This is the second of a three-part essay in which I defend my 1985 book, Evolution: A Theory in Crisis (Evolution), in light of the scientific advances and discoveries of the past thirty years.1 In Evolution, I argued that the major, taxa-defining innovations in the history of life were not derived from ancestral forms by functional intermediates.
In the first part of this essay, I considered in detail the origin of the enucleate red cell. In this part, I discuss the tetrapod limb, the feather, and flowering plants.
The Tetrapod Limb
The tetrapod limb, possessed by all members of the four classes of terrestrial vertebrates (amphibians, reptiles, birds, and mammals), is to some extent a second-order novelty, as it is derived from an antecedent structure, the fin of a lobe-finned fish. However, for our purposes, it is an important novelty because accounting for its nature and origin poses massive problems. This has long been clear. Richard Owen noted these difficulties 166 years ago, in On the Nature of Limbs.2
The mature, adult, functional forms of tetrapod fore and hind limbs differ in extraordinary ways, yet all are based on an unchanging pentadactyl design or Bauplan. This pattern is obvious in the human arm and leg. Although their morphologies are very different, both conform to the same pattern. In the arm, there is one bone (the humerus) between the shoulder and the elbow, two between the elbow and the wrist (the radius and ulna), and five fingers. In the leg, there is one bone (the femur) between the hip and the knee, two bones between the knee and the ankle (the tibia and fibula), and five toes. The same pattern is easily observed in the wings of bats and pterosaurs. Similarly, the flippers of seals and whales, the wings of birds, and the limbs of horses are built on the same underlying pattern, even though this is not obvious on superficial examination.
Likewise, the digits of the fore and hind limbs differ from one another in every known terrestrial vertebrate, and in most more than in man. Yet all are based on a common design: a succession of two, three, or four small bones of decreasing size, from proximal to distal, called phalanges.3 This is particularly obvious in the human body. Although all ten digits have a different form, the underlying digital Bauplan is essentially identical.
Owen believed that the Bauplan of the tetrapod limb must have been generated by factors other than selection for environmental fitness. Charles Darwin himself conceded in On the Origin of Species that no adaptive explanation for the underlying Bauplan could be given, for it appeared to serve no specific adaptive end.4
In Evolution, I wrote, “It is generally presumed that amphibia evolved from fish and even the order of fish, the Rhipidistia, has been specified. However, transitional forms are lacking.”5 More precisely, transitional forms leading in small successive steps from fin to limb are lacking. The earliest known amphibians, as I pointed out, had well-developed fore and hind limbs of the normal tetrapod type.
Thirty years have passed. We have still not found a single fossil with an appendage that might have bridged the gap between a fish fin and the tetrapod limb. Certainly, since 1985, a great number of early amphibian fossils have been discovered, including Acanthostega, Tulerpeton, and Ventastega, as well as several fossil fish close to the fish-amphibian boundary, including the celebrated Tiktaalik. We have also made huge advances in understanding the developmental genetics of the limb. But we are no closer to giving an account of the fin-to-limb transition in Darwinian terms.
In Gaining Ground, paleontologist Jennifer Clack considers in depth the many early amphibian species and their putative fish ancestors found since Evolution was published, but nowhere does she allude to any clear intermediate between a lateral appendage ending in fin rays and the tetrapod limb ending in digits.6 In a 2006 paper in Nature comparing the limb of the earliest amphibians with the fin of Tiktaalik (much touted as the most convincing transitional form yet discovered between fish and amphibians), Clack and Per Ahlberg conceded:
Although these small distal bones bear some similarity to tetrapod limbs in terms of function and range of movement, they are still very much components of a fin. There remains a large morphological gap between them [the distal bones of the fin] and digits as seen in, for example, Acanthostega: if the digits evolved from these distal bones, the process must have involved considerable developmental repatterning.7
The recent discovery in Poland of tracks made by tetrapods with developed digits further complicates the issue.8
Recently, Kathryn Kavanagh et al. showed that the digits of Acanthostega and other early tetrapods were not proto- or pseudo-digits, but true digits, homologues of all subsequent tetrapod digits. A special digit developmental module that regulates relative phalange size, they demonstrated, determined the relative size of the phalanges in digits among all terrestrial vertebrates, including the very earliest amphibians.9 They also showed that the phalangeal module is distinct from another developmental module that determines the size of the long bones of the hands, known as the metapodials:
We have focused on developmental processes determining proportions of phalanx size along individual digits (fingers/toes) of vertebrates. We find that phalangeal variation seen in nature is indeed constrained by an ancestral developmental program, limiting morphologies to a continuum from nearly equal-sized phalanges to a large-to-small gradient of relative sizes. … [S]uccessive phalanges within a digit exhibit predictable relative proportions, whether those phalanges are nearly equal in size or exhibit a more striking gradient in size from large to small.10
“[E]ach phalanx negatively influences the size of the next,” they conjectured, “to an extent characteristic of each digit.” What is more, “the metatarsals do not appear to influence the segmentation process forming the phalanges and accordingly appear to be a distinct developmental module.”11 In effect, two different developmental modules are involved in generating the digits and metapodials. Kavanagh et al. surmised that these developmental modules may be homologous in all vertebrate limbs:
Measurements of phalanges across species from six major taxonomic lineages showed that the same predictable range of variants is conserved across vast taxonomic diversity and evolutionary time, starting with the very origins of tetrapods [emphasis added].12
The evidence, then, suggests that the digits—phalanges and metapodials—have been determined from the beginning by two distinct, novel developmental modules. Moreover, the authors of another recent paper showed that a Turing reaction-diffusion mechanism determined digit number; the segmentation of digits into phalanges could be altered, or even abolished, by changes to the gene circuits involved in establishing the system.13
The fact that the developmental module governing phalange segmentation and digit number is used by extant tetrapod embryos suggests that this was also the case in the very first tetrapods.
And this suggests that the digits we see in Acanthostega and the earliest tetrapods are homologous to those in all subsequent vertebrate hands and feet; digit numbers are generated by the same basic Turing reaction-diffusion mechanism that has remained unchanged for 400 million years. There is no evidence from fossils or evolutionary developmental biology of a progression via proto-digits to true digits. Previous descriptions of transitional structures, such as those offered by Erik Jarvik, have not been confirmed.14 We do not know whether this pattern originated first in one limb and was then redeployed to the other, but it is generally assumed that redeployment played a crucial role.15
In every known tetrapod, the ten digits differ in morphology; this implies that each digit has a unique identity, like the hind and fore limbs themselves. While all digits are constrained by a homologous developmental module, a differential identity has been superimposed upon them.16 In extant limbs, digit identity is established in the tissues between each digit during their formation. A Sonic hedgehog (Shh) protein gradient is involved in determining digit identity and number.17 Given that digit differences are so marked in the very earliest autopods, the evidence is consistent with the idea that the Shh gradient was present from the beginning. This implies that the earliest autopods differed from extant autopods only by the absence of pentadactyly. All other developmental systems were already in place.
The final step toward the modern autopod was the adoption of pentadactyly. The very earliest autopods possessed six, seven, and eight digits. But pentadactyly is the norm for post-Devonian tetrapods. The work of Rushikesh Sheth et al. suggests that a Turing reaction-diffusion mechanism was subsequently fine-tuned to limit the number of digits to five.18
The fossil evidence, and evidence from evolutionary developmental theory (evo-devo), suggest that the basic design of the autopod may have arisen per saltum. Fish, for example, would have been obliged to lose their fin rays before replacing them with autopods.19 To ensure the proper, orderly, and coordinated growth of an organ like a hand, the position and the timing of the expression of all the genes involved must be rigidly controlled, as must the regulation of the growth and development of all the constituent tissues.
The autopod, Sean Carroll has claimed, “…evolved because a set of vertebrate Hox genes have acquired a new switch or set of switches which activate them in a new distal part of the embryonic limb.”20 But the overall process must have involved far more than a few switches, as Carroll noted. “There were many other developmental changes and genes involved in the shaping of the autopod. Other genes … acquired digit-specific switches, and all of the soft tissues—tendons, ligaments, and muscles—and the genes that control their formation and patterning evolved as well.”21
No matter the Darwinian interpretation, there is a significant break between fin and limb. The new evo-devo picture provides no support for a gradualist, functionalist scenario. Trying to envisage the process, hypothetically, as a product of gradual natural selection, poses Herculean challenges. Some of the earliest tetrapods had eight digits, some seven, some six. What adaptive purpose did the eight digits serve in Acanthostega? Why seven in Tulerpeton? What environmental pressure demanded these numbers? And what was so magical about the number five that it became the canonical digit number in post-Devonian tetrapods? Surely causal factors other than natural selection were involved? Perhaps pentadactyly arose from the pre-existing genetic architecture, which channelled the development of the limb toward the five-digit design; or perhaps selection for some unrelated aspect of early amphibian biology gave rise to the constraint. But cumulative selection acting on an interminable series of slightly different autopod designs was not the main causal agency.
This is a trivial problem compared to the nightmare of attempting to explain, in Darwinian terms, why the basic design of all ten digits is the same. Just as the fore and hind limbs always differ in all adult tetrapods, so do the digits in both the hand and foot, sometimes dramatically. In no species are the digits in the hand and the foot identical. Pterosaurs, flying reptiles that existed from the late Triassic to the end of the Cretaceous Period, had a massively enlarged fourth digit supporting their wings. In bats, the wing is supported by an enlargement of digits two, three, four, and five, although the first digit is of normal size. The thumb and the big toe in human beings have two phalanges; the other digits, three. The middle finger of the human hand is longer than the adjacent digits. In the foot, the size of the digits decreases from the big toe to the little toe, which is almost vestigial.
Not only do the digits exhibit different morphologies, each performs a slightly different function. An underlying plan or homology was adopted or redeployed in all digits. Until digit identity is imposed, none of the digits can acquire a different morphology; but digit identity only has adaptive utility when the digits exhibit a different morphology. This poses an additional challenge to Darwinian explanations.
What applies to the homology of the ten digits also applies to the fore and hind limbs, which are never identical. This would suggest that utility demands a different form for fore and hind limbs. Common sense says that if this is so, it must have been true for the mysterious ancestor that was gradually acquiring the form of its limbs under the supervision of natural selection. There cannot have been two identical but separate functional continua leading to identical fore and hind limbs.
Recent studies have shown that some of the basic developmental processes generating the fore and hind limbs in tetrapods also generate the pectoral and pelvic fins in fish. Therefore the lateral appendages are homologous throughout the vertebrate series.22 The similarity in the underlying ground plan of fore and hind limbs in tetrapods may thus be partially explained by inheritance; that is, early amphibians inherited a ground plan and developmental program from their fish ancestors, which specified the same pattern in their two main lateral appendages.
How do we explain this serial similarity?
In no adult fish are the forms of the pectoral and pelvic appendages identical.23 How could selection have led to the same ground plan for pectoral and pelvic fins in fish? If digit difference is the adaptive state, the imposition of the same Bauplan on all ten digits might be considered maladaptive. Similarly, the redeployment of the same Bauplan from fore-to-hind or hind-to-fore must likewise have been maladaptive, because no adult organism actually uses two morphologically identical fore and hind limbs.
Since I wrote Evolution, no Devonian vertebrate has been found with a partially evolved autopod; none has been found with both fin rays and digits in any lateral appendage; none has been found with a pectoral fin and a tetrapod hind limb, or vice versa; and all the new evidence from evolutionary developmental biology tells against the selectionist scenario.
The Feather
In 1914, the British naturalist Alfred Russel Wallace considered the feather, and found himself impressed:
Looking at it as a whole, the bird’s wing seems to me to be, of all the mere mechanical organs of any living thing, that which most clearly implies the working out of a preconceived design in a new and apparently most complex and difficult manner, yet so as to produce a marvellously successful result.24
The origins of the enucleate red cell and the tetrapod limb are difficult to explain because it is hard to say why they have adaptive utility. But no one doubts the utility of the feather; nor does anyone doubt the utility of the stages between the simple hollow follicle to the closed pennaceous contour feather of the modern bird. But accounting for the origin of this series of novelties in terms of cumulative selection is just as difficult.
When I wrote Evolution, the story of the evolution of the feather was dominated by the paradigm of the frayed reptile scale, described in 1926 by Gerhard Heilmann:
By the friction of air, the outer edges of the scales become frayed, the frayings gradually changing into still longer horny processes, which in the course of time become more and more feather-like, until the perfect feather is produced.25
This idea had merit, in that it suggested the means by which the novel form of the feather could be built, bit-by-bit, through cumulative selection. I expressed scepticism in Evolution about such schemes, and cited Barbara Stahl’s lucid Vertebrate History: “How they arose from reptile scales defies explanation.”26
Skeptics have been thoroughly vindicated.27
Richard Prum and his colleague Alan Brush, in a landmark article in The Quarterly Review of Biology,28 wrote that:
[o]ver the last half of the 20th century, neo-Darwinian approaches to the origin of feathers … have hypothesized a micro-evolutionary and functional continuum between feathers and a hypothesized antecedent structure (usually an elongate scale). Feathers, however, are hierarchically complex assemblages of numerous evolutionary novelties—the feather follicle, tubular feather germ, feather branched structure, interacting differentiated barbules—that have no homolog in any antecedent structures … [Such g]enuine evolutionary novelties are distinct from simple microevolutionary changes in that they are qualitatively or categorically different from any antecedent or homonomous structure.29
Although the origin of the feather appears to have arisen via a succession of novelties, the new picture provides no evidence that any of the novelties leading to the feather were the result of cumulative selection.
Many of the genes and developmental systems utilized in feather morphogenesis, such as the pattern-forming genes Shh and Bone morphogenetic protein 2, or BMP-2, predated the origin of the feather. Both are widely utilized in the development of hair, limbs, digits, and teeth.30 During feather development, as Prum and Brush pointed out in another paper, the two so-called toolkit proteins, Shh and BMP-2, “[w]ork as a modular pair … The Shh protein induces cell proliferation, and the Bmp2 protein regulates the extent of proliferation and fosters cell differentiation.”31 The two proteins are used repeatedly throughout feather development from the initial formation of the placode to the pattern for the helical growth of the barb ridges.32 These remarkable developmental genetic advances, while providing evidence that the origin of the feather involved the redeployment of existing gene circuits, provide no support for the claim that the redeployment was the result of gradual micro-evolutionary processes.
The first key innovation in the evolution of the feather was the hollow feather follicle. This has no antecedents in any reptile scale or vertebrate skin appendage. What Prum called the defining feature of the feather, is a unique epidermal invagination leading to the growth of a hollow tube. Without this innovation, there would be no subsequent development of barbs, no helical growth pattern generating the rachis, and no closed pennaceous feather. The tubular nature of the feather is, therefore, the primary novelty upon which all the subsequent innovations leading to the modern, mature pennaceous feather are built. As Prum and Bush stressed,
[a]n important implication of the developmental model of the origin of feathers is the recognition of the essentially tubular (or hollow cylindrical) nature of feathers. The cylindrical follicle and feather germ are general features that characterize all feathers, therefore they should be considered the defining features of feathers [emphasis added]. Prum defined a feather as an elongate, cylindrical, or tubular epidermal appendage that grows from an invaginated feather follicle. A feather follicle differs from a hair follicle in that the follicular invagination is not merely a depression in the epidermis but a circular trough that encircles a persistent dermal papilla.33
As Prum commented to Thor Hansen,
[i]f it’s a hollow tube it’s a feather … One thing I keep saying again and again is that there’s no such thing as a “protofeather.” No one talks about a “protolimb.” You either have a limb or you don’t. Why should feathers be any different? If it’s a tube, it’s a feather. Period.34
Scales and feathers develop from an epidermal outgrowth or placode, but unlike scales, the outgrowth takes on the form of a papilla; the epidermal tissue at the base of the papilla then undergoes a novel invagination to form a cylinder of epidermal tissue, which develops into the tubular feather follicle. The initial stages of follicle formation of the placode involve a proliferation of epidermal cells above a condensation of dermal cells. This is also a novelty.35 Neither Prum, nor any other author, has provided a Darwinian scenario for any of this.
After the hollow follicle has been established, the mature closed pennaceous feather results from a succession of fascinating, novel developmental mechanisms that have no homologue in any other avian or reptilian scale, or indeed any vertebrate integumental structure.36 These include:
- The development of parallel barb ridges in the follicular collar (observed today in the plumaceous feather).
- The helical growth of the barb ridges from the posterior midline of the collar to the anterior midline where they fuse together to form the rachis.
- The formation of barbules, distal and proximal.
- The development of hooks on the distal barbules, and interlocking grooves on the proximal barbules that hold the barbs together to form the closed pennaceous vane of the ordinary contour feather.37
These developmental stages are presumed to have been acquired stepwise during the course of evolution. A still-unresolved issue is the stage at which the barbules originated. It might have been before or after the origin of the branched barbs. In modern feathers, the barbs of both pennaceous and plumaceous feathers have barbules. Those of the closed pennaceous feather are short and interlocked by hooks and grooves on the distal and proximal barbs, respectively; those of the plumaceous feather are elongated “barbules with nodal prongs that interact among barbs to form disorderly tangles that produce a large volume.” The result is a ball of fluff with excellent insulating qualities.38
No matter their order, they are all novelties. Prum again:
Many features of feathers and feather development meet this definition and qualify as evolutionary novelties. The follicle, the differentiated sheath and feather germ, differentiated barb ridges, barb rami, barbules, differentiated pennulae of the proximal and distal barbules, and the rachis are all evolutionary novelties.39
These novelties are without antecedent, and so are “the derived mechanisms by which these novel structures develop.”40
In 2005, Matthew Harris et al. demonstrated that the parallel barbs were generated by a two-component activation-inhibition mechanism, while the helical twist and the joining of the barbs to the anterior rachis involved an additional component, making a three-component Turing reaction.41 A great many changes must also be in place to cause helical displacement toward the anterior of the feather. These are bound to include many novel gene expression patterns in addition to those associated with the regulation of two different Turing mechanisms. Some indication of their complexity may be gleaned from a 2002 paper on feather evolution in Nature:
We show that the antagonistic balance between noggin and bone morphogenetic protein 4 (BMP4) has a critical role in feather branching, with BMP4 promoting rachis formation and barb fusion, and noggin enhancing rachis and barb branching. Furthermore, we show that sonic hedgehog (Shh) is essential for inducing apoptosis of the marginal plate epithelia, which results in spaces between barbs. Our analyses identify the molecular pathways underlying the topological transformation of feathers from cylindrical epithelia to the hierarchical branched structures, and provide insights on the possible developmental mechanisms in the evolution of feather forms.42
Every aspect of the feather’s origin challenges Darwinian scenarios.
Without apoptosis of the cells between the barbs and barbules, neither barbs nor barbules could exist as separate filamentous structures. Which came first: the cellular condensations that created the barb, or the apoptosis that separated them into discrete filaments? Only if both developmental processes are in place can the adaptive end of a branched feather be realized. The reduction of feather origins to Darwinian scenarios conflicts with what is known of the developmental processes underlying the ontogeny of the feather.
Just as digits have individual identities, so do feathers. Not only are feathers individualized, so are regions of feathers—as seen, for example, in the different patterns painted on individual feathers like the eye on the display feathers of a peacock. Which came first, individualization or the imposed pattern? How can a genetic program see a region of a particular feather before the feathers and their parts have been genetically and developmentally differentiated? Of what utility is individualization, especially of different regions of feathers, before it is put to use? The individuation of one feather logically implies the simultaneous individuation of others—and thus a saltation.
Research has put paid to the idea of a frayed reptile scale. Feathers and scales are fundamentally different. Although the feather can be followed in the fossil record, there are no known adaptive sequences leading gradually to each of the novelties or new homologues.
The Flowering Plants
The sudden appearance of the angiosperms, I observed in Evolution, “is a persistent anomaly which has resisted all attempts at explanation since Darwin’s time.”43 How true. No real flowers are found in any group of plants save those extant today, and no putative ancestral group has been identified in the fossil record, or by molecular phylogenetics. There is no universally accepted set of transitional forms leading up to earliest angiosperms.44
A great deal is now known regarding the developmental genetics of the flower, including the basic, so-called ABC developmental system.45 It is clear from these evo-devo studies that the recruitment of pre-existing gene circuits and pre-existing parts in angiosperm ancestors led to the formation of the flower. But, as the evolutionary biologist Günter Wagner noted, what is unique about flowers is the deployment of the various pre-existing parts and pre-existing gene circuits, and the way they became developmentally integrated into the flower.46 These new insights say nothing about the causal mechanism that brought about this redeployment.
This is not to rule out the possibility that the evolution of the flower occurred stepwise. Wagner alluded to two probable steps. The first involved the unification of male and female organs on the same shoot axis. (In the more primitive gymnosperms, the reproductive organs are separate.) The second involved:
[I]ntegrating subdenting leaves as the perianth [the non reproductive parts of the flower, i.e. sepals and petals] into the flower … Hence the flower was the product of character integration that occurred in at least two steps: the integration of male and female organs and the integration of sterile leaves to become the perianth.47
No adaptive continuum that might have gradually led to these advances has ever been proposed. Richard Bateman et al. confessed that the “long branch separating the latest diverging extant gymnosperms from the earliest divergent extant angiosperms remains a recalcitrant barrier to confidently reconstructing the first flower.”48
In addition to the flower, another defining feature of the angiosperms is the unique pattern of cell divisions and nuclear movements in the angiosperm ovule that lead to the development of the female gametophyte and egg sac. This is the process of megagametogenesis. The most common developmental pattern of the female gametophyte is termed the polygonum type. The first phase of the polygonum-type pattern commences with a diploid megaspore mother cell, which undergoes meiosis to produce four haploid megaspores. Cell walls develop around the two cells formed after the first meiotic division, and around the four cells produced by the second meiotic division. Next, three of the megaspores undergo apoptosis. This stage—megasporogenesis—is not unique to angiosperms; it also occurs in gymnosperms, a group that includes the conifers. The subsequent cellular and cytological events leading to the generation of the gametophyte and egg sac are unique to the angiosperms.
The lone remaining megaspore then undergoes three successive mitotic divisions to produce eight haploid nuclei, enclosed within the embryo sac. The eight nuclei are arranged in precise positions. One nucleus is located near the opening of the embryo sac in the egg cell. Two nuclei are positioned on either side of the egg cell, in cells called synergids. Three other nuclei reside in cells called the antipodals, located at the end of the sac, opposite the egg cell. The remaining two nuclei, known as the polar nuclei, are positioned in the center of the egg sac, in the cell known as the central cell.49 The mature egg sac is therefore composed of eight haploid nuclei and seven cells—the egg cell and its two flanking synergids, the three antipodal cells, and the central cell.
There are many variations on this standard pattern, which is termed monosporic because only one megaspore cell survives. In bisporic species, cell plate formation occurs only after meiosis one, and results in two, two-nucleate megaspores, of which one degenerates. The tetrasporic pattern is characterized by cell plates failing to form after either meiosis one or two, and results in one four-nucleate megaspore. Therefore, each ontogenic pattern or trajectory gives rise to a single functional megaspore that contains one, two, or four haploid nuclei, respectively.50 The megaspore then undergoes gametogenesis to give rise to the female gametophyte. In monosporic species, this involves three mitotic divisions; in bisporic species, two; in tetrasporic species, one. This is only a fraction of the variety of ways of getting from the megaspore to the angiosperm gametophyte. In the peppers, there are at least seven variations on this weird cytological choreography.51
There are grounds for believing that all of these are variations on an underlying pattern or theme. In a recent paper William Friedman and Joseph Williams commented that:
Within the basal angiosperm lineages … female gametophytes are characterized by a single developmental module that produces a four-celled/four-nucleate structure with a haploid uninucleate central cell. A second pattern, typical of Amborella and the overwhelming majority of eumagnoliids, monocots, and eudicots, involves the early establishment of two developmental modules that produce a seven-celled/eight-nucleate female gametophyte with two haploid nuclei in the central cell [the polygonum type].52
The next taxa-defining event in the angiosperm reproductive cycle, after the formation of the female gametophyte, is the fertilization of the egg, which occurs when the pollen tube reaches the embryo sac’s opening. Each pollen tube contains two sperm, both haploid. In the standard polygonum model, one sperm cell fertilizes the egg cell to form a diploid zygote, from which the embryo develops. The other sperm cell fertilizes the large central cell in the middle of the embryo sac, joining the two polar nuclei. Each of these is haploid. This gives rise to a triploid cell, which develops into the endosperm. This bizarre double fertilization is unique to angiosperms.
The movement of the second sperm nucleus to the center of the egg sac necessitates direction by cytological mechanisms of great complexity.53 In 2003, writing for the New Phytologist, the biologist Valayamghat Raghaven conveyed something of the degree of complexity involved:
The journey of the two sperm deposited in the degenerating synergid to align with the egg and the polar fusion nucleus [the nuclei in the center of the egg sac] is considered an arduous one and some attention has been paid to the mechanism by which this is accomplished. It is well established that only the sperm nuclei fuse with their female reproductive targets; the rest of the pollen tube discharge and the sperm cytoplasm remain trapped in the milieu of the synergid. Two aggregates of actin filaments designated as “coronas” that presumably guide the pathway of the male gametes have been identified within the embryo sac of tobacco. One of the actin aggregates forms at the chalazal end of the degenerating synergid, extending from its middle lateral region to the region of the egg. The second band occurs in the interface between the egg and the central cell and extends from the side of the egg to the region of the polar nuclei … But actin constitutes only one of the two principle proteins of a possible actomyosin-based transport system of the sperm; in the absence of a clear demonstration of the presence of myosin on the surface of fertilization-prone sperm cells, a different type of regulatory machinery involving actin coronas to haul the sperm to their destination cannot be ruled out.54
Moreover, there are again variations on the basic theme. In the most common case, the endosperm is formed from three haploid nuclei and is triploid; in other cases, the variation in the preceding cell divisions, and the movement of nuclei during embryo sac development, result in a greater number of nuclei in the central cell. In some cases, the central cell and subsequent endosperm may contain 14 haploid nuclei.55
No doubt some of the variations have adaptive consequences. Selection might favour endosperms with higher ploidy;56 and in fact, higher ploidy does have a number of advantages. It means higher heterozygosity of the endosperm cells and hence enhanced fitness.57 Moreover, the number of gene copies is increased, and this, in turn, may increase the rate of gene transcription and endosperm growth. Increased ploidy has also been seen to have other genetic advantages.58
Even if the increase in ploidy can be rationalized in terms of selective advantages, there is no evidence that the angiosperm reproductive system came about as the result of a series of gradual adaptive steps. Consider the problem of moving the sperm nucleus to the center of the egg sac to fuse with the two polar nuclei. How could machinery of such complexity have been put in place before increased ploidy could be tested for utility? What was its function before it was redeployed to move the sperm nucleus to the center of the egg sac?
As in so many other cases, trying to envisage a series of tiny adaptive steps by which novelties arise in evolution poses intractable problems. Despite the increase in our knowledge of the cellular and molecular events involved, double fertilization remains a “puzzle, despite a century of research.”59
The Revolution in Complexity
In Evolution I wrote:
In practically every field of fundamental biological research, ever-increasing levels of design and complexity are being revealed at an ever-accelerating rate. It would be an illusion to think that what we are of at present is any more than a fraction of the full extent of biological design. The credibility of natural selection is weakened … by the expectation of [the future uncovering of] further undreamt of depths of ingenuity and complexity.60
The revolution in complexity has continued unabated. This will be the subject of the third part of my essay. Many recent papers have suggested as much.61 In a 2010 Nature article, Erika Check Hayden remarked that the “more biologists look, the more complexity there seems to be.” Biochemist Jennifer Doudna, cited in the same article, added that it “seems like we’re climbing a mountain that keeps getting higher and higher … The more we know, the more we realize there is to know.”62
This is precisely what I suggested in 1985, and nothing has caused me to revise this view.
End of Part Two.
