When? Why? and How?: Some Speculations on the Evolution of the Vertebrate Integument

AM. ZOOLOCIST, 12:159-171 (1972). When? Why? and How?: Some Speculations on the Evolution of the Vertebrate Integument PAUL F. A. MADERSON Biology Department, Brooklyn College, Brooklyn, New York 11210 SYNOPSIS. The basic structure of the vertebrate integument is briefly reviewed. The system is either scaled, non-scaled, or a mixture of the two. Scales are not appendages of the integument, but are patterned folds in which the dermal and/or epidermal components may be elaborated. An appendage is the product of specialized patterns of cell differentiation localized within the dermis and/or epidermis. Scales, and appendages (whether borne within scaled or non-scaled integuments), can only be correctly defined with reference to the chemical or molecular nature of the end-products of dermal and/or epidermal cell differentiation. Truly homologous integumentary structures probably do not exist above the class level in modern vertebrates. Anatomical, developmental, neurological, and paleontological data are presented in support of a model for the origin of mammalian hair. It is suggested that hairs arose from highly specialized sensory appendages of mechanoreceptor function which facilitated thermoregulatory behavioral activity in early synapsids. Specialization of cellular differentiation within these units led to the appearance of dermal papillae. A chance mutation led to subsequent multiplication of the originally sparsely, but spatially arranged papillae, causing the induction of a sufficient density of "sensory hairs" to constitute an insulatory body covering. The insulatory properties of this "prolopelage" were the subject of subsequent selection, but the sensory function of mammalian hairs remains important. INTRODUCTION The papers presented at this symposium have indicated the wide scope of currently available data on the vertebrate integument, which greatly facilitates an evolutionary review. We can now turn away from those treatments of the past century which have tended to focus on anatomical and embryological differences, and rarely, if ever, considered the problems of function or natural selection with reference to the origin of specific integumentary structures. Initial emphasis will be placed upon denning certain fundamental terms which are important to any discussion of the existence or non-existence of general trends. Then follows a consideration of the problem of deciding whether apparently similar structures have been retained throughout evolution the conservative interpretation or whether the known developmen- The author's studies on the reptilian integument have been supported by N. I. H. Grants CA - 10844 and 1-PO1-AM-15515. Mrs. Una Maderson kindly typed the manuscript. 159 tal plasticity of the integument has permitted the repeated appearance of analogous specializations in convergent response to functional demands the radical view. Finally, the evolution of hair is discussed to illustrate the parameters which should be considered in dealing with the origin of apparently unique integumentary modifications. FUNDAMENTALS While the "mixed" ectodermalmesodermal nature of the vertebrate integument is well-known, less emphasis is placed on the fact that of all the major phyla, only the vertebrates have a multicellular epidermis. This is significant when we recall that the vertebrate integument never forms a confining exoskeleton comparable to that of Arthropods, Molluscs, or Echinoderms. Freedom from direct association with locomotory muscle action has not meant, however, that the vertebrate integument does not reflect locomotory needs. Indeed, it is more likely that the most fundamental patterns of or-

160 PAUL F. A. MADERSON ganization of the vertebrate integument are responses to problems posed by the basic locomotory patterns. Whatever the actual protovertebrate looked like (Berrill, 1955), the small softbodied creature probably possessed an integument similar to that of Amphioxus. Millions of unrecorded years of evolution separate this ancestor from the profusion of early Paleozoic fish forms, but we know that during this period, increase in body size was accompanied by a mechanical strengthening of the body surface. While the reasons for this are debatable (see discussion, Moss, 1968a), the question presents itself as to how the integument could be strengthened at all in an animal whose fundamental locomotory pattern depended on free lateral flexure of the body (Gray, 1968). Easily envisaged intermediates, with obvious selective advantages, at least for mechanical protection, lead eventually to either a partial abandonment of the body mobility "the turtle strategy" or else folding. As a result of the latter, any one segment of the body axis became covered by two or more units which could move relatively freely over one another. Since either the epidermal and/or dermal components of such units could thereafter be strengthened, this offered possibilities for mechanical strengthening while retaining the fundamental functional requirement of lability of the organ system in toto. We recognize these folds as "scales," which can therefore be defined as serial, patterned folds of the integument in which the epidermal and/ or dermal components may be variously elaborated so that one or the other type of tissue may be present in greater quantity, or be superficially more obvious, than the other. Within the definition of a scale given above, we can describe the integument of any vertebrate as being "scaled," "nonscaled," or a mixture of the two. In the case of those forms which definitely do not have scaled integuments, e.g., cyclostomes, elasmobranchs, holocephalans, anguilliform teleosts, most modern amphibia, birds, and most mammals, it is most probable that they are derived from ancestral stocks whose integument was scaled. Furthermore, the integument of each of these taxa is characterized by the presence of complex derivatives various multicellular glands, dermal denticles, hairs, and feathers. These structures are fundamentally localized centers of specialized epidermal and/or dermal cell proliferation and differentiation, within an otherwise generalized integument, of which they may properly be described as "appendages." Analagous structures may be found within scaled integuments, in which case the appendages are borne upon (epidermal specializations) (Maderson, 1971), or contained within (dermal ossifications) (Moss, 1972), individual scales. Thus, if a "scaled integument" is made up of scales, logically any individual scale is a part of the integument, and cannot therefore be regarded as an appendage. This distinction is pertinent to any discussion of integumentary evolution. Where the adult integument is scaled, the epidermal-dermal cell populations over the embryonic body surface were originally sub-divided into developmental fields. Within these fields, appendages may subsequently differentiate. As will be discussed later, the evolution, embryogenesis, and adult distribution of hairs and feathers (Maderson, 1972a) can only be understood by relating them to such developmental fields. Vertebrate integumentary structures can only be defined accurately if one combines the descriptive terms mentioned above with a reference to the chemical or molecular nature of the material synthesized by the constituent cell populations (Table I). The term "dermal scale," so often used to describe integumentary structures in piscine vertebrates, has little meaning unless one refers to the specific end-product of the interaction between dermis and epidermis in any particular taxon (Moss, 19686, 1972). Similarly, the term "reptilian scale" has no exact meaning since the differential distribution of keratinaceous proteintypes across the lepidosaurian and ar-

VERTEBRATE INTEGUMENTARY EVOLUTION 161 TABLE 1. A general characterization, of the integument of extant vertebrates following the terminology and definitions discussed in the text. Taxon Cyclostomes Chondrichthyes Sarcopterygians Actinopterygians Amphibians Chelonia Archosauria Lepidosaurs Birds Mammals General description 1 Appendages 2 Most conspicuous features 3 Unsealed Unsealed Scaled Scaled Unsealed Scaled* Scaled Scaled Mixed Unsealed Yes Yes Yes Yes No No No No Yes Yes Unicellular epidermal mucous glands Denticles* Dermal ossification with superficial COSMINE layer* Dermal ossifications with a variety of superficial mineralizations* Weak epidermal keratinization: dermal ossifications in some scaled apodans Varied horizontal distribution of epidermal keratin types Horizontal alternation of a- and /3-epidermal keratin types: dermal ossifications in many regions Vertical alternation of a- and ^-epidermal keratin types: dermal ossifications in many lizards Feathers of j3-keratin* arising from a-synthesizing general epidermis: horizontal alternation of a-and j9-keratin types on leg scales Hairs of o-keratin* arising from a-synthesizing general epidermis: dermal ossifications in some forms 1 Applies to the great majority of species in the taxon cited. 3 Only those appendages are mentioned which are usually cited as primary diagnostic features of the group. a Structures or features which are known to involve dermal-epidermal interactions arc marked thus *. 4 The body is primarily scaled, but the development of the carapace, with its associated dermal ossifications, obviously inhibits flexibility. Data from: Alexander (1970); Baden and Maderson (1970); Moss (1968a,6); Quay (1972); Spearman (I960). chosaurian scale surfaces (Baden and Maderson, 1970) makes these units as different in their own way as are feathers and hairs. The integumentary morphology of piscine fossils is usually clearly demonstrated by impressions in the surrounding matrix, but we need some "rule-of-thumb" for tetrapod fossils. Many extant squamates have scales which do not contain dermal ossifications. However, with the exception of Dermochelys (the leatherback turtle), I know of no living tetrapod which normally has a wide-spread distribution of dermal ossifications which does not have a visibly scaled integument. While this does not necessarily indicate a 1:1 relationship between externally recognizable units and individual ossification centers (Zangerl, 1969), it does suggest that in those systems where developmental fields exist in the embryonic integument and produce a pattern of dermal ossification, similar fields influence the topography of the entire integument. Therefore, I suggest that if paleontologists describe "scales" (dermal ossifications) in their material, the forms concerned probably had scaled integuments in the sense defined earlier. Was the primitive tetrapod epidermis keratinized? Spearman (1966) indicated that the potential for keratin synthesis is widespread among vertebrates, and the reports on the ultrastructure of epidermal cells (Flaxman, 1972) show that all epidermal basal cells contain the 70-80A wide filaments which are associated with a-keratin. However, it is also known that in those tissues where the /?-protein is synthesized (characterized by 30A wide filaments), the 70A filaments occur first, and the 30A units appear later and eventually fill the cells. To me, this implies that the /?-protein is a later phylogenetic development than the a-form, and this is supported by the distribution of epidermal protein types in extant amniotes (Baden and Maderson, 1970). It appears that those lower Pennsylvanian captorhinomorphs which gave rise to synapsids and mammals possessed only the capacity to synthesize a-keratin. The remainder of the cap-

162 PAUL F. A. MADERSON FIG. 1. Sagittal section through ventral body scales of the gekkonid lizard Eublcpharis macularius just before skin-shedding. The ^-layers of the outer (/So) and inner (fii) epidermal generations are thick on the outer scale surface (OSS), but are reduced to a single layer of cells on the inner surface and in the hinge region (ISS, H). Ddermis; sc- sub-cutaneous tissue. torhinomorphs, which gave rise to all the other reptilian groups and birds (Carroll, 1969(1$/:), possessed an additional capacity for /3-protein synthesis in their epidermis, which was variously expressed in different lineages (Table I). What then of the paleozoic amphibia? Romer and Witter (1941), Colbert (1955), and Kitching (1957) described ossified units suggesting a scaled integument (see above) which was secondarily modified in their lissamphibian descendents (Cox, 1967). Findlay (1968a) suggested that haematite deposits around the matrix of the lower Triassic Uranocentrodon resulted from the decomposition of sulphur-containing epidermal proteins. While this intriguing interpretation suggests the presence of keratin, it does not reveal whether it was of the a- or /J-variety! Microscopic and ultrastructural studies indicate that the epidermal tissues on the inner surface and hinge region of amniote scales tend to be thinner, less compact, and more lamellate in their organization than those on the outer scale surfaces. Different fluorescent properties of different regions of amniote scales (Cane and Spearman, 1967; Spearman, 1964, 1966, 1967) cannot be explained by reference to the presence of a- or /J-keratins alone (Baden and Maderson, 1970). However, they may reflect differences in inter-cellular bonding, which endow the different epidermal regions with different mechanical properties, and these originally augmented the flexibility of the entire integumentary system. This end is still extremely important in squamates where numerous subtle differences in patterns of cell production and differentiation modify the basic epidermal generation pattern (Maderson, 1965, 1966; Maderson and Licht, 1967) over the inner scale surface and hinge (Fig. 1). However, the persistent a-protein in these regions in crocodiles and birds (Baden and Maderson, 1970) and centers of granular layer formation in mammalian tail scale "hinges" (Spearman, 1964, 1966) should be interpreted as relics of the ancestral functional modifications. TRENDS IN VERTEBRATE INTEGUMENTARY EVOLUTION Raising the question of the possible homology between feathers, hairs, and scales, Cohen (1964) wrote: "If by homology we mean that the organs concerned, may, we believe, be traced back along lines of ancestors until a comparable structure is reached in the common ancestor, then the assessment is always made more difficult by more facts." This conclusion is germane to the entire topic of integumentary evolution. On the basis of the facts presented above and their combination with the most conservative possible deductions regarding possible integumentary anatomy in fossil forms, we are forced to conclude that no two integumentary features in two major assemblages can be strictly considered to be homologous. This concept must be restricted to such examples as pelage hair and spines in mammals, or climbing setae and the normal Oberhautchen in lizards (Maderson, 1970). Even the recognition of general anatomical trends is of limited value. While piscine vertebrates tend to have scaled integuments or conspicuous elaborations of dermal skeletal structures or both, attempts to define the degree of homology therein are important only insofar as they lead to consideration of whether dermo-epidermal interactions have or

VERTEBRATE INTEGUMENTARY EVOLUTION 163 phological diversity. THE EVOLUTION OF HAIR FIG. 2. An epidermal "Haareorgane" from the dorsal body scales of the gekkonid lizard Gekho gecko. The epidermis shows a stage 4 condition of the shedding cycle (Maderson and Licht, 1967) and shows that the "hair" derives from a modified Oberhautchen cell (SpOb). In a sense organ of this type, although the structure of the epidermal generation is modified, the subjacent germinal cells (sg), closely resemble those of the adjacent non-specialized epidermis. Note the cluster of cells in the dermis (X) beneath the sense organ here and in Figures 4 and 5. The /3-layer of the outer epidermal generation is not seen in the photograph. Other abbreviations, here and in Figures 4 and 5: oo o-layer of the outer generation; /3i /3-layer of the inner generation; clo clear layer of the outer generation; lto lacunar tissue of the outer generation; mi mesos layer of the inner generation; Obi Oberhautchen of the inner generation; Obis spinules of the unspecialized Oberhaulchen cells. have not changed during evolution. The question "Are tetrapod scales retained from the piscine ancestors?" has no meaning except to emphasize that there is a general capacity for patterned integumentary structure in different taxa with varying degrees of phyletic affinity. Whatever general trend we define or recognize, it is always subject to major or minor revision of execution. In short, I favor the "radical" view of integumentary evolution to such a degree that I would suggest that in any instance, a functional question should be asked, a functional investigation should follow, and any subsequent detailed anatomical study should be expected to demonstrate yet another example of mor- This problem has a number of facets. First, we must ask, is hair a unique mammalian characteristic? Second, are there other structures which resemble mammalian hair in other vertebrates, or indeed in other animals? Third, have hairs always served an insulatory function, and if not, what other functions could they have served? Finally, is it possible to present a model for the steps in the phylogenetic development of hair, with plausible explanations for the accompanying selective pressures? Recent reviewers (Hopson, 1969; Hopson and Crompton, 1969; Jenkins, 1970) suggest a monophyletic origin for mammals in the late Triassic - early Jurassic. Hopson (1969) concluded: "... (anatomical, physiological and neuroanatomical studies) strongly suggest that the common ancestor of monotremes and therians was also mammalian in a majority of essential features e.g. hair, lungs, diaphragm, heart, and kidneys, to name a few." How many of these features might have characterized the early Triassic cynodonts which Hopson and Crompton (1969) proposed as mammalian ancestors? Reference to possible integumentary structures of therapsids is so common-place that we may tend to forget that there is no direct information available. Watson (1931), Brink (1956), and Findlay (1968ft) interpreted depressions in skull bones as probably having housed vibrissae or "skin glands of a sweat gland nature" (Brink, p. 87). These interpretations were extrapolated to suggest pelage hairs and normal sweat glands over the rest of the body. Repeated associations between these "extrapolated interpretations" and actual mammal-like osteological features in support of suggestions of endothermy in therapsids have produced a situation so close to circuitous argument that it is time to seek a new approach to the problem of the origin of hair. No extant vertebrates have integumen-

164 PAUL F. A. MADERSON tary appendages which anatomically resemble hairs. The structures seen in many lizards (Fig. 2), once invoked as "ancestral hairs" (Elias and Bortner, 1957), are sensory units (Miller and Kasahara, 1967) derived from individual cells of the Oberhautchen (Schmidt, 1920; Maderson and Licht, 1967; Maderson, 1971), which layer is a unique constituent of the lepidosaurian epidermis (Maderson, 1968a). While the anatomy of the individual units is certainly not homologous with that of any vertebrate epidermal derivative, a number of insects have a "pelage" (Heath, 1968). Although the pelage plays a primary role in insulation in most mammals (Ling, 1970) and in some insects, various vertebrates, e.g., lizards, or man, manifest endothermic regulatory mechanisms of varying degrees of "perfection," but do not possess a continuous body covering of this type. Conversely, the presence of a covering pelage does not necessarily indicate an absolutely constant internal temperature throughout life (Heath, 1968). There is therefore no a priori reason for assuming that therapsid thermoregulation could not have evolved in the absence of a pelage. Indeed, the physical laws which govern the functioning of a pelage indicate that each constituent unit must have a certain minimum length, and there must be a certain minimum density per unit area of the body before any selective advantage accrues with regard to insulating function (Ling, 1970). It seems most unlikely that a "preadapted proto-pelage," upon which selection could act, could have appeared via a steady accumulation of "neutral traits" affecting epidermal morphogenesis over several thousand generations. A more plausible hypothesis is that the insulating function of hair is secondary and became possible only after completely different selective advantages had favored suitable morphogenic changes in the epidermis. These primary selective pressures can be identified if we consider the probable ecology of the extinct forms concerned, and thence deduce the obligatory minimal functions of their integument. Studies of Pennsylvania reptile fossils (Carroll, 1964, 1969a,b jc, 1970a,b) suggest that they were small, highly terrestrial, forest-dwelling forms. Carroll (1970&) writes of the captorhinomorph Hylonomus lyelli: "in size and general form it resembles a medium-sized lizard. It may have had similar habits as well." I suggest that functionally the integument of such forms would have resembled that of modern lizards. The epidermis would have possessed a well-developed outer cornified region which would have provided a degree of protection against dessication (Maderson et al., 1970). Carroll's (1964) descriptions of osteoscutes suggest a scaled integument (see above), so that both dermal and epidermal components probably contributed to mechanical protection. Since holocrine secretion is a very important function in modern lizards (Maderson, 1970), this may have been true for the earliest reptiles. However, in most modern amniotes, odoriferous sources are localized on the body surface: the pheromonal function of sweat-glands in some mammals is probably secondary. My own observations on a great variety of modern lizards suggest that if behavioral thermoregulation characterized the earliest reptiles, this would not have necessitated any particular morphological structure of the integument, except perhaps with regard to the distribution of pigment cells (Porter, 1967). If the integument of primitive reptiles manifested other secondary functions (e.g., climbing claws, poison glands, sexual or territorial warning appendages), comparative observations on modern amniotes indicate that associated structural modifications would have been localized on the body surface. Apart from the "primary barrier function" of physiological and mechanical protection which influences the fundamental morphology of the entire integument (Maderson, 1971), there is only one secondary integumentary function which potentially involves the entire organ system that of sensory reception. Two quite different types of sensory stimulus have always impinged upon the terrestrial in-

VERTEBRATE INTEGUMENTARY EVOLUTION 165 <?«/*. FIG. 3. Schematic representation of a mammalian tylotrich hair follicle modified after Straile (1969). 1 - hair shaft; 2 - internal root sheath; 3 - external root sheath; 4 - germinal region of the tylotrich follicle; 5 - dermal papilla; 6 - connective tissue sheath; 7 - annular complex; 8 - epidermal pad complex; 9 - neurons associated with slowadapting mechanoreceptors; 10 - mouth of sebaceous gland (body of gland not shown) ; 11 - venular complex associated with tylotrich unit (arteriolar complex not shown). tegument temperature and touch. These categories can be further subdivided since gradual changes in ambient temperature, or casual contact with the substrate during locomotion, are interpreted by the brain quite differently than are sudden temperature changes, or sharp pressures. Any or all of any variety of types, or levels, of stimulation might challenge any part of the body surface. It is therefore predictable that there would be a spatial pattern of functional differentiation across the integument, which might be reflected anatomically in patterns of nerve distribution and/or the morphology of the receptor-transducer units. I propose that mammalian hairs are derived from complex epidermal modifications of mechanoreceptor function, which were originally "sparsely," but regularly, distributed over the surface of the body. At some stage in the evolution of the therapsid integument, the competence of the developmental fields centered around the original units changed, resulting in a multiplication of basically similar morphogenic events. These events produced a sufficient density of "sense organs" per unit area of the body to produce a "pelage," the insulative properties of which were the focus for subsequent selection. The sensory function of hair in modern mammals does not exactly resemble that of the original units, but this does not affect the morphological model which will be presented. The data supporting this hypothesis will now be discussed. Straile (1969) proposed "repeating vertical units" in the mammalian integument containing epidermal, neural, and vascular elements arranged around a "tylotrich" hair follicle (Fig. 3). The tylotrich is associated with two innervated regions, an annular complex surrounding the upper third of the follicle, and an adjacent epidermal pad complex. Although the exact construction of the vertical unit varies across the body, and between taxa, tylotrichs have been observed in monotremes and many therian mammals (Mann, 1968). While there is some disagreement as to

166 PAUL F. A. MADERSON tween the two conditions. There is an impressive variety of epidermal sensory modifications in reptiles (Miller and Kasahara, 1967) (Figs. 2, 4, 5). There are no systematic investigations of any single type available, but the distribution of "Haareorganes" suggests a function of monitoring inter-scale contact (Maderson, 1971). Bailey (1969) demonstrated fast and slow-adapting mechanoreceptors by electro-physiological techniques, but did not provide an anatomical correlation. The anatomy and functioning of the infra-red sensitive cutaneous pit organs in snakes have been extensively studied (Barrett, 1970; Meszler, 1970). Although the mm 4 data are sparse, we can say that cutaneous FIG. 4. Sense organ from a labial scale of the sensory reception does occur in reptiles, iguanid lizard Iguana iguana. The epidermis is in and the diversity of associated morphologithe resting phase of the shedding cycle (Maderson cal specializations suggests that it is an exand Licht, 1967). In this type of sense organ, the tremely important function. associated germinal cells are always columnar and have vesicles at their distal tips. The mature kerelias and Bortner's (1957) morphologiatinized elements of the outer epidermal genercal schema of hair phylogeny rests on the ation are indicated by dotted lines. /So - /3-layer o premise of a direct relationship to the lathe outer generation. certilian "Haareorgane" which is no longfig. 5. Sense organ from a lateral body scale of er acceptable (see above). While the the xantusiid lizard Xantusia vigilis. The epidermis morphogenic events in hair development is in stage 4 of the shedding cycle, and we note are difficult to relate to any possible evoluthat although this type of sense organ does not protrude above the general level of the skin tionary sequence, if one ignores the details surface, it is associated with a modification of the involved, one can derive a useful, simplyhistogenesis of the inner generation which produces stated overview. A portion of the germinal mature elements (dotted outlines) analagous population becomes specialized so that its to those seen in Figure 4. daughter cells stick together as a rod which how the electro-physiological data should projects above the skin surface, while the be related to the anatomical data (see dis- daughter cells of adjacent, unspecialized cussion, Straile, 1969), there is good evi- inter-follicular epidermis do not stick todence that both rapid and slowly adapting gether so tightly and therefore desquamechanoreceptors are represented within mate. The process of initial specialization repeating vertical units so that: "The de- and adult homeostasis involves mesoderm tection of a tactile stimulus moving from cells ("the dermal papilla"), and the two point to point probably involves the inter- cell populations influence one another via pretation of a complex series of nerve im- a sequence of inductive processes which we "epithelial-mesenchymal interacpulses that are received by the brain" term tions" (see Kollar, 1972). The elegant (Straile, 1969). My own familiarity with the scaled reptilian integument, where complexity of the hair follicle, which at even cursory examination reveals a pleas- first sight seems so difficult to explain in ing geometric order, has long made me evolutionary terms, may be readily explained. It permits a "good rod" to grow suspicious of the apparent heterogeneity of from a "good hole," a mere refinement the mammalian system. Straile's "re- which could have occurred quite late in peating vertical unit" seems to me to the phylogeny of hair. It should be noted provide the required conceptual link be-

VERTEBRATE INTEGUMENTARY EVOLUTION 16? age (Carroll, 1964, 1969a,b,c, 1970a,6; Clemens, 1970; Hopson, 1969; Hopson and Crompton, 1969; Jenkins, 1970) indicate that the creatures concerned were of small size less than 18" total length. However, throughout the geological eras concerned, related forms and other amphibian and reptilian groups radiated to produce genera of considerable size. The mammalian grade of organization with its attendant morphological characteristics, e.g., hair, was perfected over a period of 200 million years by small animals, possibly crevice dwellers, who probably attained their evolutionary destiny by exploiting nocturnal niches, following gradual refinement of thermoregulatory mechanisms. FIG. 6. Suggested model for the sequence of morphological changes in the evolution of mammalian hair, a - suggested integumentary structure of a primitive cotylosaur; b - suggested integumentary structure of a cotylosaur associated with synapsid lineage; c - suggested integumentary structure of a pelycosaur associated with the therapsid/mammalian lineage; d - e - magnified views of suggested evolutionary changes in the original "hinge" region of c (outlined). Medium dense fine stipple epidermis showing basic a-protein synthetic capacity; dense fine stipple epidermis where cell maturation involves keratohyalin; sparse fine stipple dermis; clusters of fine stipple dermal papilla; dashed lines neurons associated with fast-adapting mechanoreceptors; heavy dotted lines neurons associated with slow-adapting mechanoreceptors; cross-hatching dermal ossification. For explanation, see text. that holocrine secretion from lacertilian pre-anal organs (Maderson, 19686, 1970, 1971, 19726) frequently produces a durable "rod" of mature cells which may protrude a considerable distance from the skin surface. However, there are no specializations comparable to the various layers of the inner and outer root sheath seen in a hair follicle. Before we consider the model, it should be mentioned that all recent accounts of the amphibian-reptilian-mammalian line- A possible structure of the early cotylosaurian integument is shown in Figure 6a. The epidermis contained only a-keratin, and the tissue was thinned on the inner scale surface and hinge region. A plausible suggestion for the differential distribution of mechanoreceptors would place fast-adapting units on the outer scale surface. These monitored transient environmental contact during normal locomotion. Locomotory activities involving stretching or compression of the integument (e.g., twisting of the body into small crevices and hiding) could have been monitored by slow-adapting receptors in the hinge region. Dermal ossifications were present and possibly played some mechanical protective role. In those cotylosaurs associated with the basic synapsid stock and the derived forms, certain modifications characterized the integument (Fig. 6b). I propose that the involvement of keratohyalin in the keratinization process, at first confined to the hinge region (Spearman, 1964, 1966), enhanced the overall flexibility of the integument and permitted a reduction of scale overlap. This additional protein could have resulted from a single gene change, since similar proteins exist in the epidermis of modern reptiles (Maderson et al., 1972) and birds (Alexander, 1970). Reduction of scale overlap is suggested by the scarcity, or absence, of dermal ossifications

168 PAUL F. A. MADERSON in synapsid material. The modified pattern of protein synthesis may have permitted a thickened epidermis on the outer scale surface to provide mechanical protection and perhaps decrease percutaneous water-loss. The groups of fast-adapting mechanoreceptors probably became localized within regions of hypoplasia. The reduction of scale overlap diminished the original function of the slow-adapting receptors in the hinge region for monitoring inter-scale contact. However, the function of providing sensory data during hiding could have been maintained if the nerve endings became associated with a small epidermal papilla which protruded to a level just beneath the general level of the outer scale surfaces during normal locomotion. The widespread occurrence of thermoregulatory behavior patterns in modern reptiles (Bellairs, 1969) implies that they arose very early in reptilian evolution. If, indeed, Triassic therapsids did manifest some degree of homeothermy (Heath, 1968), we might postulate that the cotylosaur-synapsid lineage possessed some special feature permitting the precocious development of this grade of organization relative to other reptilian lineages. Bailey's (1969) data suggest that cutaneous thermoception in lizards is insufficiently sensitive to facilitate thermoregulatory behavior. Heath (1968) indicated that while peripheral temperature receptors modulate hypothalamic responses to environmental temperature change in mammals, he stated: "The cold-blooded terrestrial animals may rely largely on internal receptors." Such receptors can only provide information regarding heat energy after it has been absorbed; they cannot critically examine possible differential heat-sources in the environment on a "minute-byminute basis." For this reason, the thermoregulatory behavior patterns of modern lizards involve quite sudden movements from one type of exposure to another, followed by equally rapid increases or decreases in deep body temperature (McGinnis and Dickson, 1967). The amount of heat absorbed by the deep body tissues depends upon the cooling influences at the skin surface. Any animal which could monitor such influences, and accordingly adjust its position in the environment, could achieve more subtle temperature regulation. More importantly, it could take advantage of "heat availability situations" which would be beyond the sensory analytical capacities of other forms. Figure 6c is a suggested skin structure of a pelycosaur at the base of the therapsidmammalian lineage. Certain general trends described earlier have continued, i.e., reduction of scale overlap, spread of the granular layer, epidermal thickening. The slow-adapting mechanoreceptors originally seen in the hinge region are now associated with a longer papilla, the primary function of which is still monitoring environmental contact during hiding activity. When the body is still during sun-basking, these papillae protrude above the general body surface. Their mechanoreceptor action could detect displacement of their distal tips by air-movements. Figures 6d-f suggest how further selection might have improved the functioning of this leveractivated mechanoreceptor. Figure 6f shows a rod of cells which grows out from a follicle; movement of the distal tip of the rod distorts the entire structure leading to activation of the neurons which are now associated with the upper third of the follicle. The daughter cells arising from the germinal region of the follicle form a tightly adhering mass, and specialization of the outermost cell layers would endow this "rod" with specific mechanical properties associated with flexibility. These patterns of cellular activity are sufficiently distinct from those of surrounding "interfollicular" epidermal cells to imply the presence of a distinct morphogenetic mechanism responsible for their control. At this stage in evolution, the dermal papilla appeared. It does not matter whether this structure arose initially as an embryonic or an adult "inducer," since its fundamental role maintenance of a specialized sequence of differentiative events for a circumscribed germinal/

VERTEBRATE INTEGUMENTARY EVOLUTION 169 -.** M FIG. 7. Sagittal section through rat tail scale showing a hair follicle growing from the "hinge region." The restriction of the granular layer to the follicle mouth area is indicated by arrows. H hair shaft. daughter cell population within an otherwise homogeneous epidermal system is the same at all stages of the life cycle. The model to this point suggests that although the postulated specialized mechanoreceptor which should be compared to the tylotrich hair follicle (Fig. 3) did not evolve from a scale, it was initially associated with a morphogenic field surrounding a scale and eventually superseded it in size and importance. This premise receives support from the following data. First, tylotrichs develop first in the embryo (Mann, 1968). Second, tylotrichs are more numerous dorsally than ventrally (Mann, 1969) a similar condition is seen with regard to scales in most lizards. Third, recalling that the sequence of events under discussion concerned small animals, we note that Mann (1969) stated: "The larger the mammal, the fewer the tylotrichs per unit surface area of the skin." Fourth, the described sequence of events accounts for hair distribution across the scaled caudal integument of some modern mammals (Spearman, 1964, 1966) (Fig- 7) The integumentary structure shown in Figure 6f might have characterized a small early therapsid with a highly sophisticated thermoregulatory behavior pattern. However, the function of the spatially dis- tributed "eotylotrichs" was exclusively mechanoreceptive, and such structures could not have served an insulatory function. I suggest that this secondary function arose following the multiplication of follicular units within the original "scale morphogenic field" surrounding the eotylotrich. The model suggests an association between a certain level of morphologic complexity and the evolutionary appearance of a dermal papilla. Cohen (1964, 1969) has emphasized the similarity in organization of hairs and feathers, and Maderson (1972a) has proposed the origin of the dermal papilla of feathers for similar morphogenic reasons to those presented here. Ede et al. (1971) investigated the failure of feather development in the talpid3 mutant chick embryo and demonstrated a defect of dermal papilla formation. While comparable detailed ontogenic analyses are lacking, Mann (1969) listed recessive point mutations in mice which disturbed normal tylotrich development. I submit that it is equally possible that multiplication of follicular units could have occurred as the result of a single gene change in our therapsid ancestors. The only question is, what selective advantage accrued which favored the survival and spread of such a gene change? There are two possibilities,

170 PAUL F. A. MADERSON which are not mutually exclusive. The increase in number of units reached or surpassed the minimum density per unit area necessary to provide insulatory benefits. Alternatively, since secondary hair follicles in modern mammals are associated with rapidly adapting touch receptors (Straile, 1969), one could argue that the "leveractivated" receptor associated with the slow-adapting receptors (the eotylotrich) was so successful that selection favored the incorporation of the fast-adapting units into secondarily derived similar structures. I favor the second of these explanations since it does not necessitate quantum changes in anatomical structure and does offer possible successive levels of cutaneous organization, culminating in an insulatory pelage. The model which has been presented here is highly speculative, but this is inevitable due to the nature of the subject. The premise of a mechanoreceptor origin for mammalian hair is not new, but it has never before to my knowledge been considered in detail with reference to a series of selective pressures. I would like to emphasize in conclusion, that if the level of mechanoreceptor organization shown in Figure 6f were typical of most cynodonts, enlarged units could have formed the facial vibrissae discussed earlier. If we accept Heath's (1968) and Bailey's (1969) statements with reference to deep and cutaneous thermoreception in mammals versus reptiles, we might even suggest that it was only in the phyletic line leading to mammals that peripheral modulation of hypothalamic temperature responses developed. This could have been that last subtle refinement in endothermy which ensured the success of the lineage. Right or wrong, this discussion will have served its purpose if it stimulates further interest in mammalian cutaneous reception, but even better, it should lead to comparable studies on modern lizards, the epidermis of which is, after all, the zenith of amniote integumentary evolution. REFERENCES Alexander, N. J. 1970. Comparison of a- and ^-keratin in reptiles. Z. Zellforsch. Mikroskop. Anat. 110:153-165. Baden, H. P., and P. F. A. Maderson. 1970. The morphological and biophysical identification of fibrous proteins in the amniote epidermis. J. Exp. Zool. 174:225-232. Bailey, S. E. R. 1969. The responses of sensory receptors in the skin of the green lizard Lacerla viridis to mechanical and thermal stimulation. Comp. Biochem. Physiol. 29:161-172- Barrett, R. 1970. The pit organs of snakes, p. 247-300. In C. Gans and T. S. Parsons [ed.], Biology of the reptilia, Vol. 2. Academic Press, London. Bellairs, A. d'a. 1969. The life of reptiles. Vol. 1 and 2. Weidenfeld and Nicolson, London. Berrill, N. J. 1955. The origin of vertebrates. Oxford Univ. Press, London. Brink, A. S. 1956. Speculations on some advanced mammalian characteristics in the higher mammal-like reptiles. Palaeontol. Afr. 4:77-96. Cane, A. K., and R. I. C. Spearman. 1967. A histochemical study of keratinization in the domestic fowl (Gallus gallus). J. Zool. (London) 153:337-352. Carroll, R. L. 1964. The earliest reptiles. J. Linnean Soc. London Zool. 45:61-83. Carroll, R. L. 1969a. Problems of the origin of reptiles. Biol. Rev. 44:393-432. Carroll, R. L. 1969b. A middle Pennsylvanian Captorhinomorph and the interrelationships of primitive reptiles. J. Palaeontol. 43:151-179. Carroll, R. L. 1969c. Origin of reptiles, p. 1-44. In C. Gans, A. d'a Bellairs, and T. S. Parsons [ed.], Biology of the reptilia, Vol. 2. Academic Press, London. Carroll, R. L. 1970a. The ancestry of reptiles. Phil. Trans. Roy. Soc. London Ser. B, Biol. Sci. 257:267-308. Carroll, R. L. 1970b. The earliest known reptiles. Yale Sci. Mag. 16-23. Clemens, W. A. 1970. Mesozoic mammalian evolution. Ann. Rev. Ecol. Syst. 1:357-390. Cohen, J. 1964. Transplantation of hair papillae. Symp. Zool. Soc. London 12:83-96. Cohen, J. 1969. Dennis, epidermis and dermal papilla interacting. Advan. Biol. of Skin 9:1-18. Cohere, E. H. 1955. Scales in the Permian amphibian Trimerorhachis. Amer. Mus. Novitates 1740:1-17. Cox, C. B. 1967. Cutaneous respiration and the origin of the modern amphibia. Proc. Linnean Soc. London Zool. 178:37-47. Ede, D. A., J. R. Hinchcliffe, and H. C. Mees. 1971. Feather morphogenesis and feather pattern in normal and talpid a mutant chick embryos. J. Embryol. Exp. Morphol. 25:65-84. Elias, H., and S. Bortncr. 1957. On the ph)logeny of hair. Amer. Mus. Novitates 1820:1-15.

I VERTEBRATE INTEGUMENTARY EVOLUTION 171 Findlay, G. H. 1968a. On the structure of the skin in Uranocentrodon (Rhinesuchus) senekalensis, Van Hoepen. Palaeontol. Afr. 11:15-21. Findlay. G. H. 1968b. On the scalaposaurid skull o Oliveieria parringtoni, Brink with a note on the origin of hair. Palaeontol. Afr. 11:47-59. Flaxman, B. A. 1972. Cell differentiation and its control in the vertebrate epidermis. Amer. Zool. 12:13-25. Gray, J. 1968. Animal locomotion. W. W. Norton and Co. Inc., New York. Heath, J. E. 1968. The origins of thermoregulation, p. 259-278. In E. T. Drake [ed.], Evolution and environment. Yale Univ. Press, New Haven. Hopson, J. A. 1969. The origin and adaptive radiation of mammal-like reptiles and nontherian mammals. Ann. N. Y. Acad. Sci. 167:199-216. Hopson, J. A., and A. W. Crompton. 1969. Origin of mammals. Evol. Biol. 3:15-72. Jenkins, F. A., Jr. 1970. Cynodont postcranial anatomy and the "prototherian" level of mammalian organization. Evolution 24:230-252. Kitching, J. W. 1957. A new small stereospondylous labyrinthodont from the Triassic beds of South Africa. Palaeontol. Afr. 5:67-82. Kollar, E. J. 1972. The development of the integument: spatial, temporal, and phylogenetic factors. Amer. Zool. 12:125-135. Ling, J. K. 1970. Pelage and molting in wild mammals with special reference to aquatic forms. Quart. Rev. Biol. 45:16-54. Maderson, P. F. A. 1965. Histological changes in the epidermis of snakes during the sloughing cycle. J. Zool. (London) 146:98-113. Maderson, P. F. A. 1966. Histological changes in the epidermis of the Tokay (Gekko gecko) during the sloughing cycle. J. Morphol. 119:39-50. Maderson, P. F. A. 1968a. Observations on the epidermis of the Tuatara (Sphenodon punctatus).j. Anat. 103:311-320. Maderson, P. F. A. 19686. The epidermal glands of Lygodactylus (Gekkonidae, Lacertilia). Breviora 288:1-35. Maderson, P. F. A. 1970. Lizard glands and lizard hands: models for evolutionary study. Forma Functio 3:179-204. Maderson, P. F. A. 1971. The regeneration of caudal epidermal specializations in Lygodactylus picturatus keniensis (Gekkonidae, Lacertilia). J. Morphol. 134:467-478. Maderson, P. F. A. 1972a. The structure and evolution of holocrine epidermal glands in the sphaerodactyline and eublepharine gekkonid lizards. Copeia (In press) Maderson, P. F. A. 1972b. On how an archosaurian scale might have given rise to an avian feather. Amer. Natur. (In press) Maderson, P. F. A., B. A. Flaxman, S. I. Roth, and G. Szabo. 1972. Ultrastructural contributions to the identification of cell types in the lizard epidermal generation. J. Morphol. 136:191-210. Maderson, P. F. A., and P. Licht. 1967. Epidermal morphology and sloughing frequency in normal and prolactin-treated Anolis carolinensis (Iguanidae, Lacertilia). J. Morphol. 123:157-172. Maderson, P. F. A., W. W. Mayhew, and G. Sprague. 1970. Observations on the epidermis of desert-living iguanids. J. Morphol. 130:25-36. Mann, S. J. 1968. The tylotrich (hair) follicle of the American Opossum. Anat. Rec. 160:171-180. Mann, S. J. 1969. The tylotrich follicle as a marker system in skin. Advan. Biol. Skin 9:399-418. McGinnis, S. M., and Dickson, L. L. 1967. Thermoregulation in the desert iguana Dipsosaurus dorsalis. Science 156:1757-1759. Meszler, R. M. 1970. Correlation of ultrastructure and function, p. 305-314. In C. Gans and T. S. Parsons [ed.], Biology of the reptilia, Vol. 2. Academic Press, London. Miller, M., and M. Kasahara. 1967. Studies on the cutaneous innervation of lizards. Proc. Calif. Acad. Sci. 34:549-568. Moss, M. L. 1968a. Bone, dentin and enamel and the evolution of vertebrates, p. 37-65. In Biology of the mouth. Amer. Ass. Advan. Sci., Washington, D.C. Moss, M. L. 19686. Comparative anatomy of vertebrate dermal bone and teeth. I. The epidermal co-participation hypothesis. Acta. Anat. 71:178-208. Moss, M. L. 1972. The vertebrate dermis and the integumental skeleton. Amer. Zool. 12:27-34. Porter, W. P. 1967. Solar radiation through the living body walls of vertebrates with emphasis on desert reptiles. Ecol. Monogr. 37:273-296. Quay, W. B. 1972. Integument and environmentglandular composition, function, and evolution. Amer. Zool. 12:95-108. Romer, A. S., and R. V. Witter. 1941. The skin of the Rhachitomous amphibian Eryops. Amer. J. Sci. 239:822-824. Schmidt, W. J. 1920. Einiges iiber die Hautsinnesorgane der Agamiden, insbesondere von Calotes, nebst Bemerkungen iiber diese Organe bei Geckoniden und Iguaniden. Anat. Anz. 53:113-139. Spearman, R. I. C. 1964. The evolution of mammalian keratinized structures. Symp. Zool. Soc. London 12:67-81. Spearman, R. I. C. 1966. The keratinization of epidermal scales, feathers and haits. Biol. Rev. 41:59-96. Spearman, R. I. C. 1967. On the nature of the horny scales of the pangolin. J. Linnean Soc. London Zool. 46:267-273. Straile, W. E. 1969. Vertical cutaneous organization. J. Theor. Biol. 24:203-215. Watson, D. M. S. 1931. On the skeleton of a bauriomorph reptile. Proc. Zool. Soc. London 1163-1205. Zangerl, R. 1969. The turtle shell, p. 311-339. In C. Gans, A.d'A. Bellairs, and T. S. Parsons [ed.], Biology of the reptilia. Vol. 1. Academic Press, New York.