ELSEVIER Review of Palaeobotany and Palynology 90 (1996) 223-247 REVIEW OF PALAEOBOTANY AND PALYNOLOGY Plant paleoecology and evolutionary inference: two examples from the Paleozoic William A. DiMichele \ Richard M. Bateman ''?= ' Department of Paleobiology, NMNH Smithsonian Institution, Washington, DC 20560, USA ' Royal Botanic Garden, 20A Inverleith Row, Edinburgh, EH3 SLR, UK ' Royal Museum of Scotland, Chambers Street, Edinburgh, EHl IJF, UK Received 5 September 1994; revised and acx?pted 20 December 1994 Abstract Paleobotany can contribute much to evolutionary scenario-building. Here, we use two case studies ? the Devono- Carboniferous vascular plant radiation and the largely coeval evolution of heterosporous from homosporous life histories ? to examine the interface between phylogeny and ecology. Our observations challenge some tenets of the neo-Darwinian orthodoxy, notably the assumed role of competition mediated selection as an active driving force, rather than a passive filter, of evolution. The Devono-Carboniferous class-level radiation of vascular plants was prompted by attainment of a complexity threshold and delimited the morphological envelope that enclosed an apparently fractal pattern of subsequent, lower level radiations. The contrast of low speciation rates with exceptionally high rates of phenotypic divergence in the Devonian suggests a non-adaptive "novelty" radiation, perhaps reflecting saltational evolution via "hopeful monsters". Successive lower level radiations were more constrained by the ecological hierarchy that resulted from progressive niche differentiation and saturation. This in turn reflected increased speciation rates, thereby completing a well defined negative feedback loop in the coevolution of phenotypic and ecological diflierentiation. Heterosporous Ufe histories evolved independently in at least ten lineages. Heterospory allows the sporophyte to impose, via diiferential development, a single fixed gender on each gametophyte prior to spore release. Although the resulting life history is less flexible than homospory, which on recent evidence includes a range of subtle and sophisticated strategies, it promotes the sporophyte as the primary target for selection. Gametophytes effectively perform the role of gametes and are released into the environment prior to fertilization, thus favoring aquatic- amphibious habitats resistant to occupation by homosporous pteridophytes; terrestrial heterospory requires apomixis. Although the profound iteration of heterospory implies a strong adaptive advantage, repeated gradual evolution via inferior intermediates exhibiting exosporous heterospory seems unlikely. Seed-plant success reflects economic efficiency and the subsequent evolution of effective poUination syndromes, rather than integumentation of the ovule. Major radiations of heterosporous Uneages and subsequently of seed- plants required pertiu-bation of pre-existing commxmities by extrinsic environmental changes rather than genuinely competitive displacement. This typical manifestation of "home-field advantage" further emphasizes the intimate relationship between phylogeny and ecology, and allows us to make predictions that can be tested by further paleobotanical research. 0034-6667/96/$! 5.00 ? 1996 Elsevier Science B.V. All rights reserved SSDI 0034-6667(95)00085-2 224 W.A. DiMichele, KM. Bateman/Review of Palaeobotany and Palynology 90 (1996) 223-247 1. Introduction 1.1. Neo-Darwinism in paleobiology Fossils of vascular land plants reveal a record of vast morphological change and enormous taxic diversification following humble beginnings in the Silurian. Historically, paleobotanists have applied and extrapolated the tenets of neo-Darwinian graduaUsm to explain the increase in plant struc- tural complexity. The focus has been on morpho- logical transformation, considered in isolation from the environment and often described in terms of individual organs rather than in a whole-plant context. Most theorizing has been limited to appli- cation of the axioms of gradual change to cases of presiuned ancestor-descendant relationship, often in progressive linear sequences of several taxa (e.g. Magdefrau, 1956). With a few notable exceptions (e.g. Stidd, 1980), the phylogenies themselves are usually determined intuitively from the very same expectations of gradual descent with modification. As a consequence, the literature is filled with nimierous examples of hypothetical intermediates, inferred "mechanisms" such as phyletic slide and recurvation, and, perhaps most seriously, the misuse of the population-level concept of selection pressure, treating it as a creative force rather than as a short-hand term describing the process of sorting existing alleles by natural selection. In fact, the plant fossil record has rarely been used to enlarge our understanding of evolution as a bona fide process. With the rise of cladistics and other explicit means of objectively assessing relationships (Stein et al., 1984; Stein, 1987; Crane, 1990; Bateman et al., 1992; Doyle and Donoghue, 1992; Gensel, 1992; Nixon et al., 1994) the potential circularity between phylogeny reconstruction and evolution- ary inference has been broken. In theory, modem phylogenetic methods permit a pattern to be deter- mined without specific mechanistic assumptions; explanatory hypotheses compete post hoc for the best phylogenetic fit. Admittedly, analysis of pattern rarely specifies a single evolutionary pro- cess; rather it constrains the range of possibilities by falsifying some potential processes (e.g. FrumhoflF and Reeve, 1994; Leroi et al., 1994). At the same time the breadth of morphological hypotheses has been enriched by modem concepts imported from developmental biology (Rothwell, 1987; Trivett and Rothwell, 1988; Mosbmgger, 1990; Stein, 1993; Bateman, 1994; Bateman and DiMichele, 1994a). Results include greater empha- sis on macroevolution (Wight, 1987; DiMichele et al, 1989; Bateman, 1994; Bateman and DiMichele, 1994b), challenges to the concept of competition as the all-pervasive driving mechanism in plant evolution (Scott and Galtier, 1985; DiMichele et al, 1987; Valentine et al., 1991), and the recognition that selection may act more often as a passive filter than as an active vector of evolution (Rothwell, 1987). These great strides have been made in a relatively few years, yet one major aspect of a complete picture remains poorly understood ecology (more specifically, the particular habitat preferences and limited ecological ampUtudes of each species). Commonly, ecology is assumed to influence evo- lution through competition (e.g. Knoll, 1986), though assertions about competition rely on the usually untested assumption that species share one or more common limiting resources. Of course, inferring the ecological preferences of long-extinct plants can be an obtuse exercise, but it is possible (e.g. Scott, 1978; Bateman, 1991; Phillips and DiMichele, 1992). Certainly, such inference is essential if paleobotany is to develop credible explanations for the evolution of plant form. Ecology is even more central in attempts to use fossil plant data to test or elaborate evolutionary theories. To use the classic analogy of Hutchinson (1965), organisms are the actors in a great evolu- tionary play, and ecology is the (ever-changing) stage. In this essay we adopt an unashamedly critical and polarized position on the issue of evolutionary dynamics, stressing broad macroevolutionary pat- terns over examples of intra-generic speciation. Our objective is not to deny gradual microevolu- tion. Rather, we wish to emphasize the role of ecology in dictating new and sometimes altemative explanations for significant events in plant evolu- tion. Sadly, paleobotany barely participated in the great punctuated equilibrium debate of the 1970s and 1980s. Now that ecology has come of age in W.A. DiMichele, R.M. Bateman/Review of Palaeobotany andPalynology 90 (1996) 223-247 225 paleontology, paleobotanists again are presented with an opportunity to integrate the plant fossil record with basic studies of evolutionary processes. This can be done only by escaping the constraints of theoretical orthodoxy, notably the uncritical acceptance of insensibly gradual, competitively driven evolutionary change without serious con- sideration of the important ecological component. 1.2. Scope and objectives The goal of this essay is to examine the role of paleoecology in constructing evolutionary explana- tions for taxic-phylogenetic, morphological, and biogeographic patterns. We will advocate greater awareness of the vast amount of accessible ecologi- cal data and incorporation of "ecological think- ing" into paleobotanical scenario-building. Typical use of the term "ecology" is undesirably general; it can refer to the autecologies of mdivi- dual species, the central environmental tendencies of an entire clade of organisms, or the dynamic interactions of species at the conmiunity, land- scape, or biome levels. Excluded, however, are speculations about environmental "selection pressures" as creative forces. Ecological con- siderations, when inferred independently of the organisms in question, can constrain evolutionary scenarios in many ways, some obvious but others more cryptic. Our chosen examples focus on two areas. First, in reconstructing the controls on major morphological transformations, environmental information may modify the interpretation of the timing and causes of morphological differences between presumed ancestor and descendant (s). The evolution of heterosporous plants from homo- sporous ancestors and the ultimate evolution of seed plants serve to test the idea of a progression of biologically superior (or better adapted) life histories. Second, the fossil record contains important examples of major evolutionary radiations in which both the stem lineages and the limits of a morphological envelope are established early, fol- lowed by relatively routine filling of the morpho- logical space via less profound speciation events. The Devonian-Mississippian radiation of vascular plants, arguably the most profound evolutionary diversification in land-plant history, demonstrates that radiations can be self-limiting, constrained by the interaction of ecospace and morphological complexity. 2. Case study I: Ecological constraints on adaptive scenarios ? evolution of heterospory and the seed habit 2.1. Background: the nature of heterospory With the exception of the tree ferns, nearly all numerically abundant and taxonomically diverse Imeages of trees have been heterosporous. This observation demonstrates irrefutably that hetero- thallic heterospory, in which male and female sex organs are produced on separate gametophytes derived from small and large sized spores, respec- tively (for detailed terminology see Bateman and DiMichele, 1994b), confers significant ecological advantages. Phylogenetic patterns suggest that het- erospory evolved independently at least ten times among vascular plants (Bateman and DiMichele, 1994b; Bateman, 1995), including the zostero- phylls, selaginellalean plus rhizomorphic lycopsids, equisetaleans, several filicalean ferns, and several progymnosperms plus their descendants, the seed plants (Fig. 1). Thus, there was no single case of directional selection generatmg a highly adapted heterosporous Uneage from which all subsequent heterosporous lineages diverged. The strongly iter- ative nature of the life history lends credence to the view that it is an inevitable outgrowth of homosporous ancestors and that it confers a sig- nificant adaptive advantage (e.g. Chaloner and Pettitt, 1987). Inspection of the fossil record leads to three generally accepted conclusions: heterospory is an evolutionarily successful life history, free-sporing heterospory is clearly derived relative to homo- spory (as is the seed habit relative to heterospory, however these terms are defined), and the homo- spory-heterospory-seed habit transition sequence represents a progressive increase in reproductive sophistication (Table 1). The evolution of the seed has been the focal point of a series of intermittently 226 W.A. DiMichele, R.M. Bateman/Review of Palaeobotany and Palynology 90 (1996) 223-247 Seed-Plants Ma B.P. 300 320 - 340 _ 360 380 - = 400 - 420 1 0. 5 CO o LU STEP WEST NAMU LL z o CO < o VISE TOUR ^ STRU FAME FRAS z GIVE EIRE lU Q EMSI SIEG GbUI PRID LUDL CO WENL GYMNOSPERMOPSIDA TRACHEOPHYTA s.l. COOKSONIOPSIDA ?jfi Questionable identity of ancestral class ^ Questionable monophyiy of descendant class Fig. 1. Tentative non-numerical phylogeny of all tracheophyte classes, showing putative ancestor-descendant relationships plus divergence dates, and occurrence of heterospory. Note that (1) the entire post-cooksonioid radiation occurred within the Devonian, and (2) a minimum of ten independent origins of heterospory is indicated by the stippling, taking into accoimt recent assertions of monophyiy for the Salviniales plus Marsileales based on morphological (RothweU and Stockey, 1994) and molecular (e.g. Hasebe et al., 1994) data. (/) Some Barinophytales; (2) all Sellaginellales plus all rhizomorph-bearing lycopsids (Isoetales s.l.); (5) some Equisetales; (4) some Spbenophyllales (doubtful); (J) some Stauropteridales; (6) all Hydropteridales (Salviniales plus Marsileales); (7) some Filicales (e.g. Platyzoma); (5) some Aneuophytales; (P) some Archaeopteridales, ?all Protopityales, all Cecropsidales; (70) some Neoggerathiales. (Modified after Bateman and DiMichele, 1994b, fig. 11.) contentious debates (Thomson, 1934; Andrews, 1963; Pettitt and Beck, 1968; Pettitt, 1970; Steeves, 1983; Chaloner and Pettitt, 1987; DiMichele et al., 1989; Chaloner and Hemsley, 1991). Recently, the increasing rigor of phylogenetic reconstructions has sharpened the debate on ancestry (Meyen, 1984; Beck, 1985; Stein and Beck, 1987; Donoghue, 1989; Donoghue and Doyle, 1989; Galtier and Rowe, 1989; Crane, 1990; Doyle and Donoghue, 1992; Bateman and DiMichele, 1994b; RothweU and Serbet, 1994). Heterospory has received less conceptual attention (Willson, 1981; Haig and Westoby, 1988b; DiMichele et al., 1989; Bateman and DiMichele, 1994b), and most paleobotanical scenarios have treated it merely as an intermediate stage in the evolution of the seed. Nearly all studies of the evolution of heterospory have focused on structure alone. Little attention has been paid to the ecological preferences of the taxa or life histories in question, or even to the JV.A. DiMichele, R.M. Bateman/Review of Palaeobotony andPalynology 90 (]996) 223-247 227 breadth of variation in the ancestral, supposedly primitive homosporous condition. This problem has been compounded by an insistent desire to place all known morphological fonns (or classes of forms) into continuous progressions, despite the lack of formal phylogenetic support for such schemes (Andrews, 1963; Tiffney, 1981; Chaloner and Pettitt, 1987; Chaloner and Hemsley, 1991). These scenarios are rooted in an acceptance of gradual transformation of form, driven by "selec- tive pressures" for functionally superior morpholo- gies. The foremost assiunption of this form of gradualism is competition. Successive, increasingly complex morphologies must be competitively supe- rior to the ancestral forms, and increase in com- plexity must be associated with increase in fitness. However, for competition to favor the rise of the advanced architectures at the expense of the primitive ancestral forms, there must be Httle or no ecological differentiation between ancestor and descendant. Unfortunately, the details needed to fit this view to even the most rudimentary evolu- tionary models are rarely given. Are such trans- formations allopatric, or is mass selection across entire species involved? What are the environmen- tal conditions imder which the transformation occurred, and what kinds of advantages are con- ferred by the derived morphologies under those circumstances? 2.2. Ecological constraints and the homospory- heterospory transition Heterosporous plants evolved from ancestors that possessed a remarkably flexible and ad- vanced reproductive strategy, namely homospory (Table 1). A survey of recent literature on extant pteridophytes (Willson, 1981; Haig and Westoby, 1988a; Shefiield, 1994; Bateman and DiMichele, 1994b) indicates a wide range of variation within the basic homosporous pattern of alternation of free-living generations. For example, in many fern species the gametophytes rely on chemical signals (antheridiogens) to mediate population-level sex ratios and control the timing of gametogenesis and syngamy, and in most species at least some gameto- phytes retain their potential for self-fertilization. Thus, the gametophyte generation is closely attuned to local environmental conditions and can effectively regulate its reproductive functions. The main disadvantage of homospory seems to be the need for ecological coordination between the gametophyte and sporophyte generations. In par- ticular, the sporophyte must grow at or near the point of syngamy, a location dictated by the gametophyte. Thus, evolution away from the norm of mesic, moist environments is unusual. The evo- lution of aquatic-amphibious habit on the part of both sporophyte and gametophyte seems to have been particularly demanding once a fully terrestria- lized, homosporous life history was estabUshed. "Return" to the water requires niunerous changes in morphology and physiology of both sporophyte and gametophyte. However, as free-Uving, inde- pendent Ufe history phases, regulation of develop- ment and phenotype, and the selective factors acting on that phenotype are partially (largely?) independent, even though the two phases share a common genome. Increased terrestrialization, in contrast, is limited mainly by syngamy, which can be accommodated in many ways, including through apomixis. Overall, therefore, homospory is an effective life history in terra firma habitats. Limits on homosporous ferns, lycopsids, and sphenopsids in modem habitats may reflect the limitations of vegetative architecture and "house- Table 1 Characteristics of four categories of vascular plant life histories Gametophyte Dominant phase Sex ratio (?/? control: timing of gametogenesis Functional dependence of tJ/$ on ?; Homospory Anisospory Heterospory Seed habit Free living Free living Endosporic Endosporic Codominance Codominance Sporophyte Sporophyte Labile Fixed Fixed Fixed Labile, (J/$ controlled Labile, (J/? controlled Mainly ? controlled AVholly ? controlled Independent Independent Independent Dependent 228 W.A. DiMichele, R.M. BatenumIReview of Palaeobotmy and Palynology 90 (1996) 223-247 keeping" physiology more than reproductive biol- ogy. The common view of homospory as an inferior life history waiting to be displaced is difficult to justify, particularly in mesic habitats. In contrast with homospory, the gametophytes of heterosporous plants are strictly unisexual (Fig. 2). Sex ratios are determined by the sporo- phyte epigenetically (developmentally rather than through sex chromosomes) during sporangial mat- uration. In effect, microspores and megaspores function as gametes. The way gametophytes respond to environmental variabiUty, including conditions unfavorable for gametogenesis, is to enter diapause (suspended development) inunedi- ately after release and before "germination" (note that this ability was probably available in their homosporous ancestors - spores of most extant ferns accumulate in the soil as spore banks; Dyer and Lindsay, 1992). This is a remarkably inflexible life-history strategy, given that spores are released directly into the environment without the potential to form a photosynthetic, free-living, indepen- TERRA FIRMA HABITATS HOMOSPORY "^and^ I have similar environmenta requirements but independent development limited ^ flexibility requires apomictic reproduction -^functions as unitary organism that adapts to environment AQUATIC ?AMPHIBIOUS HABITATS ?? -^and^ must independently acquire traits that confer aquatic function HETEROSPORY ?^ must acquire traits that confer aquatic function pollination SEED HABIT delivery of O directly toy circumvents environmental limitations to syngamy Fig. 2. Constraints on life history evolution in terra firma and aquatic-amphibious habitats, as the sporophyte gradually co-opts the gender control previously exercised by the gametophyte. Note especially the ineffectiveness of (1) homospory in aquatic- amphibious environments and (2) heterospory in terra firma habitats without apomixis or reliable pollination. Dotted lines indicate transitions that have been rare in plant history due to ecological constraints. Solid lines suggest the morphological-ecological history leading from homosporous plants to seed plants. W.A. DiMichele, RM. Bateman/Review of Palaeobotany andPalynology 90 (1996) 223-247 229 dently self-supporting individual. How then did free-sporing heterosporous plants become success- ful ecologically? Current evidence suggests that heterosporous plants have been ecologically domi- nant only in environments with ample free mois- ture, notably aquatic and amphibious habitats. In such environments spores-as-gametes are most likely to encounter conditions favorable for fulfill- ing their function, though even here the extant Salviniaceae and Marsileaceae possess sophisti- cated sporocarps that allow a resting phase following sporogenesis. Admittedly, some species of extant heterosporous plants grow in mesic to xeric habitats, but in these environments apogamy predominates (DiMichele et al., 1989; Bateman and DiMichele, 1994b). What advantage is conferred by a free-sporing heterosporous mode of reproduction? The most obvious characteristic is unity of form. The sporo- phyte is the dominant phase of the life cycle and becomes the focus of selection on vegetative func- tion and environmental tolerance (Table 1, Fig. 2). However, because the gametophytes and gametes must be released into the environment, the gameto- phyte generation continues to constrain the condi- tions under which the system can function effectively. In gaining the advantages of a com- pressed life cycle, the potential to colonize terra firma habitats is severely curtailed by the sporo- phytically fixed sex ratio, the inability of gameto- phytes to adjust to most environmental vagaries, and the strict requirement for moisture during spore germination and syngamy. Although many reviewers have appealed to an "intermediate" evolutionary stage between homo- spory and heterospory (of the heterothallic, endo- sporic variety), we regard heterospory with free- living gametophytes as one of the great paleobo- tanical fantasies. These are imagined intermediate forms between homosporous ancestors and each of the independent origins of heterospory. In most interpretations of neo-Darwinism, first principles dictate gradual transformation of form and hence an insensible series of intermediates. A supposed vindication of that expectation was the discovery of a rare filicalean fern Platyzoma (Tryon, 1964) that possesses such a life history. Although Sussex (1966) soon warned against treating Platyzoma as anything more than an evolutionary curiosity, it nonetheless has risen to cult status. However, survey of the modem flora reveals no other exam- ples of this life history. Examination of the ecologi- cal implications of the platyzomoid life history indicates that heterospory with free-Uving gameto- phytes bears all the disadvantages of both homo- spory and heterospory with the advantages of neither: the ecological-evolutionary tension between the sporophyte and gametophyte remains, and the gametophytes are xmable to adjust the sex ratio and timing of reproduction in response to environmental cues. If other such "intermediates" existed they are Ukely to have occurred as small, ephemeral populations and, therefore, on statisti- cal principles alone were unlikely to survive for long periods. Allopatric isolates may be common products of the dynamics of the evolutionary pro- cess, but few appear to survive to establish long- lived clades; the ecological handicaps of the platyzomoid life history thus combine with the chance elements of evolution to make it doubly unlikely that such derived lineages would foimd new clades. Homospory in modem lower vascular plants encompasses a complex array of functional mor- phologies, some quite sophisticated in their control of sexual reproduction. Most of the biochemical adaptations found in extant fems could not be recognized in the fossil record. However, there is no reason to caricature Devonian homosporous plants. By today's fiUcalean standards, tie life history represented by Platyzoma is functionally inferior to the complex life histories typical of many homosporous species. If complex homospo- rous Ufe cycles evolved during the Devonian, free- living heterospory is an unlikely intermediate form in the evolution of heterospory, simply on ecologi- cal grounds. 2.3. An ecological perspective Heterospory is ecologically successful in a different set of environments than homospory. The heterosporous life history is advantageous in aquatic and amphibious habitats where gameto- phytes are assured of favorable conditions; hetero- sporous plants in drier habitats generally are 230 W.A. DiMichele, R.M. BatemanIReview of Palaeobotany andPalynology 90 (1996) 223-247 apomictic (Fig. 2). In contrast, homospory is rare in aquatic habitats (the semi-aquatic fern Ceratopteris is an interesting exception that can complete its life cycle rapidly under aquatic condi- tions; Tryon and Tryon, 1982; Eberle et al, 1994). The extended gametophytic life-span of almost all homosporous plants permits a wider range of response in terra firma habitats. Nevertheless, the great differences in sporophytic and gametophytic form, and the continued requirement for surface moisture to effect fertilization, restrict the abiUty of homosporous species to radiate into aquatic habitats. As with heterosporous lineages, homo- sporous plants have been able to expand to the drier end of the environmental spectrum. The transition to the aquatic-amphibious habi- tat is especially demanding. The need to modify evolutionarily both the sporophyte and gameto- phyte appears to have been a major factor that prevented homosporous plants from exploiting such environments. Thus, endospory is the single most important breakthrough that permits the invasion of aquatic habitats (Bateman and DiMichele, 1994b). Endospory leaves the sporo- phyte to face the vagaries of the environment. Only during reproduction must the gametophytes be released, and then only for a brief period. But it is this reproductive phase, despite the compres- sion of the developmental stages between meiosis and syngamy, that restricts the choice of habitats, just as it opens the aquatic habitat to invasion. We see no need to advance intermediate forms between homospory and endosporic heterospory, and have argued elsewhere (DiMichele et al., 1989; Bateman and DiMichele, 1994a,b) that the mor- phological transformation is developmentally con- trolled and thus can yield a "hopeful monster" in a single generation (see also Van Steenis, 1969; Arthur, 1984; Levin, 1993). Ecologically, early heterosporous forms probably entered and exploited an environment virtually unoccupied by other vascular plants. They were capable of passing through a selective filter almost impenetrable to homosporous plants. Consequently, there is no reason to invoke competitive superiority to explain the evolution of heterospory from homospory. Rather, it can be viewed as a happenstance that permitted escape from competition (or, in other terms, exploitation of a new resource). 2.4. Evolution of the seed The earliest known seeds occur in Late Devonian swamp habitats (Gillespie et al., 1981), providing evidence that early seed plants preferentially occu- pied the same kinds of environments that were exploited by their heterosporous ancestors. However, the great diversity of early seed plants and seed morphologies occurs in terra firma habitats (Andrews, 1963; Long, 1975; Scott, 1980; Retallack and Dilcher, 1988; Rothwell and Scheckler, 1988), suggesting that seed plants radi- ated primarily in drier habitats. As with the homospory-heterospory transition, competitive superiority has often been invoked in the hetero- spory-seed habit transition. In both cases, rudi- mentary ecological analysis offers little support to the a priori hypothesis of competition. Perhaps the most serious issue in considering the evolution of seed habit is its definition. Almost universally, seed habit is defined as a heterosporous condition in which the number of functional mega- spores per megasporangium has been reduced to one. The one viable spore develops completely within the megasporangium, which is surrounded by additional tissue that is generally considered to be of sporophytic origin. As a consequence of this morphology, the male gametophyte, also endo- sporic in development, must be deUvered to the seed, a process termed pollination. Most studies of seed evolution have focused on the elaboration of the integimients (Andrews, 1963), on the rela- tive efiiciency with which various early forms cap- ture pollen (Niklas, 1981), or on likely ancestry (Beck, 1985; Crane, 1990; Rothwell and Serbet, 1994). The evolution of the seed habit itself has been addressed only tangentially, via conceptual gradualist arguments in which the seed is assem- bled step-wise through a now familiar series of morphological intermediates. Such scenarios con- stitute an evolutionary argument only if gradual- ism is again considered axiomatic, and if the final configuration of the seed is regarded as an optimal, stable morphology that is more likely to resist further evolutionary modifications than any of the fV.A. DiMichele, KM. BatemanjReview of Palaeobotany andPalynohgy 90 (19%) 223-247 231 so-called "intermediate" forms. Once more, how- ever, most of the critical intermediates are missing from the fossil record; either they have not yet been detected or they never existed. From an ecological perspective only one "step" in this transformation is critical to the life history we view as the seed habit, namely delivery of the male gametophyte directly to the female. Thus, pollination is the key innovation, less from a morphological perspective than an ecological one. Plants with routine pollination have come as close to unifying the sporophyte and gametophyte into a single individual as can be achieved by sessile organisms (Bateman and DiMichele, 1994a,b). The gametophytes no longer place environmental restrictions on the sporophyte, which becomes limited largely by vegetative environmental toler- ances rather than constraints on reproduction. It would be impossible to distinguish this break- through in the fossil record, due to the diflBculty of distinguishing free-sporing "heterosporous" plants from structurally similar plants experiencing routine pollination. It is Hkely that the first plants with pollination evolved within the aquatic-amphibious habitats occupied by their heterosporous predecessors (Fig. 2). The fossil record documents that seed plants did not achieve dominance in these environ- ments until late in the Paleozoic. Rather, the most rapid increase in seed-plant diversity in terra firma habitats occurred through the Late Devonian and Early Carboniferous, where it rapidly became the dominant life history. Heterosporous plants con- tinued to dominate wetland habitats until removed by extrinsically induced extinction (Phillips and Peppers, 1984; DiMichele and Aronson, 1992). Enviroimiental place holding on ecological time- scales was termed "home-field advantage" by Pimm (1991). It seems that the pattern can be extrapolated to geological time intervals, especially at the level of higher taxa. As we will discuss in our second case study, the seed plant radiation may have been suppressed in aquatic-amphibious habitats by the already well established occupancy of earlier heterosporous plants (Knoll and Niklas, 1987). Certainly, the earliest seed plants probably had similar vegetative tolerances to their hetero- sporous precursors (Bateman and DiMichele, 1994b). It was their ability to reproduce under a wide range of conditions that enabled them to exploit parts of terra firma enviroimients exploited by few homosporous plants. In addition, the bifa- cial cambiimi of seed plants, permitting massive wood production while retaining effective photo- synthate transport, may have allowed them to partition terra firma habitats in ways not pre- viously possible. Their reproductive methods allowed them to pass the filter between aquatic- amphibious and terra firma settings, but it was primarily vegetative specialization ("economic factors" sensu Eldredge, 1989) that facilitated their exploitation of drier habitats. When ecological factors are considered it appears that seed plants may have come into competition with heterosporous plants-and lost! The first seed plants grew in enviroimients aheady heavily saturated with heterosporous species (par- ticularly lycopsids and progymnosperms) as place holders. Although some species may have become estabUshed, the fossil record demonstrates that seed plant species failed to dominate aquatic- amphibious environments until the latest Paleozoic. In contrast to their origin, the early radiation of seed plants may have occurred under conditions of relatively Uttle competitively medi- ated selection (note that this also apphes to several earlier radiations of heterosporous species). Pollination was a "pre-adaptation" that permitted a major switch in the ecological amphtude of an entire clade, a bona fide key innovation (a morpho- logical or physiological innovation that permits a clade to switch resources and escape competitive constraints). It was not until abiotically driven extinctions eliminated heterosporous tree forms that seed plants were able to colonize aquatic- amphibious habitats. It could be argued that the radiation of several distinct heterosporous clades in wetlands during the Paleozoic (Fig. 1) contradicts the argument that home-field advantage should dissuade multiple radiations of similar life history stategies in similar habitats (B.H. Tiffney, pers. commun., 1994). However, the radiations in aquatic-amphib- ious environments are an edaphic complex of habitats, and thus offer a number of niches for exploitation; they effectively constitute a biome. 232 W.A. DiMichele, R.M. Bateman/Review of Palaeobotany and Palynology 90 (1996) 223-247 Freshwater aquatic-amphibious habitats also are isolated relative to most drier terrestrial habitats, offering more opportunity for local estabUshment of derived clades. Although several, mainly lower- vascular plant groups radiated within wetlands during the Paleozoic, most of these radiations produced limited numbers of taxa. The seed plants were no exception to this rule; the radiation of this group in aquatic-amphibious settings pro- duced few taxa. It was not until they entered drier habitats that a major radiation took place. The aquatic-amphibious environment of the Paleozoic offered limited opportunity for radiation due both to physical restriction and the presence of numer- ous occupants. Unlike the heterospory-homospory transition, no catastrophic structural changes are required in the heterospory-seed habit transition. Rather, the structural modifications that are so well docu- mented in the fossil record are mere postscripts to seed plant origins. Certainly, integumentation affects pollination efficiency and enhances gameto- phyte protection, and various micropylar and cupule modifications were instrumental in the radi- ation of seed plants. But an understanding of the patterns of change through time in these accessory structures does not lead to a greater understanding of the origin of this subsequently dominant life history. 3. Case study 11: The radiation of vascular land plants and the emergence of the ecological hierarchy 3.1. Major land-plant radiations By the end of the Early Carboniferous, ter- restrial vascular plants had undergone two of the three major radiations evident in the fossil record (Fig. 1). The first occurred during the Siluro- Devonian and gave rise to the cooksonioids, zost- erophylls, lycopsids, rhyniophytes, and trimero- phytes (Banks, 1975; Grensel and Andrews, 1984; Knoll et al., 1984; Niklas et al., 1985). The second spanned the Late Devonian-Early Carboniferous and generated the selaginellalean and rhizomor- phic lycopsids, sphenopsids, progymnosperms, seed plants, and several "fern" groups, including the "zygopterid" and filicalean lineages (Chaloner and Sheerin, 1979; Scott, 1980; Stein et al., 1984; Galtier and Scott, 1985; Meyen, 1987; Crane, 1990). Within-clade radiations lagged behind by about one geological stage but conformed to the basic architectural themes established in the earlier major diversification (but note that this lag may be explained partly by sampling of species in the fossil record: on average, a class will be more readily detected than an order; D.H. Erwin, pers. commun., 1984). The radiation of the angiosperms is the final (and ongoing) diversification of vascular plants; it has resulted in several architectural forms not represented in the earlier radiations, such as monocotyledonous vegetative organization and the herbaceous habit, though we regard these as rela- tively minor modifications of the earlier seed-plant bauplan. Figs. 1 and 3 show that the Devono-Carboni- ferous radiation established what traditional Linnean taxonomy has regarded as classes of vascular plants: Lycopsida, Sphenopsida, Pteropsida, Progymnospermopsida, Spermato- psida (we here follow the taxonomic system of ICnoU and Rothwell, 1981). The phylogenetic studies of Crane and colleagues (Crane, 1990; Kenrick and Crane, 1991) have provided informal names that capture more accurately the phyloge- netic structure of the relationships, and conform broadly to the traditional Linnean taxa. In effect, Linnean higher taxa are short-hand for the distinc- tive vegetative and reproductive architectural char- acteristics that separate the major clades. The degree of phylogenetic resolution is highly variable both within and between major clades. The pattern attending the origin of the sphenopsids is the best studied. Stein et al. (1984) showed that the sphenopsids originated from an as yet incom- pletely resolved pool of potentially ancestral forms. Although the broad patterns of descent are clear for groups other than the sphenopsids, transitional morphologies have not been found that clearly link specific sublineages. In particular, ancestral homosporous lycopsids or trimerophytes are not well connected to the derived clades, such as the rhizomorphic lycopsids, the various fern lineages, or the progymnosperms. Relationships between the wholly extinct progymnosperms and the puta- W.A. DiMichele, KM. Bateman/ReviewofPalaeobotanyandPalynology90 (1996) 223-247 233 tively descendant seed plants remain contentious. The many morphological and phylogenetic studies have all been handicapped by the paucity of fully reconstructed progymnosperms. As indicated by the sphenopsids, the Late Devonian was a time of wide-ranging morphological innovation. The "morphological envelope" was largely circum- scribed during this time; it was filled during subse- quent lower level radiations, which played on the major structural themes that had already evolved (Bateman, 1991). This potentially fractal taxo- nomic pattern is evident in the early class level radiation and subsequent order level radiations summarized in Fig. 3, though a more rigorous analysis extended to lower taxonomic levels is desirable (cf. Erwin et al., 1987). Surprisingly little ecological attention has been devoted to this fundamental class-level radiation. Most discussion has centered on the evolution of particular morphological features, such as the tree habit, vascular morphology, or seeds. This focus overlooks their more profound context as part of the evolution of nearly all major architectural plans in vascular plants, an event that occurred during a remarkably short period relative to the total history of the tracheophytes. The literature on this interval and its evolutionary history is broad; key papers are summarized below. Many taxonomically restricted features did not occur randomly across physical habitats; rather, the radi- ation was characterized by strong clade-environ- ment interactions. Chaloner and Sheerin (1979) documented the preceding accumulation of mor- phological innovations diuing the Devonian, and Banks (1980) provided a biostratigraphic context for the major taxa. Knoll et al. (1984) described patterns of diversification during the Devonian and, most significantly, quantified the degree of morphological differentiation among the major groups. Knoll and Niklas (1987) argued for a quantitative adaptationist approach to evaluating evolutionary scenarios that encompassed biomech- anical attributes of organisms but lacked a detailed paleoecological component. More explicitly eco- logical studies, focusing mainly on floristics, include those of Scheckler (1986a,b), Bateman (1991), and Scott and coworkers (for summary see Scott, 1990). Retallack and Dilcher (1988) examined the ecological implications of inferred Early Carboniferous seed-plant biologies, and Raymond (1987) evaluated the paleobiogeo- graphic distribution of major morphological traits. Several studies used a range of approaches to evaluate phylogenetic relationships (e.g. Chaloner and Sheerin, 1979; Knoll and Rothwell, 1981; Meyen, 1984; Stein et al., 1984; Crane, 1988). It is from the totality of approaches that a picture of the radiation emerges. S/DDLCUCP'R JKTB Fig. 3. Timing of origination of class and order level taxa indicated by the fossil record. Ordinal level originations peak later and are far more extended. Classes are Cooksoniopsida, Rhyniopsida, Trimerophytopsida, Zosterophyllopsida, Lycops- ida, Progymnospermopsida, Cladoxylopsida, Sperma- topsida, Sphenopsida, Pteropsida (see Fig. 1). Orders from Knoll and Rothwell (1981); angiosperms are treated as an order. 3.2. Questions of scale We will briefly review the literature relevant to the ecological implications of the Devono- Carboniferous radiation. Invertebrate and verte- brate paleontology have developed many concepts that are broadly appUcable to interpretation of patterns in the plant fossil record. Suggestions that plant and animal evolution were guided by different suites of mechanisms should be scruti- nized in light of existing theory. The basic shapes of basal radiations, the nature of morphological discontinuity among lineages, and the basic tenets of speciation are the same for plants and animals. The literature review that follows is intended to emphasize existing explanatory models that apply as well to plants as to the animals on which they were formulated. Crucial questions about the class- 234 IV.A. DiMichele, KM. BatemanIReview of Palaeobotany andPalynology 90 (1996) 223-247 level vascular plant radiation are, what caused it to begin and what caused it to end? We will argue that upward causation is the key, especially the degree to which processes operating at the scale of local populations within communities can account for the phylogenetic shape and ecological consequences of a major phenotypic radiation. It appears likely that the ecological spatio- temporal hierarchy with which we are famiUar (community > landscape >biome) did not exist in the Late Devonian, but evolved in concert with the morphological radiation. As it developed, the hierarchy began to constrain (or perhaps contain) the spatial and temporal ranges over which popula- tion-level processes could act, and thus increasingly channelled the vascular plant radiation. Valentine (1980) suggested that during the early Paleozoic radiation in marine ecosystems specia- tion was limited by the availabiUty and size of "empty" niche space. Organisms exhibiting minor functional modifications needed only small low- competition ecospaces in which to estabhsh themselves, whereas carriers of major structural or functional deviations needed larger voliunes of space because they were likely to have very low fitness and little competitive ability. Valentine's model predicted that ecospace existed indepen- dently of the occupying organisms. It was most widely available at the onset of a radiation and rapidly filled as it became colonized via major phenotypic jumps (Fig. 4). Subsequent speciation tended to deviate progressively less from the ini- tially estabUshed architectural themes because there were no low-competition ecological voids that could accommodate major phenotypic change. This type of radiation was classified by Erwin (1992) as a "novelty" event: one that takes place under low selection and involves organisms with weak genetic and architectural constraints. The novelty radiation is followed by more typically "adaptive" radiations within the limits of the ecological space occupied by the founding lin- eage(s). Erwin (1992) noted that in the fossil record of marine invertebrates events of this type tended to occur in the Paleozoic, presumably because empty ecological space was more common then, and because genomic and developmental ^^n '^^^^^^H' ^^^^^^H" H ^^^^^B ^^^H ? ^^H J^H l~~^ ^^?^^VT^H~ B r^l r^^l ^^l_ r1 B-1-?t==!==J 1^ ? ? = = = = !? = =: = ? )_ s| > i 'si r -+- -r-j- T 3ui Si l Fig. 4. Hypothetical pattern of rapid occupation of the available ecospace following class-level originations in the Siluro-Devonian. Squares represent "adaptive" space (tesserae in the model of Valentine, 1980). Founders of a major clade require more adaptive space than subsequent speciation events that are smaller scale modifications of that same basic theme. Space filling combines with progressive canalization of development and architectural constraint to limit both the opportunity for survival of highly divergent forms and the likelihood of their occurrence. W.A. DiMichele, R.M. Bateman/Review of Palaeobotany andPalynology 90 (1996) 223-247 235 constraints were less entrenched, permitting more phenotypic variation. In Valentine's scenario there is a progressive interaction between morphological change (the engine of which is not of immediate concern; it is assimied that a spectrum of variation, from tiny differences to large discontinuities, is present at all times) and ecological space, even though the ecological space is regarded as largely predefined. Fundamentally, the ecological factors of standard neo-Darwinian evolution (principally, the chance location of a suitably low competition site by a population bearing a derived phenotype, generally via an allopatric event) are those con- trolling speciation, irrespective of the degree of phenotypic deviation from the ancestor. Significant parts of Stanley's (1975, 1979, 1990) "species selection" paradigm also focus on the interaction of extinction and origination rates in evolving lineages and thus treat ecological factors as at least partially dependent on biotic factors. Stanley demonstrated that high rates of speciation (as opposed to origination, which was assimied to be stochastically constant) within clades must be accompanied by high rates of extinction, and that large population sizes tend to reduce the likelihood of speciation. Both factors reflect the inter- actions of speciation and adaptive (niche) space. Origination may provide a wide range of variation for selection but there also must be sufficient adaptive space to allow the establishment of small derived populations. Without extinction, pheno- typically derived forms tend to encounter ecologi- cally similar ancestors; generaUsts with large population sizes are on average more resistant to extinction, so these groups tend to occupy extens- ive ecospace for relatively long periods and have relatively low speciation rates. Damuth (1985) argued for a more clearly pop- ulation-centered, and hence more ecologically cen- tered, formulation of species selection; selection among species takes place through the mteractions of "avatars": local community-specific populations of larger species ("ecotopodemes" sensu Gilmour and Heslop-Harrison, 1954). In Damuth's formu- lation avatars are to species as individuals are to populations. They provide the basis for an ecologi- cal analysis of selection among species, a model broadly consistent with that of Stanley but more explicit with regard to the all-important level in the organismal hierarchy at which constraints operate. The interaction of ecological and evolutionary dynamics on long time scales also has been addressed by other workers. As detailed by Vermeij (1987), the concept of escalation explicitly requires environmentally forced ecological disruption to break up co-evolved species complexes. Such complexes effectively exert home-field advantage and exclude "invaders" from access to resources. Extinctions open up resources and permit compar- atively brief bursts of competitively driven evolu- tion. Brett and Baird (1995) provided perhaps the most explicit link between landscape- and biome- level changes in marine faunas and synchronous, macroevolutionary changes in many of the compo- nent lineages. They argued that early Paleozoic faunas persisted taxonomically and structurally for several million years. Periods of persistence were terminated by brief intervals of rapid turnover, accompanied by the emergence of new ecological structures. Major evolutionary divergences were concentrated in several Uneages during these intervals of ecological disequilibriiun; these were "ecologically induced" radiations sensu Erwin (1992). Simpson (1944) viewed major phenotypic changes as rapid shifts from one adaptive peak to another (this is consistent with the pattern outlined in Fig. 4). Calling such changes "quantum evolu- tion", Simpson visualized a causal combination of happenstance attainment of a threshold level of "pre-adaptation" coinciding with unusually great ecological opportunity. Simpson (1953) later brought these concepts more in line with standard neo-Darwinian models of allopatric speciation, which implicitly require an ecologically uniformi- tarian viewpoint. In contrast to models in which ecology is implic- itly or explicitly central, most macroevolutionary models have focused on the origin of variation, or on the "internal" morphological constraints to phenotypic change, in which ecology plays a role only as a final, passive selective filter (sensu Rothwell, 1987). Several authors (e.g. Bateson, 1894; Goldschmidt, 1940; Schindewolf, 1950; and later Eldredge and Gould, 1972; Gould and Eldredge, 1977; Gould and Lewontin, 1979; Stidd, 236 W.A. DiMichek, R.M. Bateman/Review of Palaeobotany and Palynology 90 (1996) 223-247 1980; Arthur, 1984, 1987; Levinton, 1988; Gould, 1991; Stebbins, 1992) looked for developmental factors to explain the alternation of short-term phenotypic change and long-term phenotypic stasis. Such models view evolution within lineages as constrained principally by "historical" events, i.e. architectures created in the past that persist into the present (Gould and Lewontin, 1979; Gould and Vrba, 1982). Early, morphologically wide-ranging radiations are argued to be a con- sequence of poorly canalized developmental path- ways and limited interactions among the architec- tural components of organisms. The fixation of Uneages and architectures early in a radiation is thus strictly a consequence of ever-mounting constructional "burden" (sensu Riedl, 1979) and ever more complex developmental interactions (Gould, 1977). Erwin and coworkers (Erwin and Valentine, 1984; Erwin et al, 1987; Erwin, 1993) argued that Valentine's model needed to be modified by includ- ing such morphological and genetic constraints as those considered in more traditional formulations of macroevolution. Such constraints would reduce the likelihood of major phenotypic changes through time as developmental and genomic con- tingencies evolved ("canalization" sensu Wadding- ton, 1942; see also Arthur, 1984). The Devono-Carboniferous origin of major vas- cular-plant architectures appears to be an Erwin- style "novelty" radiation. Early in such a radiation, highly derived forms may be able to survive even if they have ecological tolerances vastly different from their ancestors (equivalent to biome-scale differences). High extinction rates (Stanley, 1990) are not required at this stage because of low utilization of the full spectrum of resources. As resources become increasingly saturated, derived phenotypes begin to encounter "occupied" adap- tive space more frequently, thus selecting against the more divergent forms. Smaller-scale evolution- ary events, which were happening all along, begin to fill in the broad types of ecological space (e.g. "tropical wetlands", "regularly disturbed stream- sides"). At this point, landscapes with multiple communities may emerge. Again, Valentine assumed that a full spectrum of phenotypic varia- tions are always Ukely to occur; hopeful monsters are always appearing, but their probability of long- term survival decreases rapidly as ecospace occupa- tion increases; consideration of constraints enforced by increasing developmental complexity alters this expectation, reducing the expected breadth of divergences. A consequence of this model, which we discuss below, is that the ecologi- cal hierarchy evolves in complexity and connectiv- ity as a major radiation proceeds. In doing so, it begins progressively to constrain that radiation in a negative feedback loop. All of these macroevolutionary scenarios, whether or not they include an ecological compo- nent, contrast with the gradualism taken as the default in most paleobotanical forays into evolu- tionary biology, the most obvious example being Zimmermann's (1959) unduly influential Telome Theory. Gradualist models are perhaps more "Darwinian" than "neo-Darwinian". They rely on the assumption that the environment provides a uniform landscape across which blow prevailing winds, the "selective pressures" for whatever mor- phological feature is under study. Guided by com- petition, organisms are assiuned to create an invariable backgroimd of minor phenotypic varia- tion that is subject to constant gentle selection pressures. The result is a stately, continuous unfolding, not the discontinuous, fractally pat- terned process envisioned by most macroevolution- ary models (Fig. 4). The "ecological" component in these graduaUst models is often in practice merely the "selective pressures" themselves. 3.3. Major plant groups: initiation of the class-level radiation "Why at that particular time?" is a question that can be asked about any phenotypic radiation at any taxonomic scale. Although the paleobota- nical record clearly documents an architectural radiation of major proportions during the Late Devonian and Early Carboniferous (Fig. 1), the vascular plant radiation has been viewed either from the perspective of phylogenetic pattern or almost completely in terms of individual attributes, often considered pivotal in the evolution of plants (for example, the evolution of tree habit, seeds, and laminar megaphyllous leaves). Competitive W.A. DiMichele, KM. Bateman/ReviewofPalaeobotanyandPalynology90 (1996) 223-247 Til interaction among plants is taken to be the driving force (Chaloner and Sheerin, 1979; Tiffney, 1981; Knoll, 1984, 1986; Knoll et al., 1984; Knoll and Niklas, 1987; Traverse, 1988; Selden and Edwards, 1989), and this unique Late Devonian origin of widely divergent, complex architectures becomes obscured by an undesirably narrow focus on indivi- dual structural components of the major plant groups. We suggest that phenotypic complexity played the most important role in the onset of the radia- tion, though detailed documentation of the asser- tion is outside the scope of this paper. Knoll et al. (1984) made the most credible attempt to date to document the increase in complexity that occurred during the Devonian (Fig. 5). For both the zoster- ophyll and trimerophyte lineages, they detected a rapid increase in morphological complexity during the Middle Devonian, reaching a plateau in the Late Devonian (and increasing again for lycopsids in the Early Carboniferous). This implies the exis- tence of a phenotypic (and hence developmental) threshold of complexity that, once crossed, per- mitted rapid diversification-essentially complexity begetting complexity. This threshold was reached by the gradual accumulation of potentiaUy adap- tive morphological structures during the early phases of vascular land plant evolution (Chaloner 20 LU IX o o ai 1- 1 5 z lU z Ui o > D < ? s = threshold of o complexity o I a. a. o Pr Loe 1 SI 1 Em Late*^,. Early El 1 Ql Middle Frl Fal Late 1 z SILURIANI DE VONIAN 1 Fig. 5. Rapid increase in "morphological" advancement following attainment of a "threshold of complexity" in the Siegenian. [Modified from Knoll et al. (1984) who listed the criteria for calculation of the morphological advancement score.] and Sheerin, 1979). Although simple early plants possessed most of the cell types found in modem angiosperms, diversity within particular categories of cell type (e.g. meristems, vascular tissues) was much lower. More unportantly, early land plants possessed few organ systems. Together, these factors limited their ability to produce highly diver- gent derivative forms (and, by implication, offered an increasing probability of producing convergent derivative forms). However, once a critical mass of heritable phenotypic variability and complex developmental regulation were attained, large steps could be made in the degree of divergence between ancestor and descendant. In other words, the possi- bility for generating "hopeful monsters" was lim- ited in the earliest phases of land plant evolution, but became increasingly likely as phenotypes and their regulation became more complex. This rela- tionship between elaboration of complex morphol- ogy and its developmental control underpinned Zimmermann's (1959) Telome Theory, which unfortunately has yet to be satisfactorily integrated with plant developmental mechanisms as they are understood today (e.g. Stidd, 1987). This explanation does not rule out competition as an important selective filter, though competition among vascular plants is indirect and is not neces- sarily involved in the establishment of a novel lineage (Bateman and DiMichele, 1994a) (Fig. 6). Diffuse competition, perhaps landscape-scale esca- lation as envisioned to occur periodicaUy by Vermeij (1987), could have driven the entire system once complex morphologies began to appeiir. However, competition and specific selective pres- sures alone are unlikely to account for the rela- tively short-term structural diversification of vascular plants belonging to widely divergent clades, unless the organisms were ecologically imdiiferentiated. In this context, early land plants (Late Siltirian and Early Devonian) appear to have undergone minimal ecological differentiation. "Commimities" commonly consisted of patchworks of monotypic stands (e.g. Andrews et al., 1977), with many of the species using common resources and competing for the same space within streamside and wet floodplain habitats (Knoll et al., 1979; Tiffney, 1981; Tiffney and Niklas, 1985; Bateman, 1991; 238 W.A. DiMichele, KM. Bateman/Review of Palaeobotany andPalynology 90 (1996) 223-247 A CO CO LU ECONOMIC REPRODUCTIVE DISPLACIVE REPULSIVE Fig. 6. Four phases of establishment of a new Uneage. Once the genetically novel propagule (asterisked) germinates, its individual survival requires only economic establishment. However, lineage establishment necessitates reproduction and population expansion. Any current occupants of the preferred niche must be displaced, and the novel lineage must then resist attempted invasions from other prospective occupants of the niche. Note that (1) botanical "competition" can only be indirect and mediated via differential utilization of the same resources, and (2) the displacive phase of establishment requires a higher level of fitness than the repulsive phase ("home-field advantage" sensu Pimm, 1984, 1991). EstabUshment of low-fitness "hopeful monsters" therefore requires a vacant niche. (See also Bateman and DiMichele, 1994a.) DiMichele et al., 1992). The dominant dynamics in these kinds of ecosystems would have been location and occupation of space. This was essen- tially an r-selected world. At this stage, the system should have been highly invadable from both ecological and evolutionary perspectives, due to the lack of integrated, multispecies communities on Early Devonian landscapes. Competition for space rather than light or nutrients should have permitted new species to invade and estabUsh on undersaturated landscapes with few mutualistic interactions among species (cf. Fig. 6). In current "ecojargon", the Early Devonian was a time of low connectance (plants had few coevolved inter- actions and interdependencies), high resilience (following disturbances, ecosystems were able to return to rapidly to the starting condition), and low resistance (ecosystems were easily invaded by species new to the area or newly evolved) (see Pimm, 1984). The long lag times in the develop- ment of ecological structure reflect natural lags in the evolutionary process; unlike today, in the Early Devonian there simply were no structurally com- plex species to invade the low resistance landscapes of the day. The distinct ecological strategies of the major clades apparently evolved in concert with increas- ing structural variability during the Early and Middle Devonian, when swamp and periswamp floras were already ecologically distinct (e.g. Tasch, 1957; Edwards, 1980; Gensel, 1987; Bateman, 1991; DiMichele et al., 1992). Thus, at this early stage in land plant history, ecology is less a con- straint than a reflection of the progress of pheno- typic evolution. When organisms began to evolve structural and physiological attributes that permit- ted them to escape the wetter landscapes, or to exploit those landscapes more fully, they encoun- tered little counter-selection from existing occu- pants (cf. Fig. 6). In such a scenario, physical tolerance predominates over biotic interaction in dictating evolutionary pattern. This is fully consis- tent with the evolutionary model developed by Valentine (1980) and elaborated by Erwin et al. (1987). 3.4. Ecological limits to the radiation Why did the radiation of vascular plants stop with the three major lineages of lycopsids (homo- sporous forms, selaginellaleans, and rhizomorphic forms) and three or four major trimerophyte- derived lines (sphenopsids, progymnosperms-seed plants, and one or more distinct types of ferns)? W.A. DiMichele, KM. Bateman/Review of Palaeobotany andPalynology 90 (1996) 223-247 239 Certainly, these groups do not exhibit the full spectrum of possible ways to build a vascular plant. The answer is probably two-fold. Phenotypic and genotypic constraints, at some level, impose limits on subsequent morphological change. Thus, the maximum extent of phenotypic divergence per clade must have decUned sharply once develop- mentally complex body plans evolved. A morpho- logical argument of this type is difficult to test directly due to the antiquity of the event and our inability to sample the full spectrum of variability produced by any extinct species (as opposed to the full extent of variability that survived). Nonetheless, attempts have been made to use phylogenetics to gauge phenotypic disparity among Paleozoic marine invertebrates (Briggs et al., 1992; Foote, 1994), and similar paleobotani- cal studies would vmdoubtedly yield interesting results. The second factor, again stated explicitly by Valentine (1980), is the progressive saturation of ecological resource space (Fig. 4). Most authors have treated ecological space as a relatively imi- form background. Valentine's (1980) preferred analogy was pre-existing tesserae on a checker- board (see also DiMichele et al., 1992; Hanski, 1994). Basal, highly divergent Uneages are envisioned as needing disproportionately large amounts of ecospace, due largely to weakly canal- ized development and their consequent inability to withstand serious competition for resources. Later- evolving offshoots would need progressively smaller ecospaces, within the larger initial space, as their competitive abilities became less hindered by weak developmental controls. Under this model, ecospace is pre-existing, predefined, and merely waiting for progressive occupation. Although heuristically useful, this convention is not particularly reaUstic, especially for the early phases of a novelty radiation into ecospace that has no prior biotic definition. In other words, the resources available to newly evolved species are not limited by the resource use patterns of pre- existing species. Examination of the Late Devonian-Early Carboniferous vascular plant radiation suggests that terrestrial landscapes were transformed from poorly defined with little hierarchical structure to increasingly well defined with distinct hierarchical structure. As major class-level lineages evolved and became established they began to partition the land surface and define the biotic characteris- tics of particular physical environments. These patterns then began to constrain the likelihood of establishment of subsequent highly divergent forms. Thus, clades and ecological space co- evolved; the morphological envelope and ecologi- cal envelope were established simultaneously (Fig. 4). As discussed earlier, an ecologically undersaturated environment favors a "novelty" radiation (sensu Erwin, 1992), particularly in the Paleozoic where complex morphologies and strong taxonomic preferences for particular ecological settings had not yet evolved. The combination of available ecological resource space and relatively simple organisms with,weak developmental canal- ization enhances the survival of highly derived morphologies, which are generally classified as the stem lineages of higher taxa. Such radiations begin to slow as ecological-adaptive space is filled, which selects proportionately more against progressively more derived forms. As the survival of the more deviant phenotypes declines, ecospace is filled by the constantly present background of less pro- foundly modified derivatives, which we recognize as species within the major taxonomic groups. Wetlands and terra firma landscapes formed the first major ecological dichotomy within what could be considered the lowland tropical biome. Early wetlands were primarily the province of hetero- sporous rhizomorphic lycopsids and possibly some heterosporous progymnosperms (Scheckler, 1986a,b; Rex and Scott, 1987; Scott, 1988). Although elements of other major Uneages occurred in such environments, the former groups appear to have been ecologically centered there. In contrast, terra fijma habitats, those with well drained to occasionally dry soils, became the prov- ince of a major seed-plant radiation during the Early Carboniferous (Bridge et al., 1980; Rex and Scott, 1987; Scott, 1988, 1990; Bateman and Scott, 1990; Galtier and Scott, 1990). Retallack and Dilcher (1988) argued that during the Early Carboniferous seed plants established the full spectrum of r to AT serai strategists, with habits ranging from small shrubs to large trees, in habitats 240 W.A. DiMichele, R.M. Bateman/ReviewofPalaeobotanyandPalynology90 (1996) 223-247 ranging from wet streamsides to dry soils in dis- turbed environments. By noting this ecological dichotomy we do not mean to unply that the evolutionary events that gave rise to the earUest species in a particular clade necessarily occurred in the environment (s) in which the group ulti- mately radiated. In some instances, such as that of the seed plants discussed in the first case study, the habitats of origin and those of ultimate ecologi- cal success appear to be significantly different. Certainly, it has been suggested that "marginal", "disturbed", or "stressful" environments, marine and terrestrial, may be crucibles for major pheno- typic divergences that later estabUsh and radiate in more favorable settings (e.g. Axelrod, 1967; Hickey and Doyle, 1977; Bottjer and Jablonski, 1988; DiMichele and Aronson, 1992). Sphenopsids, especially the larger calamitean species, were centered in aggradational environ- ments such as stream- and lake-sides, later occupy- ing parts of peat- and clastic swamps (Scheckler, 1986b). This is a narrow subset of the broader enviroimiental spectrum. Calamitean sphenopsids were the only major woody tree group in the Devono-Carboniferous with clonal, rhizoma- tous reproduction (Tiffney and Niklas, 1985). Examples of calamite burial and recovery growth are common in Late Carboniferous clastic sequences (Gastaldo, 1992); this ability may have been the key to their success m aggradational habitats, where equisetaleans (albeit herbaceous) remain common today. The ferns appear to have been a diverse (and probably polyphyletic) group that occupied a wide range of environments, from wet swampy settings (Rhacophyton, zygopterids; Scott and Galtier, 1985; Scheckler, 1986a; Scott and Rex, 1987) to fire and flood disturbed environments (Scott and Galtier, 1985; Scott and Rex, 1987; Scott, 1988). Monodominant stands were common in the Early Carboniferous, both in wet and dry sites, and fusinization (charcoal preservation) is common. This suggests that ferns radiated in disturbed envi- ronments where habitat disruption favored plants that could colonize and reproduce rapidly, and spread to new disturbed sites before suffering extirpation (Scott et al., 1984; Scott and Galtier, 1985; Rex and Scott, 1987; Bateman and Scott, 1990). Small stature, homosporous habit, high reproductive output, and rhizomatous growth may have permitted them to exploit such environments, where the potential for estabUshment of new species would have been high. These floristic patterns, recognizable in the Late Devonian-Early Carboniferous, continued into the Late Carboniferous (e.g. Scott, 1978; Pfefferkom and Thomson, 1982; Phillips et al., 1985; Collinson and Scott, 1987; DiMichele and Phillips, 1994), where such ecological differentiation has been well docimiented both quantitatively and quaUtatively. The Late Carboniferous also demonstrates the development of biomic differentiation within the tropical lowlands, where xeromorphic floras, con- sisting primarily of seed plants in general and conifers in particular, existed contemporaneously with the classic mesic-hydric biome that has long typifed the "coal age" in paleobotany texts (Scott and Chaloner, 1983; Lyons and Darrah, 1989). Conifer-rich xeromorphic floras appeared first in subtropical latitudes during the early Late Carboniferous, subsequently migrating southward into the tropics (e.g. Zhou, 1994). This observation gives credence to the suggestion of Knoll and Niklas (1987) that vascular plants colonized most of the the physiographic regions of the land surface early in their radiation. The radiation of major groups was not accompa- nied by a significant increase in alpha-diversity (diversity within individual habitats). Bateman (1991) critically re-examined the primary literature on Late Paleozoic floras, focusing on the Early Carboniferous of northern England and southern Scotland. By attempting to hold fades constant to avoid pooling richness measurements from different kinds of habitats, and by emphasizing potential whole-plant species, he inferred that species richness remained relatively low through the class-level radiation, averaging eight species per assemblage and perhaps as few as four per strictly delimited uniform habitat. However, species in each habitat differed greatly, both phylo- genetically and ecologically. This interpretation is consistent with Erwin's (1992) characterization of a novelty radiation, in which speciation rates are low but phenotypic divergence rates are high. Local habitats evidently remained relatively species W.A. DiMichele, R.M. Bateman/ReviewofPalaeobotanyandPalynology90 (1996) 223-247 241 poor, but major lineages occupied ecologically distinct parts of the landscape. 4. Conclusions The Devono-Carboniferous radiation of vascu- lar plants reveals patterns and processes operating at diiferent levels in the evolutionary and ecological hierarchy. Our chosen examples ? the evolution of heterospory and the seed habit, and the larger phyogenetic pattern and timing of the radiation ? reveal linkages between events happening at the populational and clade levels, at scales ranging from habitats to biomes. This radiation had several immediate and important consequences: (1) The terrestrial land surface was partitioned into subenvironments that conformed strongly to the ecological centroids of the major class-level taxa. Each of the major class-level taxa dominated some portion of the land surface in the Carboniferous, generating a taxonomically egali- tarian dominance-diversity spectrum unlike that of any subsequent time period. Today, for example, the world is largely dominated by seed plants, irrespective of whether one considers terra firma, aggradational, disturbed, or wetland habitats. Moreoever, in most of these habitats a single phylogenetically trivial but ecologically preeminent group, the angiosperms, is dominant. (2) Alpha-diversity remained low in nearly all Devono-Carboniferous assemblages. However, subsequent radiations in all groups elevated diver- sities. These radiations began in the mid-Early Carboniferous and were largely complete by the mid-Late Carboniferous in the humid lowlands, though in the drier uplands they appear to have continued into the Mesozoic. (3) Radiations at lower taxonomic levels remained confined, for the most part, to subcom- partments of the landscape. In the case of hetero- spory, for example, the earUest plants with this life history successfully occupied wetlands and remained the dominant elements in these habitats until major extrinsically induced extinctions eventually permitted seed plants to occupy the vacated resowce space. In this way, the early radiations of plants confined and directed later evolutionary events. "Home-field advantage" appears to operate at the level of population dynamics, but its effects are manifested most strongly at the levels of community organization and landscape partitioning by major clades. (4) Resource partitioning within communities increased, following the increase in the diversity of life habits and life histories. All landscapes included species specialized for the full spectrum of life histories, from short-lived, opportunistic to long-lived, site-occupying strategies. (5) We believe that the vascular-plant architec- tural radiation is a manifestation of "home-field advantage" (sensu Pinmi, 1991) on a scale much greater than that observed in modem ecosystems. The initial radiation appears to have occurred under low competition into a landscape that was unoccupied or undersatiu^ated (probably by primi- tive homosporous plants, possibly vascular s.l.) (Scott, 1980, 1990; Kenrick and Crane, 1991). Initial occupation interacted both with natural selection and developmental canalization. As organisms became more complex morphologically the likelihood of further major increases in com- plexity declined, reducing the average phenotypic distance between ancestor and descendant species (though this increases the potential for profound evolutionary decreases in complexity, breaking canalization and resetting the evolutionary clock; Bateman, 1994, 1995). As ecological resource space became increasingly saturated the likelihood of survival of major phenotypic deviants declined, due to progressive increase in the average intensity of selection. Also, the initial long-phase radiation occurred in a near-vaccuum; the potential niches had never been occupied (primary vacancy). However, as increasingly inhospitable habitats were colonized, the only opportimities lay in the secondarily vacant niches that resulted from extrin- sic environmental perturbations. Thus, large evolu- tionary jumps simultaneously became increasingly less hkely to occur and much less likely to survive if they did occur (Valentine, 1980; Erwin et al., 1987; Bateman and DiMichele, 1994a). (6) It is always tempting to find causation at the ecological level that confines a particular prob- lem. Our chosen examples ? the origin of hetero- spory and the seed habit ? may indeed have been 242 W.A. DiMichele, R.M. Bateman/Review of Palaeobotmy andPalynology 90 (1996) 223-247 mediated by strong proximate causes operating within local ecosystems. However, study of the hierarchy of ecological organization under which these patterns originated cautions us to consider the larger context. Biotic structure and interactions at levels above populations and communities may exert profound constraints on evolutionary patterns. (7) All of the key aspects of the above evolution- ary-ecological theory are at least testable indirectly via the plant fossil record. Three distinct types of paleontological data are required. First, reconstructed fossil plant species should be integrated with extant species in well foun- ded phylogenetic analyses (e.g. Crane, 1988; Donoghue, 1989; Doyle and Donoghue, 1992; Bateman and DiMichele, 1994a; Bateman, 1994, 1995). Including the most complete range of rele- vant coded taxa is unimportant for determining sister-group relationships and thereby classfying taxa on the basis of monophyly, but it is essential for the interpretation of underlymg evolutionary mechanisms. On average, maximizing the number of coded taxa included in an analysis minunizes the number of character-state transitions on each branch of the most parsimonious cladogram. This, in turn, maximizes the probabiUty of determining the sequence of acquisition of character states within a lineage by distributing them on successive branches of the tree (the null hypothesis must be that all of the character states on a single branch of a cladogram were acquired together as a single evolutionary event: Bateman and DiMichele, 1994a). Such sequences are especially important in cases like heterospory, where an apparently simple life- history change affects, either directly or indirectly, an extensive range of phenotypic characters (Bateman and DiMichele, 1994b). It is the highly iterative nature of heterospory, and the apparently broadly similar sequence of character-state acquisi- tion in each independently evolving lineage, that most strongly implicates an underlying adaptive drive (Bateman, 1995). An additional benefit of integrated phylogenies is the ability to quantify morphological disparity via branch lengths, and thus to compare the relative magnitudes of succes- sive radiations (e.g. Briggs et al., 1992; Foote, 1994). The relative timing of the origin and radiation of a clade is crucial to many of the arguments advanced in this paper. Thus, the second important line of evidence is the date of first appearance of a taxon (or character) in the fossil record. Given that any species will inevitably have existed prior to its earliest detected appearance, the fossil record offers only a minimum estunate of the absolute age of that species. Moreover, there are often severe contradictions between the relative tuning of appearance of taxa (1) observed in the fossil record, and (2) predicted by the relative positions of nodes (branch points) in a cladogram (e.g. Fisher, 1992; Norell and Novacek, 1992; Huelsenbeck, 1994). Despite these diflSculties, rela- tive timing of events can help to limit the range of possible interpretations of causes and constraints, specifically those of the origin of a key innovation and the subsequent evolutionary radiation. For example, the first five origins of heterospory occurred during a relatively brief period of 20 Myr in the Mid-Late Devonian (Fig. I). This suggests that several lineages that diverged during the earliest (Siluro-Devonian) vascular plant radiation crossed a vegetative complexity threshold that allowed them to occupy aquatic-amphibious habi- tats, but the origins are insufficiently closely spaced to suggest forcing by an extrinsic factor such as a catastrophic extinction event that freed niches for reoccupation. The third line of evidence necessary is the envi- ronment of grovrth of fossil species, as inferred during paleoecological studies. Given in situ fossils (e.g. DiMichele and Nelson, 1989; Wing et al., 1992) or adequate reconstruction of transported fossils, much information can be gathered about the preferred habitat (s) in which particular evolu- tionary radiations took place (admittedly this should not be equated with the habitat in which a key innovation first originated, which is far more difficult to determine due to the low probability of detection). In particular, we have argued repeat- edly for the importance of home-field advantage in determining the ecological dominants of particu- lar habitats. Home-field advantage assumes that the species occupying a particular niche in a partic- W.A. DiMichele, RM. Bateman/Review of Palaeobotany andPalynology 90 (1996) 223-247 243 ular habitat will resist invasion by similar pheno- types (even of high fitness); this can be tested in the fossil record if the plants are sufficiently well known and their preferred habitats have been estabhshed. Acknowledgements We thank Thomas N. Taylor and Ruben Cuneo for the invitation and support to participate in the symposium from which this paper is an outgrowth. Dan S. Chaney and Mary Parrish kindly prepared final drafts of the figures. We especially thank Bruce H. TiflFney and Douglas H. Erwin for detailed, thoughtful reviews. This is publication No. 25 from the Evolution of Terrestrial Ecosystems Consortium, which provided partial support. References Andrews, H.N., 1963. Early seed plants. Science, 142: 925-931. Andrews, H.N., Kasper, A.E., Forbes, W.H., Gensel, P.G. and Chaloner, W.G., 1977. Early Devonian flora of the Trout Valley Formation of northern Maine. Rev. Palaeobot. Palynol., 23: 255-285. Arthur, W., 1984. Mechanisms of Morphological Evolution. Wiley, New York, NY, 275 pp. Arthur, W., 1987. The Niche in Competition and Evolution. Wiley, New York, NY. Axelrod, D.I., 1967. Drought, diastrophisra, and quantum evolution. Evolution, 21: 201-209. Banks, H.P., 1975. Reclassification of Psilophyta. Taxon, 24: 401-413. Banks, H.P., 1980. Floral assemblages in the Siluro-Devonian. In: D.L. Dilcher and T.N. Taylor (Editors), Biostratigraphy of Fossil Plants. Dowden, Hutchinson and Ross, Stroudsburg, PA, pp. 1-24. Bateman, R.M., 1991. Palaeoecology. In: C.J. Cleal (Editor), Plant Fossils in Geological Investigation: The Palaeozoic. Ellis Horwood, Chichester, pp. 34-116. Bateman, R.M., 1994. Evolutionary-developmental change in the growth architecture of fossil rhizomorphic lycopsids: scenarios constructed on cladistic foundations. Biol. Rev., 69: 527-597. Bateman, R.M., 1995. Non-floral homoplasy and evolutionary scenarios in living and fossil land-plants. In: M.J. Sanderson and L. Hufiford (Editors), Homoplasy and the Evolutionary Process. Academic, London, in press. Bateman, R.M. and DiMichele, W.A., 1994a. Saltational evolution of form in vascular plants: a neoGoldschmidtian synthesis. In: D.S. Ingram and A. Hudson (Editors), Shape and Form in Plants and Fungi. Academic, London, pp. 61-100. Bateman, R.M. and DiMichele, W.A., 1994b. Heterospory: the most iterative key innovation in the evolutionary history of the plant kingdom. Biol. Rev., 69: 345-417. Bateman, R.M. and Scott, A.C., 1990. A reappraisal of the Dinantian floras at Oxroad Bay, East Lothian, Scotland. 2. Volcanicity, palaeoecology and palaeoenviroimients. Trans. R. Soc. Edinburgh B, 81: 161-194. Bateman, R.M., DiMichele, W.A. and WiUard, D.A., 1992. Experimental cladistic analysis of anatomically preserved arborescent lycopsids from the Carboniferous of Euramerica: an essay on paleobotanical phylogenetics. Aim. Mo. Bot. Gard., 79: 500-559. Bateson, W., 1894. Materials for the Study of Variation Treated with Especial Research to Discontinuity in the Origin of Species. Macmillan, London. Beck, C.B., 1985. Gymnosperm phytogeny?a commentary on the views of S.V. Meyen. Bot. Rev., 51: 273-294. Bottjer, D.J. and Jablonski, D., 1988. Paleontological patterns in the evolution of post-Paleozoic benthic marine inverte- brates. Palaios, 3: 540-560. Brett, C.E. and Baird, G.C., 1995. Co-ordinated stasis and evolutionary ecology of Silurian to Middle Devonian faunas in the Appalachian Basin. In: D.H. Erwin (Editor), New Approaches to Speciation in the Fossil Record. Columbia Univ. Press, New York, NY, pp. 285-315. Bridge, J.S., Van Veen, P.M. and Matten, L.C., 1980. Aspects of the sedimentology, palynology and palaeobotany of the Upper Devonian of southern Kerry Head, Co. Kerry, Ireland. Geol. J., 15(2): 143-170. Briggs, D.E.G., Fortey, R.A. and Wills, M.A., 1992. Morphological disparity in the Cambrian. Science, 256: 1670-1673. Chaloner, W.G. and Hemsley, A.R., 1991. Heterospory: cul- de-sac or pathway to the seed? In: S. Blackmore and S.H. Barnes (Editors), Pollen and Spores: Patterns of Diversification. Oxford Univ. Press, Oxford, pp. 151-167. Chaloner, W.G. and Pettitt, J.M., 1987. The inevitable seed. BuU. Soc. Bot. Fr., 134: 39-49. Chaloner, W.G. and Sheerin, A., 1979. Devonian macrofloras. In: M.R. House, C.T. Scrutton and M.G. Bassett (Editors), The Devonian System. Spec. Pap. Palaeontol., 23: 145-161. Collinson, M.E. and Scott, A.C., 1987. Factors controlling the organisation and evolution of ancient plant communities. In: G.H.R. Gee and P.S. Giller (Editors), Organisation of Communities Past and Present. Blackwell, Oxford, pp. 399-420. Crane, P.R., 1988. Major clades and relationships in higher gymnosperms. In: C.B. Beck (Editor), Origin and Evolution of Gymnosperms. Columbia Univ. Press, New York, NY, pp. 218-272. Crane, P.R., 1990. The phylogenetic context of microsporogene- sis. In: S. Blackmore and S.B. Knox (Editors), Microspores: Evolution and Ontogeny. Academic, London, pp. 11-41. 244 W.A. DiMichele, KM. BatemanIReview of Palaeobotany andPalynology 90 (1996) 223-247 Damuth, J., 1985. Selection among "species": a formulation in terms of natural functional units. Evolution, 39: 1132-1146. DiMichele, W.A. and Aronson, R.B., 1992. The Pennsylvanian- Permian vegetational transition: a terrestrial analogue to the onshore-offshore hypothesis. Evolution, 46: 807-824. DiMichele, W.A. and Nelson, W.J., 1989. Small scale spatial heterogeneity in Pennsylvanian-age vegetation from the roof shale of the Springfield coal (Illinois Basin). Palaios, 4: 276-280. DiMichele, W.A. and Phillips, T.L., 1994. Paleobotanical and paleoecological constraints on models of peat formation in the Late Carboniferous of Euramerica. Palaeogeogr. Palaeoclimatol. Palaeoecol., 106: 39-90. DiMichele, W.A., Davis, J.I. and Olmstead, R.G., 1989. Origins of heterospory and the seed habit. Taxon, 38: 1-11. DiMichele, W.A., Phillips, T.L. and Ohnstead, R.G., 1987. Opportunistic evolution: abiotic environmental stress and the fossil record of plants. Rev. Palaeobot. Palynol., 50: 151-178. DiMichele, W.A., Hook, R.W., Beerbower, R., Boy, J.A., Gastaldo, R.A., Hotton, N., Ill, Phillips, T.L., Scheckler, S.E., Shear, W.A. and Sues, H.-D., 1992. Paleozoic terrestrial ecosystems. In: A.K. Behrensmeyer, J.D. Damuth, W.A. DiMichele, R. Potts, H.-D. Sues and S.L. Wmg (Editors), Terrestrial Ecosystems Through Time. Univ. Chicago Press, Chicago, IL, pp. 204-325. Donoghue, M.J., 1989. Phylogenies and the analysis of evolutionary sequences, with examples from seed plants. Evolution, 43: 1137-1156. Donoghue, M.J. and Doyle, J.A., 1989. Phylogenetic studies of seed plants and angiosperms based on morphological characters. In: B. Femhohn, K. Bremer and H. Jomvall (Editors), The Hierarchy of Life. Elsevier, Amsterdam, pp. 181-193. Doyle, J.A. and Donoghue, M.J., 1992. Fossils and seed plant phylogeny reanalyzed. Brittonia, 44: 89-106. Dyer, A.F. and Lindsay, S., 1992. Soil spore banks of temperate ferns. Am. Fern J., 82: 89-122. Eberle, J., Hasabe, M., Linkel, C, Nemacheck, J., Wen, C. and Banks, J., 1994. Sex determining genes in the homospo- rous fern Ceratopteris. Am. J. Bot., 81(6): 124-125 (abstr.). Edwards, D., 1980. Early land floras. In: A.L. Panchen (Editor), The Terrestrial Environment and the Origin of Land Vertebrates. Academic, New York, NY, pp. 55-85. Eldredge, N., 1989. Macroevolutionary Dynamics: Species, Niches, and Adaptive Peaks. McGraw-Hill, New York, NY. Eldredge, N. and Gould, S.J., 1972. Punctuated equilibria: an alternative to phyletic gradualism. In: T.J.M. Schopf (Editor), Models in Paleobiology. Freeman, San Francisco, CA, pp. 82-115. Erwin, D.H., 1992. A preliminary classification of evolutionary radiations. Hist. Biol., 6: 133-147. Erwin, D.H., 1993. The origin of metazoan development: a palaeobiological perspective. Biol. J. Linn. Soc, 50: 255-274. Erwin, D.H. and Valentine, J.W., 1984. 'Hopeful' monsters, transposons, and Metazoan radiation. Proc. Natl. Acad. Sci. USA, 81: 5482-5483. Erwin, D.H., Valentine, J.W. and Sepkoski, J.J., 1987. A comparative study of diversification events: the early Paleozoic versus the Mesozoic. Evolution, 41: 1177-1186. Fisher, D.C., 1992. Stratigraphic parsimony. In: W.P. Maddison and D.R. Maddison (Editors), McClade 3: Analysis of Phylogeny and Character Evolution. Sinauer, Sunderland, MA, pp. 124-129. Foote, M., 1994. Morphological disparity in Ordovician- Silurian crinoids and the early saturation of morphological space. Paleobiology, 20: 320-344. Frumhoff, P.C. and Reeve, H.K., 1994. Using phylogenies to test theories of adaptation: a critique of some current proposals. Evolution, 48: 172-180. Galtier, J. and Rowe, N.P., 1989. A primitive seed-like structure and its impUcations for early gynmosperm evolu- tion. Nature, 340: 225-227. Galtier, J. and Scott, A.C., 1985. The diversification of early ferns. Proc. R. Soc. Edinburgh B, 86: 289-301. Galtier, J. and Scott, A.C., 1990. On Eristophyton and other gymnosperms from the Lower Carboniferous of Castleton Bay, East Lothian, Scotland. Geobios, 23: 5-19. Gastaldo, R.A., 1992. Regenerative growth in fossil horsetails following burial by alluvium. Hist. Biol, 6: 203-219. Gensel, P.G., 1987. The evolution of early land plants. Am. Sci., 75: 478-489. Gensel, P.G., 1992. Phylogenetic relationships of the zostero- phylls and lycopsids: evidence from morphology, paleo- ecology and cladistic methods of inference. Aim. Mo. Bot. Gard., 79: 450-473. Gensel, P.G. and Andrews, H.N., 1984. Plant Life in the Devonian. Praeger, New York, NY, 380 pp. Gillespie, W.H., Rothwell, G.W. and Sheckler, S.E., 1981. The earliest seeds. Nature, 293: 462-464. Gilmour, J.S.L. and Heslop-Harrison, J., 1954. The deme terminology and the units of microevolutionary change. Genetica, 27: 147-161. Goldschmidt, R., 1940. The Material Basis of Evolution. Yale Univ. Press, New Haven, CT, 436 pp. Gould, S.J., 1977. Ontogeny and Phylogeny. Belknap, Cambridge, MA, 501 pp. Gould, S.J., 1991. The disparity of the Burgess Shale arthropod fauna and the limits of cladistic analysis: why we must strive to quantify morphospace. Paleobiology, 17: 411-423. Gould, S.J. and Eldredge, N., 1977. Punctuated equilibria: the tempo and mode of evolution reconsidered. Trans. R. Soc. London B, 3: 115-151. Gould, S.J. and Lewontin, R.C., 1979. The spandrels of San Marco and the Panglossian paradigm: a critique of the adaptationist program. Proc. R. Soc. London B, 205: 581-598. Gould, S.J. and Vrba, E., 1982. Exaptation?a missing term in the science of form. Paleobiology 8: 4-15. Haig, D. and Westoby, M., 1988a. Sex expression in homospo- rous ferns: an evolutionary perspective. Evol. Trends Plants, 2: 111-119. Haig, D. and Westoby, M., 1988b. A model for the origin of heterospory. J. Theor. Biol., 134: 257-272. W.A. DiMichele, R.M. Bateman/ReviewofPalaeobotanyandPalynology90 (1996) 223-247 245 Hanski, I., 1994. Patch-occupancy dynamics in fragmented landcapes. Trends Ecol. Evol., 9: 131-135. Hasebe, M., Wolf, P.G., Hauk, W., Manhart, J.R., Haufler, C.H., Sano, R., Gastony, G.J., Crane, N., Pryer, K.M., Murakami, N., Yokoyama, J. and Ito, M., 1994. A global analysis of fern phylogeny based on ricL nucleotide sequences. Am. J. Bot. (suppl.), 87: 120-121 (abstr.). Hickey, L.J. and Doyle, J.A., 1977. Early Cretaceous fossil evidence for angiosperm evolution. Bot. Rev., 43: 1-104. Huelsenbeck, J.P., 1994. Comparing the stratigraphic record to estimates of phylogeny. Paleobiology 20: 470-483. Hutchinson, G.E., 1965. The Ecological Theater and the Evolutionary Play. Yale Univ. Press, New Haven, CT. Kenrick, P. and Crane, P.R., 1991. Water conducting cells in early land plants: implications for the early evolution of tracheophytes. Bot. Gaz., 152: 335-356. Knoll, A.H., 1984. Patterns of extinction in the fossil record of vascular plants. In: M. Nitecki (Editor), Extinctions. Univ. Chicago Press, Chicago, IL, pp. 21-68. Knoll, A.H., 1986. Patterns of change in plant communities through geological time. In: J. Diamond and T.J. Case (Editors), Community Ecology. Harper and Row, New York, NY, pp. 126-141. Knoll, A.H. and Niklas, K.J., 1987. Adaptation, plant evolution, and the fossil record. Rev. Palaeobot. Palynol., 50: 127-149. Knoll, A.H. and Rothwell, G.W., 1981. Paleobotany: perspec- tives in 1980. Paleobiology, 7: 7-35. Knoll, A.H., Niklas, K.J. and Tiffney, B.H., 1979. Phanerozoic land plant diversity in North America. Science, 206: 1400-1402. Knoll, A.H., Niklas, K.J., Gensel, P.G. and Tiffney, B.H., 1984. Character diversification and patterns of evolution in early vascular plants. Paleobiology, 10: 34-47. Leroi, A.M., Rose, M.R. and Lauder, G.V., 1994. What does the comparative method reveal about adaptation? Am. Nat., 143: 381-402. Levin, D.A., 1993. Local speciation in plants: the rule not the exception. Syst. Bot., 18: 197-208. Levinton, J., 1988. Genetics, Paleontology and Macroevolution. Cambridge Univ. Press, Cambridge. Long, A.G., 1975. Further observations on some Lower Carboniferous seeds and cupules. Trans. R. Soc. Edinburgh B, 69: 26-93. Lyons, P.C. and Darrah, W.C, 1989. Earliest conifers in North America: upland and/or paleoclimatic indicators. Palaios, 4: 480-486. MSgdefrau, K., 1956. Palaobiologie der Pflanzen. Fischer, Jena, 3rd ed. Meyen, S.V., 1984. Basic features of gymnosperm systematics and phylogeny as evidenced by the fossil record. Bot. Rev., 50: 1-111. Meyen, S.V., 1987. Fundamentals of Palaeobotany. Chapman and Hall, London, 432 pp. Mosbrugger, V., 1990. The Tree Habit in Land Plants. Springer, Berlin, Lecture Notes Earth Sci., 28. Niklas, K.J., 1981. Air flow patterns around some early seed plant ovules and cupules: implications concerning efficiency in wind pollination. Am. J. Bot., 68: 635-650. Niklas, K.J., Tiffney, B.H. and Knoll, A.H., 1985. Patterns in land plant diversification: an analysis at the species level. In: J.W. Valentine (Editor), Phanerozoic Diversity Patterns: Profiles in Macroevolution. Princeton Univ. Press, Princeton, NJ, pp. 97-128. Nixon, K.C., Crepet, W.L., Stevenson, D. and Friis, E.M., 1994. A reevaluation of seed plant phylogeny. Ann. Mo. Bot. Gard., 81: 484-533. Norell, N. and Novacek, M., 1992. The fossil record and evolution: comparing cladistic and paleontologic evidence for vertebrate history. Science, 255: 1690-1693. Pettitt, J.M., 1970. Heterospory and the origin of the seed habit. Biol. Rev., 45: 401-415. Pettitt, J.M. and Beck., C.B., 1968. Archaeosperma arnoldii? a cupulate seed from the Upper Devonian of North America. Contrib. Mus. Paleontol. Univ. Mich., 22: 139-154. Pfefferkom, H.W. and Thomson, M.C., 1982. Changes in dominance patterns in Upper Carboniferous plant-fossil assemblages. Geology, 10: 641-644. Phillips, T.L. and DiMichele, W.A., 1992. Comparative life- history biology of arborescent lycopsids in Late Carboniferous swamps of Euramerica. Ann. Mo. Bot. Gard., 79: 560-588. PhiUips, T.L. and Peppers, R.A., 1984. Changing patterns of Pennsylvanian coal-swamp vegetation and imphcations of chmatic control on coal occurrence. Int. J. Coal Geol., 3: 205-255. Phillips, T.L., Peppers. R.A. and DiMichele, W.A., 1985. Stratigraphic and interregional changes in Pennsylvanian coal-swamp vegetation: environmental inferences. Int. J. Coal Geol., 5: 43-109. Pimm, S.L., 1984. The complexity and stability of ecosystems. Nature, 307: 321-326. Pimm, S.L., 1991. The Balance of Nature? Univ. Chicago Press, Chicago, IL, 434 pp. Raymond, A., 1987. Interpreting ancient swamp conmiunities: can we see the forest in the peat? Rev. Palaeobot. Palynol., 52: 217-231. Retallack, G.J. and Dilcher, D.L., 1988. Reconstuctions of selected seed ferns. Ann. Mo. Bot. Gard., 75: 1010-1057. Rex, G. and Scott, A.C., 1987. The sedimentology, palaeo- ecology, and preservation of the Lower Carboniferous plant deposits at Pettycur, Fife, Scotland. Geol. Mag., 124: 43-66. Riedl, R., 1979. Order m Living Things. Wiley, New York, NY. Rothwell, G.W., 1987. The role of development in plant phylogeny: a paleobotanical perspective. Rev. Palaeobot. Palynol, 50: 97-114. Rothwell, G.W. and Scheckler, S.E., 1988. Biology of ancestral gymnosperms. In: C.B. Beck (Editor), Origin and Evolution of Gynmospenns. Columbia Univ. Press, New York, NY, pp. 85-134. Rothwell, G.W. and Serbet, R., 1994. Lignophyte phylogeny and the evolution of spermatophytes: a ntmierical cladistic analysis. Syst. Bot., 19: 443-482. Rothwell, G.W. and Stockey, R.A., 1994. The role of 2A6 W.A. DiMichele, KM. Bateman/Review of Palaeobotany andPalynology 90 (1996) 223-247 Hydropteris pinnata gen. et sp. nov. in reconstructing the cladistics of heterosporous ferns. Am. J. Bot., 81: 479-492. Scheckler, S.E., 1986a. Floras of the Devonian-Mississippian transition. In: R.A. Gastaldo and T.W. Broadhead (Editors), Land Plants: Notes for a Short Course. Univ. Tenn. Dep. Geol. Sci. Stud. Geol., 15: 81-96. Scheckler, S.E., 1986b. Geology, floristics and paleoecology of Late Etevonian coal swamps from Appalachian Laurentia (U.S.A.). Ann. Soc. G6ol. Belg., 109: 209-222. Schindewolf, O., 1950. Grundfragen der PalSontologie. Schweizerbart, Stuttgart. Scott, A.C., 1978. Sedimentological and ecological control of Westphalian B plant assemblages from West Yorkshire. Proc. Yorkshire Geol. Soc., 41: 461-508. Scott, A.C., 1980. The ecology of some Upper Palaeozoic floras. In: A.L. Panchen (Editor), The Terrestrial Environment and the Origin of Land Vertebrates. Syst. Assoc. Spec. Vol., 15. Academic, London, pp. 87-115. Scott, A.C., 1988. Volcanoes, fires, and the Lower Carboniferous vegetation of Scotland. Open Univ. Geol. Soc. J., 8(2): 27-31. Scott, A.C., 1990. Preservation, evolution, and extinction of plants in Lower Carboniferous volcanic sequences in Scotland. In: M.G. Lockley and A. Rice (Editors), Volcanism and Fossil Biotas. Geol. Soc. Am. Spec. Pap., 244: 25-38. Scott, A.C. and Chaloner, W.G., 1983. The earUest conifer from the Westphalian B of Yorkshire. Proc. R. Soc. London B, 220: 163-182. Scott, A.C. and Galtier, J., 1985. Distribution and ecology of early ferns. Proc. R. Soc. Edinburgh B, 86: 141-149. Scott, A.C. and Rex, G.M., 1987. The accumulation and preservation of Dinantian plants from Scotland and its borders. In: J. Miller, A.E. Adams and V.P. Wright (Editors), European Dinantian Environments. Wiley, New York, NY, pp. 329-344. Scott, A.C., Galtier, J. and Clayton, G., 1984. Distribution of anatomically preserved floras in the Lower Carboniferous of western Europe. Trans. R. Soc. Edinburgh Earth Sci., 75: 311-340. Selden, P.A. and Edwards, D., 1989. Colonisation of the land. In: K.C. Allen and D.E.G. Briggs (Editors), Evolution and the Fossil Record. Belhaven, Chichester, pp. 122-152. Sheffield, E., 1994. Alternation of generations in ferns: mechanisms and significance. Biol. Rev., 69: 331-343. Simpson, G.G., 1944. Tempo and Mode in Evolution. Columbia Univ. Press, New York, NY, 237 pp. Simpson, G.G., 1953. The Meaning of Evolution. Columbia Univ. Press, New York, NY. Stanley, S.M., 1975. A theory of evolution above the species level. Proc. Natl. Acad. Sci., 72: 646-650. Stanley, S.M., 1979. Macroevolution: Pattern and Process. Freeman, San Francisco, CA. Stanley, S.M., 1990. Adaptive radiation and evolution. In: P.D. Taylor and G.P. Larwood (Editors), Major Evolutionary Radiations. Syst. Assoc. Spec. Vol., 42. Clarendon, Oxford, pp. 1-15. Stebbins, G.L., 1992. Comparative aspects of plant morpho- genesis: a cellular, molecular, and evolutionary approach. Am. J. Bot., 79: 589-598. Steeves, T.A., 1983. The evolution and biological significance of seeds. Can. J. Bot., 61: 3550-3560. Stein, W.E., Jr., 1987. Phylogenetic analysis and fossil plants. Rev. Palaeobot. Palynol., 50: 31-61. Stein, W.E., Jr., 1993. Modeling the evolution of stelar architecture in vascular plants. Int. J. Plant Sci., 154: 229-263. Stein, W.E. and Beck, C.B., 1987. Paraphyletic groups in phylogenetic analysis: Progymnospermopsida and Pr6phan6rogames in alternative views of seed-plant relation- ships. Bull. Soc. Bot. Fr., 134: 107-119. Stein, W.E., Jr., Wight, D.C. and Beck, C.B., 1984. Possible alternatives for the origin of the Sphenopsida. Syst. Bot., 9: 102-118. Stidd, B.M., 1980. The neotenous origin of the pollen organ of the gymnosperm Cycadeoidea and implications for the origin of higher taxa. Paleobiology, 6: 161-167. Stidd, B.M., 1987. Telomes, theory change, and the evolution of vascular plants. Rev. Palaeobot. Palynol., 50: 115-126. Sussex, I.M., 1966. The origin and development of heterospory in vascular plants. In: E.G. Cutter (Editor), Trends in Plant Morphogenesis. Longman, Harlow, pp. 141-152. Tasch, P., 1957. Flora and fauna of the Rhynie chert: a paleoecological evaluation of the published evidence. Univ. Wichita Bull., 32: 1-24. Thomson, R.B., 1934. Heterothally and the seed habit versus heterospory. New Phytol., 33: 41-44. Tiffhey, B.H., 1981. Diversity and major events in the evolution of land plants. In: K.J. Niklas (Editor), Paleobotany, Paleoecology and Evolution. Praeger, New York, NY, pp. 193-230. Tiff'ney, B.H. and Niklas, K.J., 1985. Clonal growth in land plants: a paleobotanical perspective. In: J.B.C. Jackson, L.W. Buss and R.E. Cook (Editors), Population Biology and Evolution of Clonal Organisms. Yale Univ. Press, New Haven, CT, pp. 35-66. Traverse, A., 1988. Plant evolution dances to a different beat. Hist. Biol., 1: 277-301. Trivett, M.L. and Rothwell, G.W., 1988. Modeling the growth architecture of fossil plants: a Paleozoic filicalean fern. Evol. Trends Plants, 2: 25-29. Tryon, A.F., 1964. Platyzoma?a Queensland fern with incipient heterospory. Am. J. Bot., 51: 939-942. Tryon, R.M. and Tryon, A.F., 1982. Ferns and Fern Allies. Springer, New York, NY. Valentine, J.W., 1980. Determinants of diversity in higher taxonomic categories. Paleobiology, 6: 444-450. Valentine, J.W., Tiffney, B.H. and Sepkoski, J.J., Jr., 1991. Evolutionary dynamics of plants and animals: a comparative approach. Palaios, 6: 81-88. Van Steenis, C.G.G.J., 1969. Plant speciation in Malesia with special reference to the theory of non-adaptive, saltatory evolution. Biol. J. Linn. Soc, 1: 97-133. Vermeij, G.J., 1987. Evolution and Escalation. Princeton Univ. Press, Princeton, NJ, 527 pp. W.A. DiMichele, KM. BatemanjReview of Palaeobotany and Palynology 90 (1996) 223-247 247 Waddington, C.H., 1942. Canalization of development and inheritance of acquired characters. Nature, 150: 563-564. Wight, D.C., 1987. Non-adaptive change in early land plant evolution. Paleobiology, 13: 208-214. Willson, M.F., 1981. Sex expression in some fern gametophytes: some evolutionary possibilities. J. Theor. Biol., 93: 403-409. Wing, S.L., Hickey, L.J. and Swisher, C.C, 1993. Implications of an exceptional fossil flora for Late Cretaceous vegetation. Nature, 363: 342-344. Zhou, Y.-X., 1994. Earliest pollen-dominated microfloras from the early Late Carboniferous of the Tian Shan Mountains, NW China: their significance for the origin of conifers and palaeophytogeography. Rev. Palaeobot. Palynol., 81: 193-211. Zimmermann, W., 1959. Die Phylogenie der Pflanzen. Fischer, Stuttgart.