions—that is, underlying developmental generators are
conserved, while changes in developmental timing, etc. produce recognizably differentiated
morphology befitting of taxonomic partition. In essence, then, parallel evolution is a form of
homoplasy that depends in a somewhat ironic sense on the more general domain of homology
not only for its recognition (see above) but its raison d’etre.17
Developmental biologists have often stressed the idea that homologues, by definition
non-functional categories (Amundson & Lauder, 1994), are not mere evolutionary
epiphenomena, but rather are tied to a different but equally viable notion of function—namely
the casual roles of various biological processes, such as embryogenesis and developmental
anatomy. As Amundson notes (2001, p. 3), “they see homologues as indicators of the
commonness of the developmental processes that give rise to them.” Gunter Wagner (1989), for
example, emphasizes shared developmental constraint as the conceptual and definitional basis of
homology. The problem with the developmental definition of homology, however, particularly
in relation to the present discussion, is that it appears broad enough to include parallel
evolutionary events, which are also entrenched in common underlying pathways of development
17 Both functionalist and structuralist views of the evolution of form predict the occurrence of parallel evolution (Wake, 1991), although the former do so for reasons external and the latter for reasons internal to the lineages in question.
18
that constrain the ultimate set of forms on which selection can act. As such, both causal role
functions (e.g. embryogenesis) and selective effects functions must be invoked to explain the
general frequency and specific instances of parallel evolution. Indeed, the category of
parallelism not only vindicates Amundson and Lauder’s argument (1994) that causal role
functions have a role to play in the ‘proximate’ sciences of anatomy and physiology—but it goes
even further to demonstrate that causal role function is a requisite explanatory device in
evolutionary biology as well. Additionally, causal role functions must be taken into account in
delineating the set of potential morphologies that can underwrite a given function, particularly
within the confines of a specified Bauplan.18 For these reasons, causal role functions will be
integral to any explanation of the inhomogenous clumping of organisms in morphospace, which
results in large part from internal developmental constraints on the evolution of adaptive
optimality.19
For sure, it is analytically possible (though empirically intractable) to parse the
homological components of parallelism from those which result solely from factors external to
the organism—namely selective pressures constrained by the physical laws.20 Nevertheless,
assuming a single origin of life, true convergence conceptually requires the replacement of all
underlying developmental-genetic architecture over time through processes of selection and drift.
Defined in this way, convergence becomes a negligible or even non-existent phenomenon. This 18 Generally, ‘Bauplan’ refers to a deep and robustly persistent homology that restricts the potential set of variation on which selection can act, or as we shall see, makes the realization of certain forms more likely. 19 Incidentally, by explaining the uneven distribution of forms in morphospace, developmental canalization can refute one of the major critiques marshaled by Creationists against Darwinian theory—namely, the lack of extant transitional forms between clustered Bauplans at various levels of the genealogical hierarchy. 20 I would venture a guess that this intractability in parsing the relative significance of internal and external constraints on the evolution of organismic form lies at the root of the conflation of parallel and convergent evolution in the literature. However, as the nascent science of ‘evo devo’ comes of age and the epistemic problems begin to evaporate, these terminological distinctions will become vital to our understanding of the evolution of organismic form.
19
is because in light of recent findings concerning ‘deep homology’ (see below), even our best case
scenarios of convergent evolution may in fact be re-defined under the internally (and
contingently) constrained rubric of parallelism.
Whereas external constraints may be viewed as limiting the universe of viable solutions
to ecological design problems, internal developmental constraints may positively influence the
evolution of local optimality by establishing preferred internal channels that make good solutions
more likely to be hit upon (e.g. Diogo, 2005; Gould, 2002).21 As such, developmental
constraints on variation may limit the so-called ‘evolvability’ of lineages by reducing the overall
isotropic variation on which the optimizing agency of natural selection may act, but they may
also positively shape the evolution of organismic form by rendering accessible certain ‘good
tricks in design space,’ to use Dennett’s (1995) terminology.22
Such a positive reading of internal constraint implies that developmental processes may
bias generative mechanics so as to enhance the production of certain adaptive features,
particularly ‘key innovations’ or traits causally associated with macroevolutionary diversification
and the origins of higher taxa.23 For example, Hunter and Jernvall (1995) have shown that the
‘hypocone,’ the extra cusp characteristic of the quadritubercular (rather than triangular) molars
21 The canonical definition of ‘developmental constraint’ was introduced by Maynard Smith et al. (1985), who held it to be “a bias on the production of variant phenotypes or a limitation on phenotypic variability caused by the structure, character, composition, or dynamics of the developmental system” (p. 266). As Amundson (1994) notes, constraints on form are only constraints on adaptation when they prohibit the production of phenotypic variants that would have been favored by natural selection over variants permitted by the formal constraints. To this extent, the discourse on constraint is entrenched (forgive the pun) in a counter-factual biological analysis. 22 In fact, canalization may be more than an incidental limitation on isotropic variation; it could also be (as Waddington, 1959, suggests) an adaptation in its own right designed to ensure developmental outcomes by maintaining the robustness of ontogenetic trajectories in the face of developmental disturbances. 23 To the extent that key innovations or ‘good moves’ in design space are connected to the origins of higher taxa, one way of testing SCM’s hypothesis regarding the inevitability of certain adaptive complexes is to determine whether or not they are correlated with speciel or functional diversity (via neontological and paleontological comparison with sister taxa). Many promising candidates, such as the image-forming eye (see below), yield mixed results (see De Queiroz, 1999). This is largely because key innovations are contingent on numerous spatio-temporal and ecological factors (De Queiroz, 2002), and thus can only be recognized retrospectively (Sterelny, 2006).
20
of most therian mammals, evolved “convergently” (read: in parallel) within this order of
mammals on more than twenty occasions and is associated with marked diversification. It is
almost certain that this highly iterative and presumably useful adaptation, which greatly
increases occlusal area for crushing herbivorous material, owes its repeated evolution to a
common developmental pathway underlying the generation of the therian dental arcade. Thus,
while mammalian developmental constraints may limit the range of solutions available to
members of that class, it may also positively bias the production of key innovations such as the
hypocone. At the same time, however, it would be unreasonable to argue that the mammalian
hypocone, because it evolved numerous times independently in one order of mammals, is an
inevitable evolutionary outcome insensitive to developmental constraints—but as we shall see,
this is precisely the type of argument put forward by SCM.
Importantly, by underwriting the parallel evolution of similar forms, internal constraint
may allow for a degree of macroevolutionary predictability, at least within specified taxonomic
(Bauplanian) parameters. Viewed in this way, internal constraints are not merely the cause of
global adaptive imperfection, but also the basis for local optimality. Thus, while particular
conserved developmental pathways (e.g. Pax-6/eyeless) may represent inextrapolatable frozen
contingencies, such generative entrenchment (Wimsatt & Schank, 1988) may facilitate the
repeated parallel evolution of predictable formal solutions which are optimal given the
developmental parameters.24 Consequently, we can make at least as good a case for
developmental attractors which preserve homology as SCM can make for functional attractors
which break it.
24 The notion of developmental entrenchment may be considered analogous to (but perhaps less coarse-grained than) the concept of ‘contingent irreversibility’ with respect to the major transitions of life (Maynard Smith & Szathmary, 1995; but see McShea (2001) on reversals in major transitions).
21
Ironically, developmental constraint may be responsible for much of predictable
functionalism, while at its core embodying a form of radical contingency. This conclusion may
be unsatisfying to those with a penchant for neat dichotomies, as it provides yet another instance
of noisy biology defying our hopes for a mathematical symphony of parading forms. Yet, the
parameters of biological evolution remain not classical but improvisational—quirky, historical
movements, unintelligible in prospect but brilliant in retrospect, that would make Newton and
Mozart cringe but Armstrong and Davis proud.
As noted above, ‘convergence’ has often been touted as quintessential evidence for hard
adaptationism. However, many instances of homoplasy are better explained by a hybrid analysis
involving both functionalist (non-contingent) and historicist-structuralist (contingent)
interpretations of the constraints on available design space (Wake, 1991). Such an analysis is
more characteristic of parallelism than it is of convergence. SCM fails to recognize this
difference and therefore does not consider the possibility that frozen, canalized contingencies
may underlie many of the paradigmatic examples of ‘convergent’ evolution that he invokes.25
As a result, he repeatedly peddles putative examples of ‘convergence’ that are in fact parallel
adaptations underwritten by a homologous body plan, physiology, and genetic-developmental
architecture.
For instance, SCM appeals to numerous examples of ‘striking convergence’ within
taxonomic classes in response to common selective pressures. Such convergences better
interpreted as parallelisms (in which underlying homologous generators cannot be ruled out)
include: the ‘convergent’ saber-toothed morphology between placental and marsupial felids
(1998, p. 202-204; 2003, p. 130-132), ‘convergences’ within specific orders of mammals,
25 This crucial point is forcefully made by Sterelny (2005) in his critical review of SCM’s 2003 book.
22
including fossorial lifestyles (p. 132, 140), ‘convergence’ with respect to ‘pike morphology’ in
several genus of freshwater fish (p. 133), ‘convergence’ in raptorial forelimbs of the mantids and
neuropterans (p. 129) (two orders within the class hexapoda), ‘convergences’ within orders of
birds in plume coloring, wing shape, and hummingbird morphology (p. 138), and ‘convergence’
in stem morphology of the New World cactus and the African spurge, to name just a few (p.
134).26
As to this last example, SCM states (Ibid) that “cacti and spurges are only distantly
related, and their common convergence is because of the rigors of living in an arid environment.”
But this is only a partial truth, as a significant (if not dominant) causal determinant of the above
recognized similarities in form is a common Bauplan which restricts the set of formal solutions
to shared design problems. In such cases, both selection and development constrain adaptational
design space, and the latter may reflect contingent rather than robust macroevolutionary patterns.
Pictorial renditions of these parallelisms are on their face quite remarkable, and appear at first
blush to reflect the inexorable power of natural selection to steer form toward certain pre-
ordained evolutionary ‘endpoints’ or ‘functional islands.’ Nevertheless, they are not all that
surprising or impressive, given a more thorough appreciation of developmental pathway
conservation. As noted by Wake (1991, p. 555) in the context of the evolution of the attenuate
body form in salamanders, the fact that “related lineages independently have adapted to similar
microhabitats by assuming essentially identical ecomorphologies” is suggestive not of formal-
functional invariance but significant design limitations due to entrenched developmental
26 SCM also invokes the results of controlled experiments by Travisano et al. (1995) on populations of E. coli that he believes cut against a radically contingent view of life. After being left to diverge for 2000 generations and then exposed to a challenging biochemical substrate (maltose), adaptive responses were initially dominated by history (contingency), but (according to SCM) adaptationism ultimately prevailed via the swamping effects of convergence. This represents yet another example of parallel evolution involving underlying homologous generators whose extrapolatability to the history of life is seriously questionable. Furthermore, as Sterelny states (2005, p. 587), “[i]t is one thing to multiply evolve the ability to digest maltose; it is another to evolve meiosis.”
23
pathways. For these reasons, SCM cannot invoke many or even most of his ‘striking’ examples
of homoplasy without accounting for positive developmental constraints. SCM may be correct
that natural selection will (ceteris paribus) tend to find the optimal solution to a given design
problem, but it will be forced to do so within the strictures of a contingently pre-defined
Bauplan.
For example, selection seems to have arrived at an essentially optimal ratio of filament
length to strike speed in squid tentacles (Kier & Curtin, 2002), but such a compromise is made
within the larger, constrained and perhaps ultimately contingent framework of the molluskan, or
more specifically cephalopod, body plan. Similarly, biologists have made arguments for long-
term evolutionary progress by pointing to iterative examples of competitive exclusion or
incumbent replacement, such as the replacement of straight-necked turtles with those capable of
folding their necks underneath their shells (a parallel rather than convergent event which,
according to Rosenzweig & McCord, 1991, occurred on four to five separate occasions). While
these examples might indeed be indicative of phyletic competition driving patterns of
macroevolution, there is no reason to believe that such intra-class or intra-order homoplasy is
generalizable to higher taxonomic levels. Therefore, such touted examples of parallelism are not
going to give SCM the kind of extrapolatability that he needs to support his universal biology,
which requires a set of laws regarding the evolution of biological form that is stable across
higher taxonomic counterfactuals.27
27 For sure, the number of variants within a given Bauplan may be infinite, but it is the disparity or range of morphological variation that is of critical import for purposes of understanding the scope and severity of constraints on the evolution of form. Amundson (1994) offers an elegant analogy in this regard between Chomskian theories of language and Bauplanian constraints—he states (p. 570) that “just as all languages generated by a universal grammar are governed by certain constraints, so are all of the possible outcomes of the embryological processes of a given phylum.”
24
After discussing the remarkable case of ‘convergence’ between the marsupial fauna of
the super-islands of Australia and South America, and the placental mammals, SCM goes on to
render the following most astonishing conclusion (1998, p. 205): “Does it follow then that
contingent processes are an irrelevance in the way we see the world? I have argued that, so far
as the history of life is concerned, they are.” Simply by looking at putative homoplasies between
two mammalian orders, he concludes that contingency is not important in the history of life.
What about the evolution of body plans themselves in the late Ediacaran? What about the
seemingly arbitrary extinction of many successful phyla in the Cambrian? Why did it take
between 1 and 2 billion years for the first single-celled organisms to appear, and another billion
years or so for prokaryotic symbiosis to produce eukaryotic cells, a major transition which only
happened once?28 The answers to these questions do not necessarily invoke contingency, but
these are critical questions for which we do not currently have answers—questions that clearly
cannot be resolved definitively or even confidently by appeals to marsupial parallelisms.
That said, for the comparative physiologist, instances of marsupial-placental convergence
represent a “treasure of information distinguishing between crucial and incidental features of
mammals that have taken up different habits and habitats” (Vogel, 1996, p. 301). Any such
extrapolations, however, are taxonomically limited to the mammalian class. On the other hand,
if the most recent common ancestor of toothed and baleen cetaceans was terrestrial, their
similarities would tells us much more about the external constraints on aquatic mammalian
lifestyle then they currently do. Nevertheless, even such broader generalizations would be
taxonomically confined—for example, “[t]he fact that some members of every major arthropod
class [] have extremely long and thin legs quite likely implies some basic but as yet
28 In each of his major works discussing the philosophical implications of convergence, SCM’s fails to address or otherwise mention the term ‘major transitions’, a concept which, according to a review by Szathmary (2005, p. 851), represents a “great challenge” to SCM’s proffered view of life.
25
unappreciated functional property of the basic skeletomuscular scheme of arthropods” (Ibid), but
probably not of bilaterians generally.
While SCM does point as well to genuine instances of convergence, many of these
examples, such as the analogy between arthropod eusociality and the reproductive caste systems
of naked mole rats (2003, p. 141-142), are probably not counterfactually robust and only weakly
extrapolatable. Eusociality, for instance, is only a key innovation (a trait connected with
macroevolutionary diversification as compared to a paleontological sister taxa without the trait in
question) if at all in its arthropod instantiation(s), which in some sense militates against
recognizing it as a universal functional attractor. Moreover, as Vogel (1996) points out, the
dozen or so independent origins of sexually non-egalitarian eusociality in Hymenoptera reflect
parallel rather than convergent adaptations underwritten by a peculiar haplo-diploid genetic
system (as compared to a one single case of eusociality arising in egalitarian insects). In any
event, if eusociality is an inevitable evolutionary endpoint, why has it not arisen in other
vertebrate clades?
Problems of counterfactual instability apply to many of SCM’s proffered exemplars of
convergence, including those between beetles and spiders on ant morphology (in subverting the
‘immune system’ of the hive), which themselves may depend on layers of ecological and formal
contingency. On the other hand, SCM does draw upon some examples of genuine macroscopic
convergence that could be the subject of robust macroevolutionary generalizations, such as those
pertaining to particular sensory modalities (Ibid, Ch. 7);29 lamentably, he offers no principled
method for comparing the extrapolatability of different types of homoplasy.
29 Sterelny (2005) suggests that there may be significant limitations on sensory adaptations given the constrained set of reliable signals an organism can get about its environment, but such constraints are probably much weaker than SCM contends.
26
SCM must find some way, then, to rescue the philosophical import of many of his
favorite examples of convergence properly reinterpreted as developmentally-entrenched (and
positively constrained) instances of parallel evolution. One possibility is to offer the following
argument: In order to develop a homology-free perspective on macroevolution, particularly in
relation to the evolution of form, we must (ironically) accede to the developmental-structuralist
contention that conserved internal channels underlie many homoplastic events; however, this
need not entail capitulation to a historicist, contingency-infested view of life, so long as each
entrenched developmental component of any given homoplasy arose not by contingent but
predictable, functional processes. Lather rinse repeat, so the argument goes, we arrive at the
current inhomogenous distribution of morphospace occupation through the optimizing hand of
selection working synergistically with the ‘ratcheting’ effect of internal Bauplanian constraints.
There are good reasons for rejecting this argument, however. While there are instances in
which natural selection has been shown to act as an optimizing agent (see examples involving
squid strike speed and folding turtle necks above), there are innumerable occasions in the
evolution of form in which a simple engineering analysis demonstrates that selection acts more
often as a tinkerer; that is by tweaking existing sub-optimal developmental plans to meet
transient local adaptive challenges. In fact, it is the ubiquitous sub-optimality of organismic
design that is often showcased (by Darwin, no less) as among the best evidence for evolution by
natural selection, a blind mechanistic process that works with what is at hand and cannot plan
ahead. Indeed, as Vermeij states (1994, p. 233), “the fact that degrees of specialization [i.e.
optimality] among ecologically similar species vary widely in the world today strongly implies []
that design standards in nature are relative, not absolute.” Thus, “given that the correlation
between structure and function is often far from precise, we should expect organisms to embody
27
ad hoc and often rather clumsy solutions to functional demands, solutions that bear a deep stamp
of history and ancestry” (Ibid, citations omitted).30 In sum, there are few good reasons for
believing that selection has produced primarily optimal, counter-factually stable developmental
constraints, particularly given the stochastic and ephemeral nature of the selective environment.
Furthermore, SCM has failed to demonstrate that the specific ineluctable evolutionary
endpoints suggested by widespread homoplasy are associated with diversification (as measured
either by species number or functional disparity), nor has he established that lineages bearing
these ‘inevitable’ adaptive complexes enjoy longer evolutionary durations than those without the
traits. To the contrary, Van Valen (1973) famously showed that the probability of extinction for
any group is steady through time, irrespective of its phylogenetic age. These findings have been
interpreted as the result of a perpetually decaying selective environment, due in part to strategic
interactions between co-evolving species, in which global evolutionary progress is virtually
impossible. At the same time, the ability of selection to subvert contingent Bauplanian
constraints has not been shown, and in fact the opposite conclusion flows from the massive
conservation of kernels and the prevalence of sub-optimal design. Having failed to establish
these two propositions, SCM’s central argument—and by implication his view of life—cannot
stand.
Thus, we may properly conclude that SCM cannot appeal to parallelisms to bolster his
core contention (2003 p. 11) that ‘[while] the landscape of biological form...may in principle be
almost infinitely rich...in reality the number of roads through it may be much, much more
30 Other factors interfering with selection, such as gene flow between neighboring populations, have been shown to undermine the evolution of local optimality. For instance, genes selected for in contiguous but ecologically different environments can hybridize during inter-populational mating, impeding optimal adaptation to local ecological challenges (Brandon, 1990). And there is evidence to suggest that even low levels of gene flow can have significant homogenizing effects on gene frequencies within a population (e.g. Maruyama, 1972).
28
restricted.’ A better explanation of irregular morphospace occupation is reflected in Gould’s
(2002, p. 1178) remarks:
I do not doubt that many discontinuities in morphospace represent the colonization by optimal phenotypes of widely dispersed peaks in maximal biomechanical efficiency. But I am equally confident that more of nature’s evidently nonrandom, and oddly dispersed, clusters in morphospace, bearing such enormously different weights ranging from single ‘outliers’ to millions of species, primarily record the historical constraints imposed by workable solutions with adaptive origins—developmental designs that then congealed, enforcing reiteration and change within their internally directed channels forever after.
Importantly, some of SCM’s claims tiptoe around the evolution of organismic form per se
into the more Platonic realm of disembodied functional properties. Nonetheless, convergence
between phyla or other distant taxonomic groups on certain broad functional solutions, such as
‘mastication,’ will not admit of interesting predictions regarding the evolution of form absent
some specification of a particularized Bauplan that natural selection can tweak in furtherance of
the common ecological task. Beyond forecasting a greater hardness of the proximate organs of
mastication relative to the objects being masticated (a ratio with otherwise unspecified
parameters), in addition to the ‘muscles’ and ‘neural’ architecture necessary to correlate their
grinding movements, not much can be said about the evolution of formal solutions to mastication
disembodied from an explicit body plan. This is because the limitations and positive (non-
isotropic) channels imposed by internal homologous generators are a much more significant
determinant of the evolution form then the invariant physical laws. Although not totally
insignificant, the latter constraint alone will not restrict ecologically viable morphospace to a
manageably finite set of formal solutions. While exobiologists would likely be able to identify
organs of mastication regardless of the particularized form they assumed, there is no reason to
conclude that vertebrate teeth, arthropod mandibles, or mollusk beaks are inevitable solutions to
29
such a broad ecological design problem as ‘chewing.’31 Thus, we may be able to conclude a
priori that big fierce animals will be relatively rare in any ecosystem (Colinvaux, 1980), but this
generalization tells us virtually nothing about what they will look like.32
What in any event are these Platonic functional attractors that SCM refers to? He
believes that they are “biological properties” that represent the predictable evolutionary
endpoints or functional attractors toward which biological form gravitates. There are deep
problems with this view, however, and precisely how deep will depend on how we cash in the
term biological property. If we define it broadly to include such mechanisms as ‘predator
evasion,’ ‘metabolism,’ ‘thermo-regulation,’ or ‘propulsion,’ such properties may indeed be
universal but they would be of no use whatsoever in predicting the evolution of form if there is
an unmanageably large set of configurations that can realize the same function. The biological
property of ‘a behavior that increases one’s chances of reproduction’ would be about as
uninformative.33 The concepts of Batesian and Mullerian mimicry, for instance, in and of
31 Another mistake that SCM makes in citing a litany of putative examples of ‘convergence’ is that he simply assumes without argument that the morphological similarities he cites represent similar formal solutions to similar functional design problems. In fact, the strict morphological definition of homoplasy at work in phylogenetic reconstruction includes no such assumption. That is, similar non-homologous features are convergent or parallel regardless of their particular functions—“the convergent forms may function similarly, but similarity in physiological performance or ethological or ecological biological role is neither assumed not expected” (Simpson, 1961). Thus, if SCM endeavors to show that the examples of convergent (often actually parallel) evolution that he cites imply particular invariant constraints on adaptational design space, he must show that these instances of homoplasy involve common selective effects. 32 Although a good case can be made for significant non-contingent (chemico-physical) constraints on biological mechanisms (Sterelny & Griffiths, 1999), this paper is concerned with the evolution of form, to which mechanism does not necessarily speak. 33 SCM suggests that the optimizing agency of natural selection biases macroevolutionary processes toward certain predictable evolutionary ‘endpoints.’ It is true that even if throughout the course of evolution we do not see a repetitious return to the same old formal functional solutions, we do find an undeniable trend in certain meta-variables, such as the increase associated with hierarchical structure and complexity of design. However, pioneering work by Dan McShea (1994 / 2001), operationalizing complexity to determine whether an evolutionary tendency (bias) toward complexity is present across all hierarchical levels of all major taxa, has shown the evidence for a driven (rather than passive) trend to be inconclusive. As such, SCM must countenance the adequacy of stochastic, ‘random walk’ null models of macroevolutionary change to explain the emergence of (homoplastic) complexity without upward biases toward particular ‘functional attractors.’ In other words, while many instances of homoplasy
30
themselves offer nothing by way of the evolution of form because they are contingent on the
sensory capabilities of predators, rather than any factors intrinsic to mimicry (Vogel, 1996).
Viewed as such, SCM’s conception of biological properties is at best overly broad and at worst
tautologous and devoid of interesting conclusions. Moreover, it is particularly odd that SCM
would argue for the existence of disembodied functional attractors largely by appealing to the
literature on parallel and convergent evolution, as functional convergence without formal
convergence does not confound phylogenetic reconstruction and is generally not considered
homoplasy.34
In any event, SCM offers no reason to believe that the set of functional attractors, even if
it is significantly less than the infinite morphospace of the logically possible, is small enough that
it will allow for robust generalizations. Given the complexities of local, Earth-based ecology, it
is easy to envision a near-infinite set of functional attractors in the hyper-heterogeneous cosmic
ecology—one so large that science can never make a dent in it, let alone invoke local ecological
conditions to formulate laws of form on its basis. Even many of the adaptive complexes which
can be defined solely by their selective effects because they represent functional solutions to
widespread design problems in the animal kingdom, such as ‘gut’ or ‘heart,’ are probably not
restricted to a manageable set of forms that can realize the relevant function. For this reason, as
far as the evolution of organismic form is concerned, purely functional biological categories of
may be the result of similar selective pressures, overall macroevolutionary trend dynamics may be random with respect to those particular instances. Moreover, convergence in one-dimensional variables such as hierarchy would be no more impressive than that with respect to say body size or color, since none of these features enable biologists to make robust predictions regarding the evolution of particularized form. Thus, even if a selection-biased trend toward hierarchical complexity were established, this would be but a very small brick in foundation of a universal biology. 34 SCM also points to convergences in behaviors, such as ‘agriculture’ (e.g. between certain hymenoptera and modern humans). While behavioral homoplasy may be more amenable to operationalized comparison than disembodied functions, it nonetheless presents definitional difficulties that make the comparison more recalcitrant than traditional investigations in structural homoplasy.
31
the sort relied upon by Neander (1991) and referenced by SCM do not have much to contribute
to the quest for non-accidental macroevolutionary generalizations.
To be clear, I am not arguing that the holy grail of biology—a universal research
program—is but a pipe dream. In fact, I have put forward ideas elsewhere (Powell, 2005,
unpublished manuscript) designed to develop a conceptual-empirical research program toward
precisely that effect. In order to overcome the obstacles imposed by functional supervenience on
a heterogeneous formal base, and strategic interaction operating in a stochastically fluctuating
selective environment, we must identify more than simply nominal instances of homoplasy that
are merely suggestive of design constraints, as SCM has done. Rather, we must identify
iteratively evolved traits that are robust and Bauplan-independent, in that they represent the
recurrent independent realization of adaptive solutions that, due to unusually severe physical
constraints and near-universal ecological applicability, are limited to a manageable set of forms
and largely impervious to strategic interaction. If we could confidently identify such an adaptive
complex,35 it would offer a powerful refutation of radical contingency, or the view that not only
is the emergence of specific taxa highly contingent on initial conditions, but so too are the
complex adaptive solutions hit upon by those taxa (as espoused by Gould, 1989, p. 290).
Unfortunately, SCM’s unsystematic work on convergence takes us no closer to realizing
this goal. He maintains that unlike the specific adaptations of a particular species, general
35 Candidates that are in my view probably robust across higher taxonomic counterfactuals and constrained to a limited set of potential forms include (inter alia) image-forming eyes (see e.g. Land, 1992), circulatory systems which depend on bulk flow rather than simple diffusion and hence require nonlinear elasticity (Vogel, 1996), and load-bearing structures which depend on a crucial twist-to-bend ratio (Vogel, 1998). The latter two examples involve mechanical variables that can be (and indeed have been) achieved in forms as diverse as the hind-tibia of Orthoptera to tree trunks and flight-associated feathers.
32
biological properties are robust, since they will inevitably manifest at some spatio-temporal
point in the unfolding of the evolutionary process.36 He states (2003, p. 308):
[A]lthough any history is necessarily unique, the resultant complex end-form is not simply the contingent upshot of local and effectively random processes. On any other suitable planet there will I suggest be animals very much like mammals, and mammals much like apes. Not identical, but surprisingly similar.
In essence, then, SCM’s challenge is not directed at uncontroversial theories of contingency
associated with “the destiny of a given lineage” (1998, p. 201), but toward the more radical
Gouldian notion that most or all interesting biological properties are themselves highly
sensitively dependent on initial conditions. He states (2003 p. 307) that “whenever the known
edge of the evolutionary envelope is reached...then it will be explored independently several
times.” However, he does not offer any concrete guidelines as to how in any particular case we
are to know when the putative “edge” is reached. How many cases of convergence are to
suffice, and how are we to determine the degree of similarity necessary to recognize them as
meaningful homoplastic events? In this vein, how similar must the structural solutions be
(formally and functionally) for our projections to be meaningful? Does parallel evolution count
for or against the inevitability of evolutionary endpoints? Even more importantly, why are
certain endpoints at the end? These questions confront any strong appeal to the frequency of
homoplasy as evidence for a highly non-contingently constrained adaptational design space.
Finally, SCM appeals to widespread exaptation as evidence of the supremacy of natural
selection to escape developmental constraints in driving the compass of life toward particular
36 SCM would argue that although we might be unable to predict which particular lineages (e.g. small theropod dinosaurs with feathers aiding in thermoregulation) will co-opt existing structures for particular predictable functions (e.g. powered flight), we need only show that it is virtually inevitable that some lineage will evolve the relevant biological property.
33
formal-functional solutions.37 His view seems to be that ubiquitous pre-adaptation implies that
life will find a way to ‘leap frog’ as it were from one developmental constraint to another,
shifting functions and co-opting spandrels, inexorably moving toward more complex functional
solutions embodied by homoplastic events. As such, he views exaptation as a positive causal
contributor to the frequency of homoplasy. Gould (2002) also emphasizes the
macroevolutionary role of exaptation, but he argues that it points instead toward a severely
contingent view of life. He maintains (p. 1160) that the ubiquity of “quirky functional
shift...inevitably suggests (but admittedly does not prove) a high degree of fortuity, as implied by
the required capacity of features built for one function to act in another way that could not have
influenced or regulated their original construction by any functional evolutionary mechanisms
like natural selection.” Thus, one can just as easily read the ubiquity of exaptation as frustrating
rather than facilitating homoplastic events by forcing natural selection to work within severely
sub-optimal developmental constraints. As a result, we are unable to analytically or empirically
infer either predictability or contingency from pervasive exaptation.
3. DEEP HOMOLOGY AND ITS IMPLICATIONS FOR MACROEVOLUTIONARY EXO-BIOLOGICAL EXTRAPOLATION
Deep homology refers to the recent surprising discovery in the emerging field of
evolutionary developmental biology (‘evo-devo’) that all major phyla share certain key
developmental pathways whose origin dates back to the Pre-Cambrian, well over 500 million
37 Throughout his 2003 book (e.g, p. 4-5), SCM vaguely refers to a concept he calls “inherency,” which interpreted mechanistically appears to refer to the notion of an ‘exaptive pool’ (sensu Gould, 2002, p. 1270) and is often invoked in the context of discussions of phenomena that are best described as exaptation (a term conspicuously absent from an otherwise thorough index). This reading is supported by a definition SCM offers in another publication that same year (2003b, p. 505), in which he characterizes inherency as “the potentiality of structures central to evolutionary advancement…[which] are already ‘embedded’ in more primitive organisms.” Szathmary (2005, p. 856) has recently also interpreted SCM’s use of ‘inherency’ in this way.
34
years ago, a notion unexpected by the Modern Synthesis. The presence of deep homology
suggests a much more prominent role for parallel evolution in shaping the history of life than
previously thought. This raises the possibility that bilaterian diversity, as Gould (2002, p. 1159)
suggests,
represents an extensive set of modifications and tinkerings upon a basic pattern set by history at the outset, and then adumbrated in one geologically brief episode to establish all fundamental building plans. Forever after, for more than half a billion years, the subsequent evolution of complex animals [] has been restricted to much more limited permutation within the confines of these early congealed designs.
The core question, then, is whether the metazoan Bauplane represent a “subset of workable, but
basically fortuitous, survivals among a much larger set that could have functioned just as well,
but either never arose, or lost their opportunities, by historical happenstance” (Gould, 2002, p.
1160-1161). This question is currently unanswerable, although the foregoing analysis
concerning the decaying nature of the selective environment combined with the radical
entrenchment of kernels suggests that the various morphospatial clusterings are unlikely to
Research probing the phylogenetic depth of homologous development, such as work done
on the inter-phylum substitutability of master control genes such as Pax-6/eyeless (Zuker, 1994)
and Hox genes (Carroll, 2005), and their ability to generate ectopic or supernumerary organs in
different phyla, demonstrates the substantial macroevolutionary importance of phylogenetic
conservation. Take, for instance, that locus classicus of convergent evolution in the camera-type
eyes of vertebrates and cephalopods. While the macroscopic arrangements of such eyes may be
convergent, their underlying neural-developmental architecture is homologous, dating back to
35
the Pre-Cambrian. SCM notes as much (2003, p. 239-241), but does not follow the implications
of deep homology to its natural, larger conclusion. As Gould (2002, p. 1069) states,
Eyes of such strikingly similar design owe their independent origin as much to genetic and developmental parallelism, based on internal constraints of homologous genes and developmental pathways, as to selection’s capacity for iterating nearly identical adaptations from scratch by convergence.
SCM notes that there may be a common organization for highly developed sensory systems
which he believes, if true, suggests that the evolution of the various sensory modalities may be
surprisingly likely once the neural architecture is in place for one system. However, he offers no
guess as to the evolutionary probability of that initial suite of neural architecture which he
believes was placed into the exaptive pool for later cooption into various sensory modalities as
selection sees fit. Presumably, he would argue that the massive conservation of entrenched
developmental pathways speaks not to the existence of frozen accidents but rather to the power
of natural selection to stabilize good adaptive solutions.
Nevertheless, deep homology raises the following striking possibility: at the heart of many or
even most of the quintessential examples of convergence is a massively conserved, monophyletic
developmental architecture. If this is the case, and the evolution of such underlying homologous
constraint (what Saether (1989, p. 619) referred to as the “capacity to develop synapomorphy”) is
itself sensitively dependent on initial conditions, then adaptationists such as SCM cannot freely
cite such examples of parallel evolution as evidence of non-contingent constraints on
adaptational design space. Deep homology may indeed imply deep contingency—and it only
takes one major contingency to collapse the regime of adaptive predictability!
Nevertheless, non-accidental generalizations regarding the evolution of form can obtain even
if the entire history of life is derived from a single massive contingency (say at the earliest stages
36
of replicating molecules, 3.5 bya), or even from several phylum-level contingencies during the
Cambrian ‘experimentation phase’ which resulted in the entrenchment of only a small subset of
equally likely metazoan phyla. This is because our extrapolations could specify particular
Bauplanian constraints (which function as a limited set of ceteris paribus qualifications), thereby
presupposing certain major transitions in the evolution of life or the origins of major body plans.
For example, given the developmental underpinnings of the mammalian dental arcade, and an
epistemologically manageable set of ecological factors, we could perhaps make certain robust
predictions regarding the evolution of mammalian molars. This may be done even if one
monumental contingency (e.g. a massive bolide impact 65 mya) lies at the casual heart of
mammalian adaptive radiation in the Paleocene.
The precision of our generalizations will, of course, be inversely correlated with the
taxonomic abstraction at which they are made—that is, intra-order extrapolations will fare better
than those made within phylum, class, etc., due to the increased significance of shared
homology. Indeed, it is for this reason that we should expect to find higher frequencies of
homoplasy at lower taxonomic categories, as we in fact do. If this analysis is correct, I would
regard it not as a failure of a priori biology per se, but rather as an important (if ineliminatable)
limitation on the scope of exobiological generalizations.
4. THE NEED TO SPECIFY HIERARCHICAL LEVELS OF ANALYSIS IN ASSESSING HOMOLOGY AND HOMOPLASY
Homology has always been a non-functional category, identified by anatomical,
morphological, or developmental criteria alone, irrespective of selective-functional
considerations (Amundson & Lauder, 1994). Importantly, organisms possess character traits at
37
multiple levels of organization and can be examined at different structural tiers (Roth, 2001). As
developmental biology has matured, it has become increasingly clear that “homology at one level
does not necessitate homology at another” (Abouheif et al., 1997). According to Van Valen
(1982), one fundamental property of homology is “continuity of information.” Because different
aspects of structures are controlled by different developmental programs (Roth, 1984), changes
at one level, such as genotype or gross phenotype, may be ‘screened off’ (sensu Brandon, 1990)
from changes at other levels of the hierarchy (Roth, 2001 / 1991). Phenotypic traits may be
decoupled, as it were, from their underlying patterns of gene expression and development.38 For
these reasons, we are compelled to identify our level of inference in assessing homology, be it
genes, their patterns of expression, their developmental functions, or the structures to which they
ultimately give rise.
A similar analysis should apply to our identification of homoplasy. Not surprisingly, a
number of authors (Gould, 2002; Bolker & Raff, 1996; Roth, 1991; Wagner, 1989) have
emphasized that the assessment of homoplasy can only be done effectively by specifying a
hierarchical level of analysis. For example, many aspects of the wings of birds and bats are
homologous—for they share a common suite of vertebrate plesiomorphic traits—however, the
macroscopic modification of those traits (for powered flight) evolved homoplastically between
these lineages. Similarly, a given morphology may be parallel, while the underlying molecular-
developmental generators are homologous. Thus, in order to claim that two traits are
homoplastic, we must specify particular hierarchical levels of complexity, such as the molecular-
genetic, cellular, tissue, organ, morphological suite, or ecological archetype. This is a crucial
38 Roth (2001) offers the example of a gut, which is a synapomorphy shared by all vertebrates (and hence homologous), but with patterns of gastrulation which differ between vertebrate clades.
38
element missing from many discussions of convergence, including that of SCM. To be fair,
however, the champions of contingency are frequently guilty of the same confusion by also
failing to consider traits as multi-dimensional, heterogeneous entities. In sum, any similarity, be
it homologous or homoplasious, can be said to disappear at different levels of detail. Thus, as
Gould (2002, p. 1128) notes, “convergence may prevail on one level, and constraint at another.”
Furthermore, in order to identify true homology, we need only show “overall historical
continuity in developmental mechanics,” not uniformity of underlying development; that is to
say, we must only demonstrate that in “stepwise substitution of some elements of the process [],
correspondence remains for other elements” (Roth, 2001, p. 94).39 Presumably, since ‘true
homology’ requires contiguity in developmental mechanics, ‘true homoplasy’ requires a
discontinuity in developmental mechanisms, which is supposedly reflected in phyletic disparity.
However, as we have seen, parallel evolution involves both a continuity and discontinuity in
underlying developmental generators, which compels us to recognize this gray area between
homology and homoplasy—a category that (as findings in deep homology suggest) is becoming
ever more important for understanding the inhomogenous occupation of morphospace.
Even if we find substantial convergence at higher vertical levels of analysis, such as the
macroscopic arrangement of the camera-type eye, the generalizability of such homoplasy may be
undercut by the presence of an earlier horizontal contingency, such as the monophyletic origin of
the underlying neural-developmental architecture or, even more broadly, the advent of
prokaryotic symbiosis to form eukaryotic cells (the single pre-cursor to multi-cellularity, as
noted above).
39 Thus, according to Roth (2001), X is homologous in all instances of the following four-step evolutionary transition sequence: (1) a,b,c produce trait X; (2) a,b,k produce X; (3) p,b,k produce X; (4) p,g,k produce X.
39
5. DIFFICULTIES IN OPERATIONALIZING THE DEGREE OF HOMOPLASY
As intimated above, homology and homoplasy are not all-or-nothing phenomena (Roth,
1983), and the degree of homology may depend on the extent to which two traits share a
developmental pathway. In the weakest forms of homology, for example, two traits may merely
derive from the same germ layer. Just as it is important to specify the hierarchical levels under
consideration, it is necessary to assess the degree of homoplasy when appealing to examples of
convergence as evidence for limitations on the evolution of organismic form. In listing
numerous examples of homoplasy, SCM makes no attempt to assess their respective strengths,
and offers no method for quantifying the significance of the examples he cites.
Given the inherent difficulties of assessing homoplasy in multi-dimensional structures, it
is not surprising that no method to my knowledge has yet been conceived by which to quantify
the degree of homoplasy between two lineages or forms. In theory it is possible to operationalize
the degree to which a given set of organs have become more or less similar over time—this
could be accomplished by methods analogous to the determination of morphospace occupation
(but in reverse, since disparity measures morphological divergence, rather than convergence).
Even under such an analysis, however, instances of meaningful homoplasy will often be limited
to examples of parallel rather than truly convergent evolution, since (as noted above) it is only
within lower-taxonomic contexts that form will significantly converge on multiple hierarchical
levels.
For example, the parallel evolution of saber-toothed morphology between orders of
mammals will garner a higher homoplasy rating than the convergence between mollusk and
vertebrate camera-type eyes with respect to the macroscopic arrangement of structural parts, due
in large part to the higher propinquity of descent in the former. In order to compensate for this
40
problem, we might attempt to calibrate for phyletic distance (say, for instance, between the
incisors of mammals and the mandibles of hexapods) by adjusting for the importance of
between-phyla convergence. This might allow us to better control for homology and assess the
significance of external, functional constraints on the evolution of form. While it is certainly
worth a try, it is presently unclear how such calibration can be carried out in a principled way.
Another potential method for operationalizing the degree of homoplasy is to construct a
null model of morphological change over time, assessing the probability that by chance alone
two lineages ‘randomly walked’ (according to purely stochastic processes) toward one another in
morphological variation.40 Although the null model of morphological change does not posit an
underlying cause of the homoplasy, where two lineages resemble each other more than either
does their common ancestor, the most parsimonious presumption is that such commonalities
result from similar selective pressures (e.g. Vogel, 1996). This presumption is rebuttable,
however, and must be supported by an ecological engineer’s analysis.
Apart from issues regarding its causal agnosticism, however, another limitation of the
null model of morphological change is that it does not take into account the various hierarchical
components of parallelism—that is, the contribution of underlying monophyletic developmental
pathways to homoplastic trends. As discussed above, developmental canalization may act as a
positive constraint on the evolution of form and a facilitator of homoplastic events by
functioning as a bias in an otherwise stochastic trend. Therefore, assuming that selection is
equally strong in each case, trends involving parallel and convergent evolution respectively may
be differentially biased by their corresponding degrees of developmental entrenchment.
40 For this point I am indebted to Dan McShea, personal communication. While random walks will frequently admit of statistically significant regressions and obvious trends, they often generate fallacious intuitions about bias, since such trends can emerge even where ‘ups’ and ‘downs’ are equally probable.
41
CONCLUDING REMARKS
A contingent view of life does not entail denying a significant role to natural selection in
governing evolutionary change. However, it does claim that selection is forced to respond to
stochastically changing ecological environments and to work within entrenched, sub-optimal
developmental constraints. Rather than snubbing the only known mechanism for producing
function, the historicist may grant some importance to the “striking phenomenon” of
convergence, but maintain that SCM “overestimates its sway” in guiding evolution to particular
endpoints (Gould, 1998, p. 57). In searching for the holy grail of biology, we must be cautious
to avoid what we might call the ‘anti-naturalistic fallacy,’ or the penchant to believe that nature
(in this case macroevolution) is how we would like it to be—orderly, intelligible, and governed
by a manageable set laws. I leave it to the reader to decide whether Charles Darwin himself was
a functionalist or historicist at heart—that is, whether the oft-quoted final passage of the Origin
(reproduced below) makes an analogy or disanalogy between the invariance of the physical laws
and the evolution of biological form:
There is grandeur in this view of life, with its several powers, having been originally breathed into a few forms or into one; and that, whilst this planet has gone cycling on according to the fixed law of gravity, from so simple a beginning endless forms most beautiful and most wonderful have been, and are being evolved (1859).
42
BIBLIOGRAPHY Abouheif, E. et al. (1997). Homology and developmental genes. TIG. 13: 11. Alberch, P. (1982). Developmental constraints in evolutionary processes. In Evolution and Development (J.T. Bonner, ed.). Springer-Verlag, Berlin. Beatty, J. (1995). The evolutionary contingency thesis. In Concepts, Theories, and Rationality in the Biological Sciences (G. Wolters and J. Lennox eds.). Brandon, R.N. (2005). The principle of drift: biology’s first law. (In press) ____________(1990). Adaptation and environment. Princeton Univ. Press, Princeton Bolker, J.A., & Raff, R.A. (1996). Developmental genetics and traditional homology. Bioessays 18: 489-494. Carroll, S.B. (2005). Endless forms most beautiful: the new science of evo devo. W.W. Norton & Co., Inc., New York. Conway Morris, S. (2003). Life’s solution: Inevitable humans in a lonely universe. Cambridge Univ. Press, Cambridge _______________(2003b). The Cambrian ‘explosion’ of metazoans and molecular biology: would Darwin be satisfied? Int. J. Dev. Biol. 47: 505-515. ________________(1998). The crucible of creation: The Burgess Shale and the rise of animals. Oxford Univ. Press, Oxford _______________(1994). Why molecular biology needs paleontology. Development. O(Supp.): 1-13. Conway Morris, S., and Gould, S.J. (1998). Showdown on the Burgess Shale. Nat Hist. 107: 48-55. Corning, P.A. (1983). The synergism hypothesis: a theory of progressive evolution. McGraw Hill, New York. Danforth, B.N. (2002). Evolution of sociality in a primitively eusocial lineage of bees. Proc. Nat. Acad. Sci. 99: 286-290. Darwin, C. (1859). On the origin of species by means of natural selection. 2nd ed., Murray, London. (reprint, 1963, Heritage Press, New York).
43
Davison, E.H. and Erwin, D.H. (2006). Gene regulatory networks and the evolution of animal body plans. Science. 311: 796-800. Dennett, D.C. (1995). Darwin’s dangerous idea. Simon & Schuster, New York. Diogo, R. (2005). Evolutionary convergences and parallelisms: their theoretical differences and the difficulty of discriminating them in a practical context. Biology and Philosophy. 20: 735-744. Dobzhansky, T. (1951). Genetics and the origin of species. Columbia University Press, New York. Duntley, J.D. & Buss, D.M. (2004). The evolution of evil. In The Social Psychology of Good and Evil (A.G. Miller, ed.). Guilford Press, New York. Endler, J.A. (1986). Natural selection in the wild. Princeton Univ. Press, Princeton, N.J. Foley. R. (1999). Pattern and process in hominid evolution. In Structure and Contingency: Evolutionary Processes in Life and Human Society (J. Bintliff, ed.) Leicester Univ. Press, London Goldsmith, T.H. (1990). Optimization, constraint, and history in the evolution of eyes. Q. Rev. Biol. 65: 281-322. Gould, S.J. (2002). The structure of evolutionary theory. Harvard University Press, Cambridge _________(1996). Full House: the spread of excellence from Plato to Darwin. Three Rivers Press, New York. _________ (1989). Wonderful Life: The Burgess Shale and the nature of history. W.W. Norton and Co., New York. Gould, S.J. & Calloway, C.B. (1980). Clams and brachiopods--ships that pass in the night. Paleobiology. 6(4): 383-396. Haas, O., & Simpson, G.G. (1946). Analysis of some phylogenetic terms with attempts at redefinition. Proc. Amer. Phil. Soc. 90: 319-349. Hennig, W. (1966). Phylogenetic systematics. Univ. Illinois Press, Urbana, Ill. Hunter, J., Jernvall, J.P. (1995). The hypocone as a key innovation in mammalian evolution. Proc. Natl. Acad. Sci. U.S.A. 92: 10718-10722.
44
Kauffman, S. (1995). At home in the universe. Oxford University Press, Oxford Kier, William M, & Curtin, Nancy A. (2002). Fast muscle in squid (Loligo pealei): contractile properties of a specialized muscle fibre type. Journal of Experimental Biology 205: 1907-1916. Land, M.F. and Fernald, R.D. (1992). The evolution of eyes. Annu. Rev. Neurosci. 15: 1-29. Lankester, E.R. (1870). On the use of the term homology in modern zoology, and the distinction between homogenetic and homoplastic agreements. Ann. Mag. Nat. His. 6: 34-43. Maruyama, T. (1972). Distribution of gene frequencies in a geographically structured finite population. Ann. Human Genetics. 35: 411-423. Maynard Smith, J. (1999). The origins of life. Oxford Univ. Press, Oxford, U.K. Maynard Smith, J., and Szathmary, E. (1995). The major transitions in evolution. W.H. Freeman, New York. Maynard Smith, J., Burian, R., Kauffman, S., Alberch, P., Campbell, J., Goodwin, B., Lande, R., Raup, D., & Wolpert, L. (1985). Developmental constraints and evolution. Quart. Rev. Biol. 60: 265-287. Mayr, E. (1963). Animal species and evolution. Harvard Univ. Press, Cambridge, MA. McShea, D.W. (2005). The evolution of complexity without natural selection (or, a possible large-scale trend of the fourth kind). Paleobiology, 31(Supp.): 146-156. _____________(2001). The “minor transitions” in hierarchical evolution and the question of directional bias. J. Evol. Biol. 14: 502-518. _____________(1994). Mechanisms of large-scale evolutionary trends. Evolution. 48: 1747-1763. Neander, K. (1991). Functions as selected effects: the conceptual analyst’s defense. Philosophy of Science. 58: 168-184. Patterson, C. (1988). Homology in classical and molecular biology. Molec. Biol. Evol. 5: 603-625.
45
Reichenbach, H. (1956). The direction of time. Univ. of California Press, Berkeley, CA. Richerson, P.J. & Boyd, R. (1999). The evolutionary dynamics of a crude super organism. Human Nature. 10: 253-289. Rosenberg, Alex (2001). How is biological explanation possible? Brit. J. Phil. Science. 52(4): 735-760. Rosenzweig, Michael L., & McCord, Robert D. (1991). Incumbent replacement: evidence for long-term evolutionary progress. Paleobiology. 17(3): 202-213. Roth, V.L. (2001). Character replication. In The Character Concept in Evolutionary Biology (G.P. Wagner, ed.). Academic Press. _________(1991). Homology and hierarchies: problems solved and unresolved. J. Evol. Biol. 4: 167-194. _________(1984). On homology. Biological J. Linnean Soc. 22: 13-29. Saether, O.A. (1983). The canalized evolutionary potential: inconsistencies in phylogenetic reasoning. Systematic Zool. 32: 343-349. Salmon, W. (1984). Scientific explanation and the causal structure of the world. Princeton Univ. Press, Princeton, N.J. Salthe, S. (1994). Evolution and development. M.I.T. Press, Cambridge, MA. Simpson, G.G. (1961). Principles of animal taxonomy. Columbia Univ. Press, New York. Sterelny, K. (2006). Macroevolution, minimalism, and the radiation of the animals. In Cambridge Companion to the Philosophy of Biology (D. Hull and M. Ruse, eds.). __________(2005). Another view of life. Stud. Hist. Phil. Biol. & Biomed Sci. 36: 585-593. Sterelny, K. & Griffiths, P.E. (1999). Sex and death. Univ. Chicago Press, Chicago, Ill. Sober, E. (1988). Reconstructing the past: parsimony, evolution and Inference. M.I.T. Press, Cambridge Szathmary, E. (2005). Life’s solution. Biology and Philosophy. 20: 849-857.
46
Travisano, M., Mongold, J.A., Bennett A.F., & Lenski, R.E. Experimental tests of the roles of adaptation, chance, and history in evolution. Science. 267: 87-90. Van Valen, L. (1973). A new evolutionary law. Evol. Theory. 1: 1-30. ___________ (1982). Homology and causes. J. Morphology. 173: 305-312. Vermeij, G.J. (1994). Evolutionary interaction among species: selection, escalation, and coevolution. Ann. Rev. Ecol. Systematics. 25: 219-236. Vogel, S. (1998). Convergence as an analytical tool in evaluating design. In The Optimization and Symmorphosis Debate (E.R. Weibel, C.R. Taylor, and L. Bolis, eds.). Cambridge Univ. Press. ________ (1996). Diversity and convergence in the study of organismal function. Israel J. Zool. 42: 297-305. Waddington, C.H. (1959). Canalisation of development and the inheritance of acquired characters. Nature. 183: 1654-55. Wake, D.B. (1991). Homoplasy: the result of natural selection or evidence of design limintation. Amer. Naturalist 138: 543-567. Wagner, G.P. (1989). The biological homology concept. Ann. Rev. Ecol. Systematics 20: 51-69. Willmer, P. (2003). Convergences and homoplasy in the evolution of organismal form. In Origination, developmental and evolutionary biology (Muller, G.B. and Newman, S.A. eds., pp. 33-49). Bradford Book, Cambridge. Wilson, D.S. (2002). Darwin’s Cathedral. Univ. Chicago Press, Chicago. Wimsatt, W.C. and Schank, J.C. (1988). Two constraints on the evolution of complex adaptations and the means of their avoidance. In Evolutionary Progress (M.H. Nitecki, ed., pp 231-75). Univ. Chicago Press.