Xylem heterochrony: an unappreciated key to angiosperm origin and diversifications

Authors


*E-mail: s.carlquist@verizon.net

Abstract

All angiosperms can be arranged along a spectrum from a preponderance of juvenile traits (cambial activity lost) to one of nearly all adult characters (cambium maximally active, mature patterns realized rapidly early in ontogeny). Angiosperms are unique among seed plants in the width of this spectrum. Xylem patterns are considered here to be indicative of contemporary function, not relictual. Nevertheless, most families of early-divergent angiosperms exhibit paedomorphic xylem structure, a circumstance that is most plausibly explained by the concept that early angiosperms had sympodial growth forms featuring limited accumulation of secondary xylem. Sympodial habits have been retained in various ways not only in early-divergent angiosperms, but also among eudicots in Ranunculales. The early angiosperm vessel, relatively marginal in conductive abilities, was improved in various ways, with concurrent redesign of parenchyma and fibre systems to enhance conductive, storage and mechanical capabilities. Flexibility in degree of cambial activity and kinds of juvenile/adult expressions has been basic to diversification in eudicots as a whole. Sympodial growth that lacks cambium, such as in monocots, provides advantages by various features, such as organographic compartmentalization of tracheid and vessel types. Woody monopodial eudicots were able to diversify as a result of production of new solutions to embolism prevention and conductive efficiency, particularly in vessel design, but also in parenchyma histology. Criteria for paedomorphosis in wood include slow decrease in length of fusiform cambial initials, predominance of procumbent ray cells and lesser degrees of cambial activity. Retention of ancestral features in primary xylem (the ‘refugium’ effect) is, in effect, a sort of inverse evidence of acceleration of adult patterns in later formed xylem. Xylem heterochrony is analysed not only for all key groups of angiosperms (including monocots), but also for different growth forms, such as lianas, annuals, various types of perennials, rosette trees and stem succulents. Xylary phenomena that potentially could be confused with heterochrony are discussed. Heterochronous xylem features seem at least as important as other often cited factors (pollination biology) because various degrees of paedomorphic xylem are found in so many growth forms that relate in xylary terms to ecological sites. Xylem heterochrony can probably be accessed during evolution by relatively simple gene changes in a wide range of angiosperms and thus represents a current as well as a past source of variation upon which diversification was based. Results discussed here are compatible with both current molecular-based phylogenetic analyses and all recent physiological work on conduction in xylem and thus represent an integration of these fields. © 2009 The Linnean Society of London, Botanical Journal of the Linnean Society, 2009, 161, 26–65.

INTRODUCTION

Evolutionary interpretation of wood has seemed elusive to most botanists because wood is the most complicated tissue in plants. Both long-term evolutionary trends and more localized ecological adaptations are evident in secondary xylem. Data on wood are available in huge compendia and numerous monographs and thus are difficult to access and many xylary features defy easy categorization and thus do not lend themselves to simple definitions or inclusion in data matrices. Few wood anatomists have attempted to present coherent and easily understood accounts of wood evolution or wood anatomy in relation to physiology. Study of wood anatomy has, for most of its history, been skewed in favour of woodier species, partly because many wood anatomists have been trained in schools of forestry and financial support for study of woods has been, directly or indirectly, related to woods that have commercial applications. Doubtless many botanists would like to understand the functional and evolutionary nature of xylem in angiosperms and other groups of vascular plants, but feel excluded because the complexity of xylem and the diversity of literature on xylem make entry into this field difficult for those who cannot devote large amounts of professional time to developing an understanding of xylem structure and diversity. Attempts at unifying knowledge of xylem and making xylem patterns lucid and readily understandable have been few. The result has been neglect and avoidance of xylem by many who would like a simple yet comprehensive picture of how primary and secondary xylem function and evolve. The present paper is an attempt to offer an account of unappreciated unifying features and is intended to help make understanding of xylem less elusive. Heterochrony can be seen in terms of comparative anatomy, which shows the character combinations in which vessels function successfully, and comparative anatomy is a necessary component of this study.

Understanding of heterochrony in xylem has been slow to develop. In an earlier paper (Carlquist, 1962) I introduced the idea that paedomorphosis occurs in certain angiosperms, most notably rosette trees, ‘woody herbs’ and other plants that are not typical trees. Secondary woodiness in angiosperms, at that time a virtually unexplored concept, inspired my 1962 paper to no small degree. Because the perception of paedomorphy (or protracted juvenilism) in secondary xylem was new at that time, the 1962 paper took the form of presenting what I thought were a few plausible instances. That the concept of xylem paedomorphosis (the retention of juvenile traits throughout the life of a tree) might include not only dicots with secondary woodiness but also be applicable to flowering plants as a whole occurred to me only recently. Heterochrony is applicable to the entire range of xylary expressions in angiosperms, from those without cambial activity such as monocots (but not excluding other instances of cambial cessation) to those with accelerated adulthood (woody trees and other woody plants that attain adult patterns rapidly). All angiosperms can be ranged along a wide spectrum between those two extreme conditions. Because this hypothesis is a new one, documentation is necessary. Therefore, the present paper includes descriptions and photographic illustrations (Figs 6–50) of a full range of paedomorphic phenomena (and the opposite: progenesis or accelerated adulthood) in angiosperm xylem and schematic summaries (Figs 1–5) to which these photographic examples can be referred.

Figure 6–9.

Sections of wood of basal angiosperms. Figures 6–7. Amborella trichopoda Baill. (Amborellaceae; Carlquist 5333, RSA), wood from a larger stem. Fig. 6. Tangential section. Most rays shown are uniseriate and are composed only of upright cells. Fig. 7. A radial section of a multiseriate ray cell cells are square or upright. Figures 8–9. Thottea siliquosa (Lamk.) Ding Hou (Aristolochiaceae; Fosberg 57358, RSA), wood from the base of a canelike stem. Fig. 8. Transection showing secondary growth as well as the periphery of the pith (bottom). The wide rays do not change during secondary growth. Fig. 9. Tangential section. All rays are multiseriate and composed of markedly upright (vertically elongate) cells. Scale bar, Figures 6–9, scale at lower left in Figure 6, 80 µm.

Figure 10–13.

Wood sections of Illicium, showing the outermost (and therefore the mature wood patterns) of two species. Figures 10–11. Illicium ridleyanum A.C.Sm (Carlquist 4442, RSA), a small tree. Fig. 10. Tangential section. Uniseriate rays outnumber multiseriate rays; because most ray cells are upright, many are similar in appearance to the background of tracheids. Fig. 11. Radial section of a multiseriate ray. All cells are square or upright; note prominently upright cells at top. Figures 12–13. Illicium anisatum L. (Aw-8950), a medium-sized tree. Fig. 12. Tangential section. Uniseriate rays are scarce; multiseriate rays are not as tall as in I. ridleyanum. Fig. 13. Radial section. All cells in the multiseriate ray (bottom) are procumbent except for cells at the top of the ray. Uniseriate rays (portion at top of photograph) consist of square to upright cells. Scale for Figures 10–13 shown in Figure 6.

Figure 14–18.

Wood sections of Austrobaileyales. Figures 14–16. Austrobaileya scandens C.T.White (Carlquist 1390, RSA). Fig. 14. Tangential section, showing a wide multiseriate ray, such as are found in many lianas; uniseriate and biseriate rays at left. Fig. 15. Radial section, showing the centre of a multiseriate ray. There is a rapid transition in ray cell shape from the pith (not shown, to the left of the left edge of the photograph) toward the cambium (not shown, to the right of the right edge of the photographs). The ray changes abruptly from upright cells (left) to procumbent cells (right). Fig. 16. Transection of pith and earlier-formed secondary xylem. The multiseriate rays of the secondary xylem are continuations of the wide primary rays. Fig. 17. Schisandra sphenanthera Rehder & E.H.Wilson (Carlquist 8051, RSA). Transection showing early-formed secondary xylem and adjacent pith. Primary rays do not continue into the secondary xylem; changes in the cambium result in production of imperforate tracheary elements in potential ray areas. Fig. 18. Illicium ridleyanum (Carlquist 4442, RSA). Transection showing early-formed secondary xylem and adjacent pith; as in Figure 19, potential multiseriate ray areas are rapidly converted into imperforate tracheary elements. The narrow vessel size is characteristic of a non-scandent plant as compared with those of scandent relatives (Fig. 17–18). Scale, Figures 14–18 as in Figure 6.

Figure 19–23.

Wood sections of Sarcandra glabra (Thunb.) Nakai (Chloranthaceae), Carlquist 15680 (RSA). Figures 19–20. Sections of stem wood. Fig. 19. Transection, showing vesselless condition; multiseriate ray to left of centre. Fig. 20. Radial section of multiseriate ray; all ray cells are upright. Figures 21–23. Sections of root wood. Fig. 21. Transection; arrows indicate three vessels (overlap area of end walls subdivide the vessels). Figures 22–23. SEM photographs of perforation plates of vessels from radial section. Fig. 22. View of most of the length of a perforation plate. Bars are wide, perforations narrow. Fig. 23. Portion of perforation plate, showing the threadlike pit membrane remnants. Scale bars, Figures 19–20, as in Figure 6; Figure 21, 40 µm; Figure 22, 25 µm; Figure 23, 10 µm.

Figure 24–28.

Wood sections of Chloranthaceae. Figures 24–25. Hedyosmum scabrum Solms (SJRw-28642). Fig. 24. Tangential section, showing two multiseriate rays; at extreme right, perforation plate in a vessel and, to the left of it, a uniseriate ray two cells high. Fig. 25. Radial section of a multiseriate ray; all ray cells are upright. Figures 26–28. SEM views of perforation plates from radial sections of wood. Fig. 26. Hedyosmum cumbalense H.Karst. (Todzia 2904, MO); transition in degree of pit membrane presence from end of perforation plate (which would be above the top of the photograph) to the central portion of the perforation plate (below). Figures 27–28. Ascarina lanceolata Hook.f. (SJRw-29993). Fig. 27. The entirety of a perforation plate, showing the large number of bars. Fig. 28. Ascarina philippinensis C.B.Rob. (MADw-2830); portions of five perforations, showing extensive presence of pit membranes perforated by holes of various sizes. Scale bars, Figures 24–25, scale as in Figure 6; Figure 26, 15 µm; Figure 27, 100 µm; Figure 28, 10 µm.

Figure 29–32.

Transections of stems of Saururaceae, phloem to the left (Figs 29–30) and monocot stems (Figs 31–32). Fig. 29. Saururus cernuus L. (collected by Shirley Tucker, Shreveport, Louisiana), two bundles from stem, some secondary growth present; cambial activity in primary ray areas (arrows) is vestigial. Fig. 30. Houttuynia cordata Thunb. (cultivated at University of California at Santa Barbara); sclerenchyma present as a sheath around bundles and in primary ray areas; little or no cambial activity present. Fig. 31. Clintonia borealis Raf. (collected near Ripon, Wisconsin), bundles of scape; bundles are surrounded by sclerenchyma, no cambial activity present. Fig. 32. Yucca brevifolia Engelm. (cultivated at Rancho Santa Ana Botanic Garden, Claremont, California); bundle with phloem above, large number of tracheids below (no sclerenchyma present). A few tangential divisions (arrows) may represent either a vestigial cambium or the last divisions of procambial cells, but few if any phloem or xylem cells have been derived from those divisions. Scale, Figures 29–32, scale at lower left in Figure 21.

Figure 33–37.

Wood sections of Winteraceae (Figs 33–35) and Lauraceae (Figs 36–37). Figures 33–35. SEM photographs of radial sections from Tasmannia lanceolata (Poir.) A.C.Sm. (Carlquist 8212, RSA). Fig. 33. Section showing pith and early protoxylem at left; scalariform pitting on tracheids extends from primary xylem into secondary xylem; the scalariform condition yields to opposite and circular pitting in the tracheids at far right; the circular pattern continues into the remainder of the secondary xylem. Fig. 34. Reticulate pit membranes on scalariform pits of tracheid end wall; pit membranes are not visible above the pit borders because they adhere to the pit borders. Fig. 35. Pit membrane of lateral wall of scalariform tracheid; no pores are present (small tear at left is an artefact). Figures 36–37. Cinnamomum camphora (L.) J.Presl (cultivated, Claremont, California). Fig. 36. Tangential section; all rays shown are biseriate or multiseriate; enlarged cells (top, near centre) are ethereal oil cells of axial parenchyma. Fig. 37. Radial section; all ray cells are procumbent. Enlarged cells, top, are ethereal oil cells in axial parenchyma. Scale bars, Figure 33, 60 µm; Figures 34–35, scale shown in Figure 34, 10 µm; Figures 36–37, scale as in Figure 6.

Figure 38–41.

Wood sections, showing ray conditions, of eudicots. Fig. 38. Transection of stem of Tropaeolum majus L. (cultivated in Santa Barbara, California). Wide rays, consisting of non-lignified cells, extend without changes from near pith (bottom) to cambium (top). Fig. 39. Artemisia abrotanoides Jacq. ex DC. (L. Anderson 2889, RSA). Transection; there are no rays in the first-formed secondary xylem; origins of rays in later-formed secondary xylem are indicated by arrows. Figures 40–41. Lasthenia macrantha Greene ssp. macrantha (Ornduff 4140, UC), a short-lived perennial. Fig. 40. Transection, showing pith at left, cambium and cortex at right; no rays are present in secondary xylem. Fig. 41. Tangential section, showing rayless condition. Scale bars, Figure 38, 100 µm. Figures 39–41, scale as in Figure 6.

Figure 42–45.

Sections of wood of eudicots. Figures 42–43. Pipturus albidus A.Gray ex H.Mann (USw-15340). Fig. 42. Tangential section; rays are mostly multiseriate, difficult to discern because ray cells resemble the septate fibres comprising the background of the wood. Fig. 43. Radial section of multiseriate ray; ray cells are mostly upright, only a few square cells are present. Figures 44–45. Ulmus thomasii Sarg. (collected near Ripon, Wisconsin). Fig. 44. Tangential section.; rays are multiseriate; note small size of ray cells. Fig. 45. Radial section of multiseriate ray; all ray cells are markedly procumbent. Figures 42–45, scale as in Figure 6.

Figure 46–50.

Wood sections. Fig. 46. Mammillaria hemisphaerica Engelm. (A. Gibson 560, RSA). Tangential section of secondary xylem; storied vascular tracheids have helical thickenings. Fig. 47. Ceratiola ericoides Michx. (Carlquist 8037, RSA). Tangential section of wood; almost all rays are uniseriate and composed of upright cells. Fig. 48. Aristolochia sipho L'Hér. (Carlquist 8059, RSA). Transection of outer wood and adjacent secondary phloem; wide ray (which extends to pith, not shown) below; abrupt origin of a wide ray, above. Figures 49–50. Schisandra sphenanthera (Carlquist 8051, RSA). Vessel elements from radial section. Fig. 49. Scalariform perforation plate (bars intact because paraffin sectioning was employed) from a vessel close to the pith. Fig. 50. Simple perforation plate from a vessel a little farther from pith, same section as in Figure 49. Figures 46, 49, 50, scale as in Figure 21; Figures 47, 48 scale as in Figure 6.

Figure 1.

Graph showing changes in vessel element length from pith (at left). The curve for Liriodendron tulipifera L. of Magnoliaceae (from Bailey & Tupper, 1918) represents a ‘typical tree’ curve. Tetrameles nudiflora R.Br. of Datiscaceae (from Davidson, 1976) has no vessel element change, as often found in woody plants with storied cambia. The curves for Cistanthe guadalupensis (Dudley) Carolin ex Hershk. (Portulacaceae; from Carlquist, 1962) and Delissea undulata Gaud. (Campanulaceae; from Carlquist, 1975) show decrease in vessel element length outward from the pith and illustrate juvenilism. Delissea undulata is a Hawaiian rosette tree; Cistanthe guadalupensis is a small stem and leaf succulent shrub from Guadalupe Island, Mexico.

Figure 2.

Comparisons based on a spectrum from juvenile (extreme left) to adult (extreme right). Placements of features or groups within this spectrum are intended to be generalizations related to concepts discussed in the text. The graphs shown under ‘Vessel element length’ relate to curves shown in Figure 1.

Figure 3.

A scheme showing the ray types of Kribs (1935) as well as ‘paedomorphic’ types and their phylogenetic relationships (modified slightly from Carlquist, 1988a). This scheme reflects the concept, prevalent until recent times, that characteristics found in ‘typically woody’ plants are ancestral ones in angiosperms. Thus, heterogeneous type I here is indicated as the ancestral type, following Kribs (1935).

Figure 4.

A re-interpretation of the scheme of Figure 4, indicating that the ancestral ray type of early sympodial basal angiosperms may have been paedomorphic (= juvenilistic). From such juvenilistic rays, heterogenous type I may have been derived in trees. Heterogeneous type I remains a central plexus from which other types of non-paedomorphic ray types may have been derived. Juvenilistic rays may thus have occurred in ancestral non-tree angiosperms, but they also occur in angioperms that are secondarily woody.

Figure 5.

A simplified phylogenetic tree of angiosperms, based on Moore et al. (2007), and typical of recent phylogenetic hypotheses based on molecular data. Branches of this tree have been subdivided by broken lines in order to indicate the probable ancestral status of sympodial types from which woodier lineages have been derived. Lineages with cambial loss (monocots, Nymphaeaceae) have also been derived from such an ancestral type. Within eudicots, many shifts in degree of woodiness are indicated by arrows.

An attempt is made in the present paper to show modes of xylem structure that do not bear any evident relationship to xylem characters heterochrony. The phenomena of paedomorphosis (and those of accelerated vegetative adulthood) cannot be appreciated unless their limits and variations can be demonstrated. The first introduction of this phenomenon was, as seen in retrospect, a limited one. Perhaps this may explain why acceptance of the idea of xylem paedomorphy was rarely appreciated. Only instances thought clearly to represent paedomorphy were included in the original exposition (Carlquist, 1962). Many studies have affirmed the concept of protracted juvenilism or paedomorphosis (see Carlquist, 1980; Lens et al., 2009), although, at first, there was some misinterpretation of the idea. Bierhorst & Zamora (1965) considered paedomorphosis to be merely a restatement of Bailey's (1944a) concept that primary xylem is a refugium of primitive characters. The reverse is true: in paedomorphosis, juvenile characters are extended into the secondary xylem as they are, for example, in Valerianaceae (Carlquist, 1983a) and lobelioid Campanulaceae (Carlquist, 1969). Olson (2007) and Olson & Rosell (2006) have presented some particularly elegant examples in Moringa Adans. (Moringaceae), in which growth forms and xylem of species with relatively small, succulent habits correspond to juvenile stages in form, anatomy and development of species that are larger and tree-like. In the ‘refugium’ concept, juvenile characters are not expressed in the secondary xylem, only in the primary xylem.

Paedomorphy in xylem is not expressed as static adaptations. Rather, the fact that angiosperms have so readily shifted growth forms and degree and types of woodiness is a key to the importance of paedomorphy and its role in angiosperm success. Families that have closely related taxa exhibiting a wide range of growth forms (annuals to monopodial trees, e.g. in Asteraceae, Fabaceae, Polygonaceae, Violaceae) should have alerted us that there are remarkable mechanisms for changing what the cambium does. DNA phylogenetic results (e.g. Soltis et al., 2000) have forced us to realize that close relationships probably exist between plants of remarkably diverse habits and anatomy. Examples of these include closely related genus pairs (closeness unsuspected in earlier classifications) such as Akania Hook.f. (Akaniaceae; forest tree) and Tropaeolum L. (Tropaeolaceae; herb with annual stems); Gunnera L. (Gunneraceae; herb with no cambial activity in bundles) and Myrothamnus Welw. (Myrothamnaceae; small woody shrub); and Begonia L. (Begoniaceae; sympodial herb) and Tetrameles R.Br. (Tetramelaceae; monopodial tree). Xylem heterochrony explains most of the differences in habit among these closely related pairs. Xylem heterochrony is basic to the ability to shift degree of woodiness as well as kinds of woodiness.

If xylem heterochrony explains much about diversification in habit within eudicots, could it have a bearing on origin of angiosperms? Although comparisons of flower structure have become a search for relictual conditions, wood is comparatively poor in historical markers; instead, the xylary designs we see today represent optimal structural configurations that tell us much more about efficiency and modes of conduction rather than ancient conditions. There are only a few exceptions to which one can point. For example, the loss of cambial activity that led to monocots is an event indelibly printed in all monocot bundles. Long perforation plates with numerous bars and pit membrane remnants in the perforations can be read as relictual with reasonable assurance because such vessel elements are not present in dominant woody angiosperm groups and only in groups that represent early branches from angiosperm clades (when compared with phylogenies such as those of Soltis et al. (2000). Such vessel elements are related to mesic ecology (Carlquist, 1975) and the groups that contain such vessels may have had long histories of occupation of mesic habitats. Xylary patterns in the basal angiosperms present a story readable in phylogenetic terms. The early-diverging angiosperms share a sympodial construction (often associated with adventitious roots), expressed in diverse ways. Sympodiality has yielded to monopodiality in only a few families in these groups, such as Magnoliaceae and Lauraceae. Paedomorphosis is a mechanism that may account for achievement of this diversity of sympodiality in basal angiosperms. The relatively few escapes from sympodiality in that group can be understood in terms of heterochronic evolutionary change.

Discussions of origin of angiosperms prior to the advent of molecular phylogenies focused on two rather polar extremes as angiosperm archetypes: for most of the 20th century, certain woody (usually arboreal) groups (e.g. Takhtajan, 1987) were believed ancestral. More recently, ‘herbs’, not categorized as to size or other vegetative characteristics, have been claimed to be basal (Taylor & Hickey, 1992). DNA-based phylogenetic trees have shown that both of these extreme scenarios are incomplete. Basal angiosperms with enormous diversity in growth form and anatomical structure are evident in early-diverging clades of the molecular trees. That diversity and its themes should have caught our notice, but the appealing simplicity of working with data matrices based on binary definitions became a preoccupation during the era in which cladistics was dealing with morphological characters rather than DNA information. The perceptions of the present paper would not have been possible without the reinforcement and inspiration of DNA-based phylogenetic trees. The reader should note at this point that the use of categories such as herb, shrub and tree, or woody and herbaceous, can be misleading, even if there is no alternative to the use of these terms. By accepting these terms, we tend to believe that the categories are more useful than they are. In fact, there are many types of herbs and woodiness and they confuse at least as often as they help perceptions [Carica L. (Caricaceae), the papaya, has no fibres at all in its wood, only vessels and parenchyma, and thereby must ironically be called a non-woody tree if traditional definitions are to be used].

Attention has naturally focused on fossils of the earliest known angiosperms. Although interest in floral features is acute, there have been reports of vegetative structures. Angiosperms apparently were limited in secondary growth prior to Campanian time (Wheeler, Lee & Matten, 1987; Wing & Tiffney, 1987a, b; Herendeen, 1991; Wheeler & Herendeen, 1993). Early angiosperms have been thought to occupy stream and lake margins (Doyle & Hickey, 1976; Doyle, 1978; Upchurch & Wolfe, 1987; Hickey & Taylor, 1992). No author says that secondary xylem is absent from the earliest fossil angiosperms. The term ‘limited secondary growth’ (Taylor & Hickey, 1992) is a typical assessment. We lack descriptions of wood histology of the earliest angiosperms. If we look at early-diverging angiosperms today in terms of heterochrony, we see diversity in xylem and that in itself may be instructive. However, when viewed in the context of heterochrony, early-diverging angiosperms show underlying unifying patterns. These patterns, and the role they play throughout angiosperms, form the central theme of this paper. In order to clarify the picture presented, anatomical expressions of xylem that do not play a role in heterochrony (e.g. vessel characters related to xeromorphy or mesomorphy) are explicitly mentioned.

Little has hitherto been published on heterochrony in xylem of plants at large, perhaps because those familiar with systematics and phylogeny in a global way are not conversant with details of xylary anatomy. A survey of heterochrony in plants (Li & Johnston, 2000) shows a preoccupation with floral rather than vegetative features. The essay by Takhtajan (1976) is similarly oriented toward floral phenomena and did not have the benefit of data sets or molecular phylogenetics.

TERMINOLOGY AND MATERIAL

As noted by Olson (2007), heterochrony in plants differs from that in animals because plants do not have a developmental endpoint. In animals, timing of sexual maturity is a criterion to which somatic development is compared (among members of a clade or lineage) and such a comparison in plants is different from that in animals for designating instances of heterochrony. I adopted the term ‘paedomorphosis’ for juvenilism in wood and that term is also used here. I follow Olson (2007) by adopting also the all-inclusive term, heterochrony, and then indicating points along a continuum from permanent vegetative juvenilism (cambial cessation) to accelerated vegetative adulthood (mature pattern of secondary xylem attained quickly), as indicated in Figure 2. The reader will note that only vegetative characters are discussed here, so that, for example, floral parts of a plant and timing of flowering onset are not discussed. Paedomorphosis (paedomorphy) is used here for protracted vegetative juvenilism. Progenesis is used here to denote accelerated appearance of adult vegetative patterns. Because plants have an open system of growth as opposed to the closed system in animals, longer duration of apical growth is here correlated with adulthood. Thus, the short-lived primary root of a monocot becomes an indication of paedomorphy, as does the complete absence of cambium in monocot bundles (possibly almost complete in bundles of a few monocots). Attempts to cross-correlate these terms with usages in zoology should not be made because the criterion of sexual expression and the nature of the growth systems are different. Wood anatomy terminology is according to Carlquist (1988a).

Collections and authors of binomials are cited in captions (or in the text, if not illustrated). Expanded collection data and preparation methods can be found in my earlier papers related to the plant families to which the species illustrated belong. Field observations on plant size, form and ecology are original unless otherwise stated.

THE CONCEPT AND CRITERIA

1. Juvenilism in vessel element length

My earliest ideas about heterochrony in woods (Carlquist, 1962) came from noticing the age-on-length graphs for tracheary elements offered by Bailey & Tupper (1918). The curve for vessel elements of Liriodendron tulipifera L. (Magnoliaceae) from that source is included in a graph here (Fig. 1). My work in wood anatomy has included studies of secondarily woody plants and I discovered that the wood of these did not follow the curves given by Bailey & Tupper (1918). I concentrated on lengths of vessel elements because vessel elements are an accurate reflection of fusiform cambial initial length in any given stem or root. Vessel elements undergo little intrusive growth as they mature from fusiform cambial initial derivatives. The imperforate tracheary elements they accompany elongate to various degrees, ranging from very little to as much as 9.5 times the length of a fusiform cambial initial from which they were derived, depending on the species (Bailey & Tupper, 1918). I repeatedly found that, instead of the rise in vessel element length at the beginning of secondary growth as in the woody species studied by Bailey & Tupper (1918), there was shortening or no change during secondary growth. Two such curves, one for Cistanthe guadalupensis (Dudley) Carolin ex Hershk. (Portulacaceae; = Talinum guadalupensis Dudley) (from Carlquist, 1962) a stem succulent, the other for Delissea undulata Gaud. (from Carlquist, 1975: 219), a rosette tree, are included in Figure 1. Such a curve was also observed in the careful study by Cumbie (1963) for Hibiscus lasiocarpus A.Gray. Also included in Figure 1 is a curve for a eudicot tree with storied wood, Tetrameles nudiflora R.Br. (from Davidson, 1976). The initial decrease in vessel element length in primary xylem is characteristic of all species studied (Bailey & Tupper, 1918; Bailey, 1944a; Bierhorst & Zamora, 1965). We know the patterns of vessel element length change with age for stems of only a small number of species, enough to validate the idea of wood juvenilism in plants with various growth forms other than monopodial woody trees and to confirm the existence of accelerated adulthood in ‘truly woody’ species such as those studied by Bailey & Tupper (1918). Various curves probably exist in eudicots with non-arboreal woody habits. Flattish age-on-length plots have been commonly demonstrated for plants with storied cambia, but also occur in some species with non-storied cambia (Cumbie, 1963).

2. Degree of cambial activity

How many xylem cells are produced by the cambium in a given species is an easily used criterion of juvenilism vs. maturity in angiosperms. The nature of cambial activity in non-arborescent species has been least studied in annuals and short-lived perennials. One common assumption is that cambial activity is markedly less in annuals and that ‘herb’ can be used as a synonym for a minimal degree of cambial activity and can be contrasted with ‘woody’ as a character state. The study of numerous eudicot species by Krumbiegel & Kästner (1993) shows that few annuals lack cambial activity in fascicular areas, although production of secondary xylem by cambial activity in fascicular areas may not be abundant. Development of cambia in interfascicular areas does occur in some annuals, but is less common and, where present, less extensive than in the fascicular areas. Cambial activity is evident even in those fascicular areas that are sheathed by sclerenchyma.

One can suppose that the lower the amount of xylem produced by a cambium, the more likely a juvenilistic xylem pattern is present. However, few species with such limited wood accumulation have been studied. In fact, non-arborescent habits characterize some of the large families of eudicots, including the largest, Asteraceae.

Special attention should be focused on the idea that addition of large numbers of secondary xylem cells by highly active and ontogenetically changing cambia constitutes an opposite form of heterochrony. In fact, if one considers eudicots at large, ‘truly’ woody species may be in the minority, so that the monopodial arborescent habit in eudicots should not be considered the typical condition. Therefore, adult patterns in cambia are just part of a continuum, the other end of which is represented by species with little or no cambial activity.

3. Cessation of cambial activity and organographic heterochrony

The idea that angiosperms that have lost cambial activity altogether are permanently juvenile (e.g. monocots, Nymphaeales) is novel and may seem unexpected (aside from the often-cited loss of cambial activity in a few aquatics such as Ceratophyllum or Myriophyllum). This idea does, however, lead to an even more important concept: the continuity of vessels or tracheid lineages from roots into stems is broken. Roots are adventitious and growth is sympodial in vascular plants without cambial activity. Potential disabling of stem vessels by spread into them of air from embolized root xylem is not a problem in non-cambial angiosperms. Thus, the histological nature, and the degree of juvenilism, in stem xylem can differ from that in roots. For example, this permits roots of a species to function as short-term water-collecting organs, while the stems and leaves can function as water-storing organs (bulbs). Although monocotyledons are the largest category of plants with cambial cessation, there are others (Gunneraceae, Nelumbonaceae, Nymphaeaceae, etc.). Discontinuity between root and stem conductive systems in angiosperms without cambia, rather than a limitation, leads to a large number of possibilities (e.g. epiphytes).

4. Change in ray histology

Since the work of Kribs (1935) on ray histology in relation to evolution, the diversity of ray types has been evident. The most important theme of the heterogeneous and homogeneous ray types (Fig. 3) proposed by Kribs is that there has been a trend toward a higher proportion of procumbent cells, especially in multiseriate rays, during evolution in woody angiosperms. Functional reasons for this have not received extensive comment, but they have to do with radial conduction of solutes (Braun, 1970; Carlquist, 2007a). In ‘ordinary’ woody angiosperms, upright cells can be found in rays of the earliest formed secondary xylem, but those of more mature wood contain a higher proportion of procumbent cells (Barghoorn, 1941a). During my studies of secondarily woody angiosperms, especially those on islands, I found numerous instances of multiseriate rays that were composed wholly or predominantly of upright cells. In secondarily woody species, upright cells predominate or are exclusively present, even in relatively large stems, such as those of the Hawaiian lobelioids (Carlquist, 1969). Although such rays were not mentioned in the scheme of Kribs (1935), I added them to a scheme of rays (Carlquist, 1961). Multiseriate rays consisting wholly or predominantly of upright cells were interpreted as paedomorphic and incorporated into my theory of paedomorphosis (Carlquist, 1962). Such rays are considered indicators of juvenilism (paedomorphy) here, but with the additional concept that there is no decisive line between paedomorphic rays (all cells upright) and progenetic rays (all cells procumbent): all are points along a spectrum of heterochrony (Figs 2, 4). Changes in degree of ray maturity within a plant during ontogeny of the cambium are clearly possible and must be taken into account. The scheme of Kribs (1935) essentially did not take these into account, however, and sought to classify rays as seen in ‘mature’ woods. Kribs included only definitively woody plants in his scheme (a bias found also in the work of Braun, 1970). Study of non-tree species was limited at best; lianas and ‘woody herbs’ were among the types not considered at all. The information presented by Kribs (1935) and Braun (1970) thus deliberately avoided important phylogenetic and ontogenetic views of angiosperm rays. The probable evolution of ray types has been revised here. Figure 4 is a recasting of ray types taking into account heterochrony in rays. There is a tendency for workers in wood anatomy to think in terms of a named ray type characterizing a particular species. Although that is an appealing initial approach for those unfamiliar with wood anatomy, the failure to include ontogenetic dimensions and juvenilistic types of rays leads to an inability to understand ray diversity and even how rays function. Note should be taken that if procumbent cells are present in rays of a particular species, they tend to be in the central portions of multiseriate rays. Both radial and tangential sections of rays are shown here in photomicrographs for a number of species. The photographs of radial sections are selected to include the central portions of rays; procumbent cells, if any are present, are thus depicted. Once one has seen procumbent and upright ray cells in a radial section of a ray, one can correlate the vertical height of those types of cells with cells of rays as seen in tangential sections. Tangential sections thus permit one to guess with fair accuracy which cells may be upright, which procumbent, but radial sections remain the basis for definition of those cell types. Ray cells that are square as seen in radial section are traditionally grouped with upright ray cells rather than procumbent cells in defining whether a ray is ‘heterogeneous’/‘heterocellular’ (containing both upright and procumbent cells) as opposed to ‘homogeneous’/‘homocellular’ (containing only procumbent cells).

5. Continuation of primary ray patterns into the secondary xylem

In some angiosperms with secondary growth, wide primary rays are present between bundles of the primary stem. In those species that develop marked woodiness, such rays are rapidly dissected into smaller rays by changes in the vascular cambium with the onset of secondary growth (Barghoorn, 1941a). However, in less woody dicots, the cambium in interfascicular areas may merely continue a ray and not subdivide the ray to any marked degree. Lack of change in such ray areas (‘rays wide and high’) seems obviously to represent a form of heterochrony and was included as an example of juvenilism earlier (Carlquist, 1962). The elliptical spaces one sees in skeletons of woody cacti are one example. The rays of Aristolochia L. and other lianas (Fig. 48) are also good examples. There are notably wide rays in wood of some eudicots that do not result from extension of primary rays, however.

6. Raylessness

Raylessness was not included as one of the criteria for juvenilism earlier (Carlquist, 1962), although I considered it a criterion later (Carlquist, 1970). Barghoorn (1941b) showed that raylessness results from substitution of fusiform fibres for ray cells in the potential ray areas of certain eudicot species. This happens early in ontogeny, so one may well ask, how can this be an expression of juvenilism? It happens only in a small number of eudicot species and in some of these rays eventually form. The fact that ray formation is delayed is thus the expression of juvenilism. The commonly cultivated species of Veronica L. formerly treated in the genus Hebe Comm. ex Juss. (Albach et al., 2004) have rayless wood (Meylan & Butterfield, 1978).

7. Retention of primary xylem patterns in lateral walls of vessels

This was used as a criterion for juvenilism earlier (Carlquist, 1962). In fact, the instances cited there were probably not indicative of heterochrony, but rather of laterally wide pits, resembling scalariform lateral wall pitting on vessels of succulent plants. Helical thickenings, such as those seen on protoxylem tracheids and vessels, can occur in secondary xylem of a few eudicots, as described below. These constitute a genuine expression of juvenilism. There is a tendency for lateral wall pits on vessels to be laterally wide, often with large pit apertures, in succulents (e.g. Crassulaceae). In Crassulaceae, this tendency is best considered as related to water economy of the plant rather than juvenilism, as demonstrated by presence of large parenchyma cells adjacent to vessels.

8. Primary xylem as arefugiumfor ancestral conditions

Bierhorst & Zamora (1965) claimed that the theory of paedomorphosis (Carlquist, 1962) was, in fact, a restatement of Bailey's (1944a) refugium concept. Bailey's idea was that specializations occurred first in the secondary xylem in woody dicotyledon stems and roots and then spread to the primary xylem. Although one sometimes sees this in angiosperms, it is a valid form of heterochrony. In fact, it is the opposite of juvenilism: the refugium idea represents accelerated onset of adult characters. Instead of primitive features persisting from the primary xylem into the secondary xylem, these features give way in the secondary xylem of a given species to derived characters.

9. Independence of criteria for juvenilism or adulthood in woods

There is a tendency for associative thinking in wood evolution (e.g. ancestral ray conditions and primitive perforation plate configurations should coexist in a given species). Such character associations can be demonstrated as statistical generalizations, but a number of wood characters do not seem linked. This is true also with characters on the juvenile–adult spectrum.

10. Excluded anatomical conditions

Long-term vessel element length trends

Bailey & Tupper (1918) envisioned a gradual shortening of tracheary elements during the course of evolution. The data they supplied do suggest that this is true as a generalization, but the explanation of what drives this shortening is ecological, related to adaptation to cold and xeric conditions (Carlquist, 1966, 1975; Carlquist & Hoekman, 1985). This trend holds true independently in various lineages of seed plants. There need be no confusion between such phyletic shortening and heterochronic changes in vessel element length. One may think of paedomorphy or progenesis as overlays on the global patterns of shortening in fusiform cambial initials. Vessel element length changes related to heterochrony can and should be demonstrated within a plant by the age-by-length curves. Because vessel element length tends to increase during secondary activity in typically woody plants, as a result of changes in length of fusiform cambial initials, earlier-formed wood (e.g. Liriodendron tulipera in Fig. 1) will have shorter vessel elements than more recently formed wood. The gradual decrease during evolution in fusiform cambial initial length (and therefore vessel element length) noted by Bailey and Tupper for vascular plants as a whole were based on comparisons among species, using what they presumed to be mature wood patterns (the most recently formed wood in large stems). In angiosperms, they did not sample species that show paedomorphy in secondary xylem; species that were not ‘typically woody’ were omitted from their study.

Other vessel features related to ecology

Changes in vessel diameter, density and grouping within a wood may be characteristic of particular woods, but they are related to ecological and seasonal factors, not to heterochrony (Carlquist, 1966, 1975, 1984). Vessel diameter may increase over time within a wood sample (e.g. Carlquist & Grant, 2005), but this is related to a larger leaf crown and a larger transpirational demand as a tree grows. In fact, vessel diameter can also decrease as a shrub ages (e.g. Artemisia tridentata Nutt., S. Carlquist, unpubl. data), reflecting adverse change in water availability.

Imperforate tracheary element types

On a phylogenetic basis, tracheids (sometimes termed ‘fibres with fully bordered pits’) have been thought to precede fibre-tracheids and libriform fibres, the three stages in specialization of imperforate tracheary elements in angiosperm wood (Metcalfe & Chalk., 1950, 1982; Carlquist, 1988a). Although there may be a few exceptions to this scenario, the patterns of occurrence of these three types of imperforate tracheary elements does not seem to bear any relationship to heterochrony in xylem.

Storying

Storied woods as a whole tend to have shorter fusiform cambial initials and shorter vessel elements than non-storied woods (Bailey, 1923; Carlquist, 1988a). Thus, when plotted in age-by-length curves (Fig. 1, Tetrameles nudiflora), woods with storying have flat lines, not changing much from metaxylem outwards as compared with woods that have non-storied cambia. Some woods with storied vessels (and associated axial parenchyma strands) may not have storied fibres (e.g. Asteraceae, notably Senecioneae; Carlquist, 1966), but that is because of intrusive growth of fibres, which therefore do not exhibit the storied pattern of the fusiform cambial initials from which they are derived. Delayed onset of storying ontogenetically is a theoretical instance of heterochrony, but, in fact, rapid onset of the storied condition during secondary growth has been reported instead (Cumbie, 1963, 1967; Davidson, 1976).

Axial parenchyma

Axial parenchyma contains starch and conducts sugar, depending on the physiological activity of the wood (Braun, 1970, 1983; Sauter, Iten & Zimmermann., 1973). This physiological activity is linked to photosynthate conduction in rays (Carlquist, 1988a; Sauter & van Cleve, 1989). However, the two kinds of parenchyma are not directly linked in evolutionary specialization (Kribs, 1935, 1937), but show changes in patterns independently. Axial parenchyma does not show patterns of heterochronic change in histology, although rays do.

Successive cambia

Successive cambia arise not from a vascular cambium, but from a master cambium produced by tangential divisions in cortical parenchyma cells (Carlquist, 2007b). The master cambium produces vascular cambia. Thus, longevity in each of the vascular cambia is finite. Although ontogenetic changes could theoretically occur within each of the vascular cambia in a stem (or root) with successive cambia, thus far no such phenomenon has been shown. For example, in a stem with successive cambia, there is no change in length of vessel elements over time.

EXAMPLES AND DESCRIPTIONS OF HETEROCHRONY

Examples are shown here in a phylogenetic context, related to the schematic phylogenetic tree shown in Figure 5. This is a way of demonstrating the range of heterochrony within basal angiosperms and more briefly eudicots. Juvenilism and its frequent interfaces with sympodial structure in the basal angiosperms is thus explored in a detailed manner. The phylogenetic tree in Figure 5 is based on Moore et al. (2007). On it have been superimposed patterns that indicate kinds of growth forms. This drawing can be used as a key to the concepts below.

1. Early-diverging angiosperms: variations on sympodial structures

Amborella Baill. (Figs 6, 7)

Amborella trichopoda Baill. is not a tree. On the Plateau de Dogny in New Caledonia, older individuals are multi-stemmed and produce new stems from the base. These new stems have secondary growth and feature woody cylinders up to 6 cm in diameter. Most stems are smaller and Amborella can be seen as a scrambling undergrowth shrub. Feild & Arens (2005) have figured a somewhat tuberous root bearing several young stems (probably a young plant). Amborella is clearly sympodial, whether in the wild or in cultivation (University of California at Santa Cruz Arboretum).

Two things are evident in the photomicrographs of stem secondary xylem. Firstly, uniseriate rays are exceptionally abundant compared with multiseriate rays (Fig. 6). This suggests paedomorphic type I rays, with a near approach to paedomorphic type III rays (Fig. 3). Secondly, radial sections of multiseriate rays show mostly upright cells (Fig. 7). The ray cells of uniseriate rays are all upright cells. Thus, wood of Amborella is paedomorphic and does not correspond to any of the heterogeneous or homogeneous ray types that Kribs (1935) intended to describe woods of all angiosperms. Wood sections from the largest collected stems of Amborella still show juvenile ray patterns.

Aristolochiales: Thottea siliquosa (Lamk.) Ding Hou

Thottea Rottb. (including Apama Lam.) is a genus of Aristolochiaceae that grows as sympodial clusters of cane-like stems, similar to the habit of a ginger in general appearance. Each stem has a finite duration; inflorescences are lateral. In transection, the stems have fascicular areas of secondary xylem separated by wide rays that are extensions of the primary rays of the stem (Fig. 8). The vascular cambium in both fascicular and interfascicular areas changes little over time. Thus, there is no change in ray number and little in ray width during secondary growth. In tangential section (Fig. 9), one sees that the rays of Thottea siliquosa are tall (none included within the limits of the photograph) and wide. More significantly, the cells are markedly upright. Although upright cells are best seen in a radial section, one can see in the tangential section that the ray cells are about at least five times taller than wide and the edges of the ray are indistinct because the fibres resemble the ray cells in shape. As seen in radial section, the ray cells of T. siliquosa are definitely all upright, corresponding to paedomorphic type II (Fig. 3). No uniseriate rays are present.

Austrobaileyales: Illicium L.

A tangential section of wood from a small tree of Illicium ridleyanum A.C.Sm. (Fig. 10) shows rays 1–3 cells wide, with uniseriate tips prominent on multiseriate rays. As seen in tangential section, most ray cells seem to be upright cells and the radial section (Fig. 11) confirms this: upright cells and square cells but no procumbent cells are present in the radial section. The wood from a mature tree of I. anisatum L. (Figs 12, 13), however, is different. As in the section of I. ridleyanum, rays are 1–3 cells wide in I. anisatum (Fig. 12). Most ray cells in I. anisatum appear circular in shape in the tangential section. Such ray cells prove to be procumbent in a radial section of a multiseriate ray (Fig. 13, below); multiseriate rays have upright ray cells only at tips. Uniseriate rays (Fig. 13, above) are composed of square to upright cells). Thus, wood from a mature tree of I. anisatum shows a mature pattern of ray cells, corresponding to the heterogeneous type I of Kribs (Fig. 3), whereas the wood of a clearly woody but relatively small stem of I. ridleyanum (corresponding to paedomorphic type I on Fig. 3) remains juvenile for a period of years.

Austrobaileyales: Austrobaileya C.T.White

Austrobaileya is a liana in which rays (Fig. 14) are tall, wide and multiseriate or short and uniseriate/biseriate. The latter are composed of upright ray cells. The wide multiseriate rays, however, rapidly change during secondary growth from strictly upright ray cells (Fig. 15, left) to procumbent ray cells only (Fig. 16, right). Thus, an adult pattern is rapidly achieved. As seen in a transection of the central portion of a stem (Fig. 16; pith is at left), the wide multiseriate rays of Austrobaileya have their origin in the primary stem and continue into the secondary xylem. This pattern is common in lianas.

A similar transection of a stem of Schisandra sphenanthera Rehder & E.H.Wilson (Fig. 17) shows a different pattern: the rays in wood close to the pith all appear to be uniseriate. This is also true in Illicium ridleyanum (Fig. 18). Schisandra sphenanthera does develop wide multiseriate rays like those of Austrobaileya as stem diameter increases (Carlquist, 1999). These comparisons are offered to illustrate that not all lianas begin with wide rays often thought typical of lianas. The ray patterns of Kadsura Kaempf. ex Juss. and Schisandra Michx. begin like those of Illicium, illustrating thereby a paedomorphosis like that of shrubs. Indeed, Kadsura and Schisandra can be regarded as scrambling shrubs that eventually become lianas (Smith, 1947; S. Carlquist, pers. observ.). Trimeniaceae are not described here because the patterns in Trimenia Seem. resemble those of Illicium and features of Piptocalyx Oliv. ex Benth. are similar to those of Schisandraceae.

Chloranthales: Chloranthaceae

Sarcandra Gardn. (Figs 19–23) was thought by Bailey & Swamy (1950) to be vesselless. When one views a stem transection with secondary xylem (Fig. 19), one can see why: the fascicular xylem is homogeneous, apparently composed only of tracheids. Rays are multiseriate and uniseriate, but both types of rays are composed only of upright cells (Fig. 20). This is true in the root, too. Wide multiseriate rays in Chloranthaceae are common and are extensions into the secondary xylem of the primary ray pattern (Carlquist, 1992a, b). Secondary xylem of Sarcandra is thus clearly juvenilistic and corresponds to paedomorphic type I (Fig. 3). In the roots of Sarcandra glabra (Thunb.) Nakai, a few vessels were located in secondary xylem. Three are shown in Figure 21 in transectional view (arrows). These appear as pairs because the overlap area is sectioned. A perforation plate of a vessel is shown in face view in Figure 21; the perforations of the perforation plate are scalariform, much like end-wall pitting on a tracheid, but slightly wider laterally and vertically. The perforations often contain thread-like pit membrane remnants (Fig. 23), a feature found in less specialized vessels (Carlquist, 1987, 1992c). Thus, Sarcandra does have vessels, but probably only a few and only in roots with secondary growth. Takahashi (1988) claimed that vessels occur in metaxylem of stems of Sarcandra, but not in secondary xylem. I have been unable to confirm this report, which contradicts what is found in other Chloranthaceae (tracheids in primary xylem, vessels in secondary xylem). Hedyosmum Sw. has notably wide, tall multiseriate rays in addition to a few uniseriate or biseriate rays (Fig. 24). These rays begin during primary growth and widen centripetally; they are rarely dissected during secondary growth. All ray cells in all rays are square to upright, sometimes markedly upright (Fig. 25). Some perforation plates in Hedyosmum (Fig. 26) are not only long, they tend to have intact or semi-intact membranes where the perforations are in transition to the intervascular pitting in the overlap area (Fig. 26, top). The vessels of Ascarina J.R.Forst & G.Forst. (Figs 27, 28) are similar to those of Sarcandra, but with long scalariform plates (Fig. 27). In Ascarina, vessels occur in both roots and stems (Carlquist, 1992b). One peculiarity of perforations in Chloranthaceae is that pit-membrane remnants are abundant, sometimes merely threads, as shown for Sarcandra (Fig. 23) or sometimes a sheet pierced with porosities of various sizes, as in Ascarina lanceolata Hook.f. (Fig. 28).

Chloranthus Sw. and Sarcandra are short, sympodial shrubs branched from the base. Ascarina and Hedyosmum tend to be larger, variously woody and definitely sympodial, with no obvious single trunk or taproot. Older plants may bear numerous shoots branching from the base and shoots produced from roots.

2. The Monocot Type of Organization: Organographic Heterochrony in Xylem

A module for monocot origin: Saururaceae

In Figures 29 and 30 are depicted transections of stems of two species of Saururaceae, a genus usually placed in Piperales (which are not considered a close relative of Chloranthaceae). Figure 29 shows Saururus cernuus L., in which collateral bundles are well spaced. In this species at least half of each bundle has resulted from secondary growth, judging from the clear radial alignment of cells in files in secondary xylem and secondary phloem. Cambial activity is much less between the vascular bundles; it is limited to a few tangential divisions in the primary ray areas (arrow). Although there are fibres at the adaxial and abaxial surfaces of the bundles in S. cernuus, the bundles are not ‘encased’ in fibres and thus can accommodate secondary growth. The bundles in Houttuynia cordata Thunb. (Fig. 30) show less cambial activity (Carlquist, Dauer & Nishimura, 1995). One can guess that, in some bundles, the vascular cambium has produced one to three layers of secondary xylem or else that the vascular cambium has ceased and that the few xylem cells in radial files are the product of tangential divisions in procambium. The number of layers of xylem (or phloem) produced by vascular cambia in H. cordata is not of any significance functionally because not enough secondary xylem is conceivably produced to add in any meaningful way to the primary xylem. The primary rays between bundles are composed of sclerenchyma and obviously no interfascicular cambial activity occurs. The fact that bundles are encased in fibres and primary rays are composed of fibres and sclereids shows that there has been a near-definitive cessation in cambial activity and that secondary tissues do not play a role in H. cordata. With development of minimal amounts of cambium, interconnection with anything like a taproot becomes unlikely and, in fact, Saururaceae are sympodial plants with adventitious roots. Similar tendencies could be cited for Piperaceae. Piperales exhibit cambial cessation, adventitious roots and sympodial plant form which are inevitably correlated with cambial cessation, all of which occur in parallel in the monocots.

Clearly there is no secondary growth in the scape bundles of a lilioid monocot, Clintonia borealis (Aiton) Raf. (Fig. 31). The encasement of bundles in sclerenchyma and conversion of primary rays to sclerenchyma are features that resemble those observed in stems of Houttuynia cordata. The vascular bundles of Yucca brevifolia Engelm. (Fig. 32) are enormous, composed mostly of tracheids. One bundle is illustrated here to take into account the idea, developed by Arber (1919), that one may find residual cambial activity in bundles of monocots. One can see a few tangential divisions between xylem and phloem of the bundle in Figure 32 (between tips of arrows), as in the examples offered by Arber (1919). Plant anatomists have not continued searching for such examples, probably because such divisions, if they represent cambial divisions at all (they could be the last divisions of procambium instead), do not produce any appreciable amount of secondary xylem or phloem. Thus, we can assert that monocotyledons have achieved essentially complete cessation of vascular cambia in bundles. This is important to realize because cambial activity is necessary to form secondary xylem and secondary phloem that are intercontinuous from stem to root, as in monopodial tree eudicots. Special secondary thickening mechanisms that produce new bundles in monocots such as Aloe L., Agave L., Cordyline Comm. ex R.Br., Dracaena Vand. ex L., etc. (all members of Asparagales) form an entirely different topic not considered here.

In monocots, the cessation of cambial activity means that vascular intercontinuity between stems and roots cannot occur (except for the short-lived seedling root, which soon ceases to function) and that the sympodial form, with adventitious roots borne on branching stems, is the basic mode of construction. Xylem and phloem cells of adventitious roots abut on xylem and phloem of stem bundles, but pit membranes separate these respective cells of root and stem and, for example, there are no perforation plates that connect root vessels with stem vessels. Perhaps the most important xylary fact that emerges from this scenario is that, in monocots, types of xylem cells and degrees of xylem development, as well as times of maturations in roots as opposed to stems, are independent from each other, whereas in a woody eudicot, secondary xylem and phloem production in root and stem are synchronized and intercontinuous. The xylary disjunction in monocot organs, which is considered organographic heterochrony here, has permitted monocots to produce a number of growth forms and enter a wide range of habitats, as discussed below.

Vascular bundles that lack cambia are not unique to monocots; they may be found in some annual eudicots as well as in Gunnera L. (Gunneraceae) and Piperaceae, for example (Metcalfe & Chalk, 1950; Wilkinson, 2000). Cambium is also absent in Nymphaeaceae (Metcalfe & Chalk, 1950) and Hydatellaceae (Carlquist & Schneider, 2009)

3. Magnoliales: monopodial xylary traits, acceleration of adult patterns

Canellales

Winteraceae are distinctive in being vesselless (Bailey, 1944b; Carlquist, 1989) and phylogenetic analyses have indicated that vessels were lost in the ancestors of Winteraceae (Young, 1981; Soltis et al., 2000, and others). This idea can be accepted, but one notes that thin and highly porous pit membranes can be found in tracheids of Winteraceae; fracturing of a large proportion of these thin membranes occurs with drying of preparations for scanning electron microscopy (SEM). Such apparent absence of membranes can be seen in preparations of winteraceous wood (Meylan & Butterfield, 1978; Carlquist, 1983b) and in Amborella (Feild et al., 2000), and such membrane absence should be considered, contrary to Feild et al. (2000), as an artefact (see Carlquist & Schneider, 2001). However, heterochrony can be found in tracheary elements of Winteraceae. Primary xylem of Tasmannia lanceolata (Poir.) A.C.Sm. (Fig. 33) shows more radial layers of scalariform tracheids than in an ordinary sequence of metaxylem (e.g. Bierhorst & Zamora, 1965) and the scalariform pattern yields eventually (= paedomorphy) to tracheids with circular bordered pits (Fig. 33, extreme right). The circular pits are formed in the remainder of the secondary xylem produced in the stem of T. lanceolata, illustrated. On what appear to be tracheid end walls of the scalariform tracheids, highly porous pit membranes may be found as shown in Figure 34 (if pit membranes have not been destroyed altogether during drying procedures). Apparent presence of the porous pit membrane only above the pit aperture is because of adherence of the remainder of the pit membrane to the pit border, a common phenomenon seen in SEM studies of tracheary elements in angiosperms (various degrees of such pit membrane adherence may be seen, depending on the species). Some intact pit membranes may be seen on what appear to be lateral walls of the scalariform tracheids (Fig. 35). Even supposing that the porous qualities of the end-wall pit membranes are exaggerated by the drying process, the thinness and vulnerability to collapse during drying of pit membranes in Winteraceae are unusual among angiosperms.

Procumbent ray cells do occur in multiseriate rays of certain trees in Winteraceae (Meylan & Butterfield, 1978), but in the family as a whole upright ray cells are remarkably common. For example, procumbent cells are lacking in rays of small cloud-forest trees of Tasmannia piperita Miers (Carlquist, 1989) and even in large multiseriate rays of Winteraceae most cells that one would have thought to be procumbent on the basis of tangential sections prove to be square when viewed in radial section (original observation). The large multiseriate rays in Winteraceae mostly begin at the pith in small-stemmed species such as Tasmannia xerophila (Parment.) M.Gray.

Laurales, Magnoliales

Multiseriate rays in Eupomatiaceae are juvenilistic, with few files of procumbent cells (Carlquist, 1992d). This is not surprising because Eupomatia laurina R.Br. is not really a tree in the ordinary sense: it branches from the base and may be considered a sympodial rather than a monopodial plant. This could be said of Annona L. and numerous genera of Monimiaceae. However, other families of Magnoliales and Laurales are clearly monopodial trees, such as most Lauraceae (excepting the parasitic vine Cassytha L.), Magnoliaceae and Myristicaceae. The wood of Cinnamomum Schaeff. (Figs 36, 37) shows adult-type rays in which all cells are procumbent. Moreover, no large rays originating at the pith are present (thus, more change in cambial organization toward an adult condition has occurred). This accords with active radial translocation of photosynthates in a trunk. Although vessels in the inner part of a monopodial tree may no longer be active in conduction, axial and ray parenchyma cells may still be (data are lacking because so few wood samples are preserved in liquid and examined for the presence of nuclei in parenchyma of various parts of a tree trunk).

4. Eudicots: examples and pervasiveness of heterochrony

Sympodiality retained: the early-diverging eudicots

Ranunculales have now been shown to be sister to the rest of the eudicot clade (e.g. Soltis et al., 2000). All families of the order are basically sympodial. Eupteleaceae, a family in this order, are woody, but Euptelea Siebold & Zucc. has the aspect of an understory tree that begins as a shrub. Its primitive vessels (Carlquist, 1992c, 1995) may have been retained because of history of occupancy of mesic habitats (Carlquist, 1975). The other families are more clearly sympodial in habit and form a coherent pattern of increasing wood specialization beginning with Lardizabalaceae (Carlquist, 1995). Lardizabalaceae are a family of vines and lianas, except for the shrubby Decaisnea Hookf. & Thomson, but lianas that branch from the base as sympodial shrubs do. Juvenilistic tendencies and secondary woodiness are common in Ranunculales. For example, a number of genera of Papaveraceae (Bocconia Plum. ex L., Dendromecon Benth.) are secondarily woody and there have been repeated shifts in degree of woodiness within Ranunculaceae and Berberidaceae, judging from comparisons of wood anatomy to a DNA-based phylogenetic reconstructions (Kim et al., 2004). The family thus has a mix of juvenile and adult patterns in wood anatomy.

Wide-ray patterns: cable construction

Key early-diverging groups of angiosperms show extension of wide primary rays unchanged into the secondary xylem (Piperaceae, Chloranthaceae, Aristolochiaceae). This could be a symplesiomorphy in early angiosperms and related to the limited secondary xylem of sympodial early-diverging angiosperms. However, it is a pattern that has occurred numerous times, probably as an apomorphy, in eudicots. As evidence, one needs only mention the numerous origins of annuals with limited accumulation of secondary xylem, such as Tropaeolum L. (Fig. 38). Although this is a common pattern, not all annuals have wide rays in secondary xylem (Krumbiegel & Kästner, 1993). Wide rays in secondary xylem, little changed from the primary rays (and thereby juvenilistic), often seem to have a mechanical significance: cable construction (segments of mechanically strong tissues embedded in a background of soft-walled parenchyma). Cable construction is widespread in eudicots; it is a term for a group of phenomena that represent distinctive adaptive modes (Carlquist, 2009). It is especially common in vines and lianas. Note should be taken that rays continuing with little alteration from pith to cambium were not included in the original concept of wood juvenilism (Carlquist, 1962), but this is added here because persistence of rays unaltered during ontogeny is by definition an example of a juvenile character.

Rayless woods: a short-term strength strategy

Rayless eudicot woods are not found in a large number of species in the aggregate, but they are cited here as examples of juvenilism because of their conspicuousness and because they are striking instances of cambial juvenilism. Raylessness represents the opposite of unaltered rays: libriform fibres are formed in potential ray areas, so that there is no appearance of ray cells, at least at first (Fig. 39); the conversion to an all-fibre formula takes place rapidly (Fig. 40), even in a short-lived perennial (Fig. 41) or annual (Phacelia Juss. Carlquist & Eckhart, 1984). Barghoorn (1941b) mentioned that raylessness is an alteration of ray ontogeny. Indeed, in some species in which wood is rayless at first, rays are eventually formed (Fig. 39), so raylessness is an expression of heterochrony, more juvenilistic in genera that stay rayless indefinitely, such as Veronica (as Hebe; Meylan & Butterfield, 1978) than in species that develop rays eventually, such as the Canarian species of Plantago L. (Carlquist, 1970).

Juvenilism in eudicot rays: a mechanism for diversification

Too often, we think that there is not much difference between wood of a shrub and that of a tree and that the difference is merely in terms of quantity of wood rather than the histology. A pair of species from two closely related families in Rosales shows that qualitative differences in wood are often heterochronic. The wood of Pipturus albidus A.Gray ex H.Mann. of Urticaceae (Figs 42, 43) is from a cloud-forest tree (‘non-herb’) c. 5 m tall, whereas the wood of Ulmus thomasii Sarg. of Ulmaceae (Fig. 44) is from a ‘typical tree’ of deciduous North Temperate forest. The rays of P. albidus are mostly 3–4 cells wide (Fig. 42), as are those of U. thomasii (Fig. 44). However, radial sections show that the ray cells of P. albidus are prominently erect, with a few files of square cells (Fig. 43), whereas the ray cells of U. thomasii are all markedly procumbent. Pipturus albidus has wood that exhibits juvenile characters, whereas the wood of U. thomasii is markedly ‘adult’ in ray structure. Is only one of these ‘heterochronic’? In my interpretation, both represent instances of heterochrony; P. albidus is essentially a secondarily woody member of the frequently herbaceous nettle family and U. thomasii is a woody member of Ulmaceae, a stereotypically woody family. Numerous such pairs could be shown as examples of different degrees of heterochrony in wood within a clade, even within a single family. Genera that show a range from herb to ‘typical’ tree include Hypericum L. (Hypericaceae), Senecio L. (Asteraceae) and Euphorbia L. (Euphorbiaceae).

Primary xylem patterns extended: advantages in succulents

Woods of succulents often have scalariform lateral wall pits on vessels, with wide ‘gaping’ pit apertures or pseudoscalariform pits. These may simply be expressions of non-xeromorphy in the xylem, related also to expansion and contraction of the secondary xylem during wet and dry periods, respectively. However, in a few succulents, primary xylem patterns of vessel secondary wall architecture are definitely extended into the secondary xylem. Those succulents that produce secondary xylem vessels that show no sign of alternate pitting, merely scalariform, may be heterochronic, as in the species comparisons of Olson & Carlquist (2001) and Olson (2007). However, the most dramatic examples of heterochrony in vessels of a succulent may be found in Cactaceae (Mauseth, 2004), in which virtually all species begin with tracheary elements that have helical thickenings. Later increments of secondary xylem may include libriform fibres and vessels with wide scalariform pits: this happens in the shrubby and arborescent cactus forms. In globular cacti, helically thickened tracheary elements may persist for the life of the plant, as in Mammillaria Haw. (Fig. 46), a juvenilism protracted indefinitely. A similar situation obtains in Anacampseros L. of Portulacaceae and Hectorella Hook.f. of Hectorellaceae, both of which families are close to Cactaceae (Carlquist, 1998).

Special ray types in eudicots

A scattering of eudicots have only uniseriate rays in which all ray cells are upright (Fig. 47; see also Fig. 2). These rays were designated as paedomorphic type III (Carlquist, 1988a), a kind of extension of Kribs's (1935) heterogeneous type III (in which procumbent or square cells are present). Paedomorphic type III rays are present in an array of eudicots (see Carlquist, 1988a) that could be called small shrubs (e.g. Empetrum L.). The explanation is apparently that such shrubs are permanently or indefinitely juvenile in wood structure, a character state derived as a form of heterochrony from species that have uniseriate rays with an abundance of upright ray cells plus multiseriate rays. In those species, uniseriate rays in earlier portions of secondary xylem lack procumbent cells and multiseriate rays often are not formed in the earlier secondary xylem (S. Carlquist, pers. observ.).

A curious ray formulation that has received little comment may be found in various angiosperms lianas: wide multiseriate rays with two types of origin in a single stem. Some rays can be traced unchanged back to the pith (Fig. 48, bottom); other multiseriate rays originate not as narrower rays that widen, but as a result of an abrupt conversion of fusiform cambial initials to ray initials (Fig. 48, above). Thus, we could say that some rays are extensions of primary rays and thus juvenilistic, whereas others represent a rapid supplanting of fusiform initials with ray initials, not a juvenilistic happening.

The refugium idea: vestiges retained briefly

The idea that specializations originate in the secondary xylem and are, in the phylesis of a lineage, extended into primary xylem was originated by Bailey (1944a). This idea can be stated in heterochronic terms by saying that juvenile characters are more likely to be found in primary xylem, whereas derived ones occur in secondary xylem in a given wood sample. This is illustrated here for Schisandra, in which vessels near the pith have scalariform perforation plates (Fig. 49), but vessels in the secondary xylem have simple perforation plates (Fig. 50). In the closely related Illicium (Schisandraceae), vessels have scalariform perforation plates both in primary xylem and in secondary xylem. Lianas as a whole are characterized by simple perforation plates in mature xylem and, although a number of them have simple perforation plates in primary xylem as well, some do not. Actinidia Lindl. (Actinidiaceae) and Tetracera L. of Dilleniaceae (Carlquist, 1988a) show long scalariform perforation plates in primary xylem but simple perforation plates or perforation plates with few bars in secondary xylem. A particularly nice example may be found in the shrub Crossosoma Nutt., in which scalariform perforation plates occur in the early secondary xylem, but only simple perforation plates occur in vessels formed after that (Carlquist, 2007c).

Bierhorst & Zamora (1965) dealt exclusively with primary xylem but were concerned with ontogenetic changes or tracheary elements within primary xylem: they reported instances of change and some instances of no change in tracheary element types during the progression from early protoxylem to late metaxylem. In numerous families, vessels with simple perforation plates are present in protoxylem and metaxylem. These families, as it happens, are also characterized by simple perforation plates in secondary xylem (e.g. Asteraceae). In other families, only tracheids were present in protoxylem and various portions of the metaxylem, but late metaxylem was composed of vessels with scalariform perforation plates. The list of these families includes Aquifoliaceae, Cornaceae, Cunoniaceae, Ericaceae, Hamamelidaceae, Lauraceae, Magnoliaceae, Pittosporaceae and Styracaceae. In all of these families, some scalariform perforation plates may be found in secondary xylem as well, although in others (Pittosporaceae and Styracaceae) there are simple perforation plates in secondary xylem, so various degrees of heterochrony are represented in secondary xylem in these families. One can think of these families with simple perforation plates (sometimes along with some scalariform perforation plates) in secondary xylem as having an advantage in their simple perforation plates and an accelerated adulthood in secondary xylem as compared with the families with scalariform perforation plates throughout the secondary xylem. Simple perforation plates are generally related to more strongly seasonal water availability (Carlquist, 1975) and these families do characteristically occupy habitats with such seasonality in moisture availability.

Bierhorst & Zamora (1965) developed considerable evidence for the primary xylem Refugium Theory, as they stated. They offered no evidence against paedomorphosis, which they confused with the Refugium Theory at that time. In instances of paedomorphosis, primary xylem features (such as upright cells in ray areas) are continued indefinitely into the secondary xylem, regardless of axis diameter.

CONCLUSIONS

Sympodial advantages and disadvantages in early-diverging angiosperms with secondary xylem

Sympodial early-diverging groups of angiosperms with cambial activity have some potential advantages.

  • 1By means of multiple stems, they can quickly (near or at ground level) spread over a wider lateral area for photosynthesis at appropriate levels into areas not too bright or too shady. Monopodial trees are faced with vertical opportunities and only the crown level can be varied.
  • 2If they grow along ground level or have rhizomes underground (Piper L., Chloranthus, Hedyosmum), branches can root, not only securing footholds over a much larger area, but also by rooting of new branches tap new sources of moisture or, conversely, escape from inundation.
  • 3By producing numerous branches that can potentially root, they can escape from failures of the conductive system in any one stem or root. Even if the base of the plant does not ramify laterally, continual production of innovations from a caudex-like root (e.g. Amborella, Chloranthus) multiplies the chances than any given branch might succeed.
  • 4By having some degree of cambial activity, upright canelike branches of limited duration (e.g. Thottea of the Aristolochiaceae, as well as various Chloranthaceae and Piperaceae) can be produced. These can access better light levels.

There is a tradeoff: a plant with a few tall stems has access to better light but is more vulnerable to failure of the conductive system. Not surprisingly, a moist understory environment, the characteristic habitat for many sympodial early-diverging angiosperms with wood, will offer the most potential for success. The multiplicity of stems (and roots) of limited duration in sympodial but woody plants correlates with a greater degree of juvenilism in the wood. This is true not merely because stems are smaller in diameter but because conduction of photosynthates will mostly be vertical: from leaves to roots, from base of plant to inflorescence. The abundance of upright ray cells not only correlates with predominantly vertical movement, but also suffices to transfer photosynthates from outer stem zones to inner stem zones.

An alternative to the production of numerous canes, as in Piperaceae and Chloranthaceae, is for any given plant to develop a multiplicity of stems, any or all of which can ascend rapidly so as to reach the photosynthetic compensation point for the plant and escape illumination limitations of the understory habitat: the liana habit (Austrobaileya, Kadsura, Schisandra, Aristolochia, Piper).

Vessel origin and loss: problems and possibilities of vessel presence

Marginal nature of the primitive vessel element

Sperry, Hacke & Wheeler (2005) and Sperry et al. (2007) provided some fascinating keys to evolution of conducting systems in angiosperms. They showed that primitive vessels have severe drawbacks in resistance of long scalariform perforation plates (which often have pit membrane remnants) to conductive flow and in the lack of safety (prevention of entry into vessels of air bubbles that would cause embolisms). The relatively poor solutions of many early angiosperms to these problems would explain why early angiosperms were arrested in the evolution of their conductive systems and why their sympodial growth forms (less risky than the monopodial tree habit) were retained for so long. With the retention of less woody sympodial growth forms, secondary xylem in early angiosperms is inevitably juvenilistic to various degrees. I am constantly reminded of the wilting of the new growth of all plants of Illicium tenuifolium (Ridl.) A.C.Sm. on an unusually warm afternoon on Maxwell Hill, Malaya (depicted photographically in Carlquist, 1975). One rarely sees wilting of native plant species in the wild when other species in the area are not experiencing such wilting. Illicium has notably long scalariform perforation plates with extensive pit membranes remnants (Carlquist, 1982; Carlquist & Schneider, 2002). Although a vessel, being wider than a tracheid, has greater conductive capabilities (the Hagen-Poiseuille effect), the vessels of Illicium (and other woody basal angiosperms) offer exceptional resistance because of end walls. Such end walls might have the effect of prevention of spread of air bubbles from one vessel element to another, a safety device that keeps conduction lower. Porosity of lateral wall pits in vessels is considered a risk for embolism formation (Sperry & Hacke, 2004) and is certainly a consideration. The change to bubble-impervious walls is a problem for primitive vessels; pit membranes of lateral walls can be redesigned for greater safety, although fewer and smaller pores in them do seem to offer less conductive capacity. Lateral wall design seems less a limiting factor than end-wall design because vertical conduction is the predominant function of vessels (see, however, comments below regarding the torus-margo system in conifer tracheids).

The wood of Illicium and other such angiosperms consists of tracheids in addition to vessels. Tracheids are less embolism-prone than vessels because they have bordered pits on end walls, not perforations. However, tracheids are, as noted, less efficient at conduction. There is good comparative evidence that tracheids serve as a subsidiary conductive system, capable of maintaining water columns when vessels are embolized (Carlquist, 1984). However, this does not solve the problem of protecting the vessel-element conducting system and maintaining its function.

Secondary vessellessness

Sperry et al. (2007) provided us with a picture of vessel elements as an invention that, in its earliest versions, confers little advantage and needs a panoply of changes in its structure, together with changes in associated cells, before it becomes a genuine asset. Hence, the slow climb of early angiosperms into more stressful environments and the seeming confinement of species with long scalariform perforation plates to mesic environments is understandable (Carlquist, 1975). Not surprisingly, some clades of early-diverging angiosperms under such constraints may have lost vessels elements; an all-tracheid formula could for a given taxon prove a safer although less than ideal solution for effective conduction. Loss of vessels in Winteraceae and Trochodendraceae under such circumstances is indeed conceivable. This did not prove a good long-term strategy for Winteraceae, which account for only a tiny proportion of angiosperms and which are confined to areas that stay moist longer because of coolness. Such areas do permit frost and, not surprisingly, Tasmannia R.Br. ex DC. wood samples from montane Australia and Tasmania frequently show frost damage, damage repaired by formation of callus tissue followed by resumption of cambial activity (Carlquist, 1989). Another strategy has been the development of warty lumen surfaces on tracheids in montane species of Drimys J.R.Forst. & G.Forst (Carlquist, 1988b) and Tetracentron Oliv. (Carlquist, 1988a), a feature found in tracheids of conifers and vessels and tracheids of angiosperms in areas where freezing and drought are likely to occur (Carlquist, 1988a). Evolution of Winteraceae has been seriously confined by even moderate fluctuations of moisture, drought, cold and humidity. Helical thickenings in vessels of Illicium from more temperate localities (Carlquist, 1982) are a similar adaptation that likely helps resist embolisms caused by frost. Pit membranes of tracheids of Winteraceae (as with Amborella) are delicate, often fracturing during the preparation process (Carlquist, 1983b; Carlquist & Schneider, 2001) and, although relatively good at conductive flow (Sperry et al., 2007), are not well designed for safety. One is not surprised that Amborellaceae and Winteraceae have been unable to radiate ecologically. An all-tracheid system of conduction would not be advantageous in drier montane areas where risk to water columns is greater. The assessment of Feild, Brodribb & Holbrook (2002) of the ecological diversity of Winteraceae (“hardly a relict”) does not seem confirmed by the relatively limited ecological and geographic distribution of this family.

Sarcandra, although it has vessels, seemingly has few of them (Carlquist, 1987) and may be an example of near loss of vessels. We can assume that there is secondary absence of vessels in Winteraceae and Trochodendraceae, but the explanation is not chance; it probably was retreat in certain habitats from vessel elements with little conductive efficiency because these plants did not evolve the wood features that support safety in angiosperm woods (e.g. axial parenchyma distribution, suitable foliar apparatus, etc.). There is less difference between tracheids in vesselless angiosperms and vessels in such families as Illiciaceae and Chloranthaceae than one might think and strict binary thinking is only a detriment to understanding transitions between tracheids and vessel elements. Loss of vessels in woody angiosperms has received much comment, but possible similar loss of vessels in monocots (see below) has escaped notice.

Changes in patterns of axial and ray parenchyma in relation to conduction

Modifications to the vessel element to make vessel-bearing woods competitive include alterations of the perforation plate. Indeed, perforation plates with no bars (simple) are common in terms of species and those with relatively few bars are not common in terms of the world flora (Carlquist, 1975). However, the problem with primitive vessels may not be so much the resistance to flow of the end wall or the vulnerability of lateral wall pit membranes. Maintenance of water columns in vessels (and refilling of embolized vessels, often as a result of root pressure; Tyree & Zimmermann, 2002) is evidently a function of axial parenchyma more than any other factor. Braun (1983) and Tyree & Zimmermann (2002) illustrated legumes in which vessels are completely surrounded by axial parenchyma. In fact, this axial parenchyma formulation is not a common one. Rather, we should consider that any type of axial parenchyma distribution, as well as the upright cells of rays, can play a part in starch storage and release of sugars into vessels. Release of sugars into vessels increases their osmotic pressure and maintains the water column (Braun, 1970, 1983, 1984; Sauter et al., 1973). There is no indication that this process is limited to particular taxonomic groups of angiosperms, nor does it function only in trees of seasonal habits. It may be as much a phenomenon in lianas and their water column maintenance as in aquatic or desert plants: trees of seasonal habitats happen to have been used in the first experiments.

The role of living fibres

Living fibres near vessels may substitute for axial parenchyma in starch storage and release of sugars into vessels (Wolkinger, 1969, 1970, 1971). Thus, design of a parenchyma or parenchyma-like system that can support water column maintenance in vessels may have been a prime problem to be solved in early angiosperms. One can note, for example, that the massive multiseriate rays of Piper, supplemented by axial parenchyma adjacent to vessels and living (septate) fibres (Meylan & Butterfield, 1978) may constitute such a system. Piper shows root pressure indicative of a system that could maintain water columns and repair their embolisms (Tyree & Zimmermann, 2002). Such factors would explain the success of Piper and how it has managed to speciate so abundantly and occur in habitats other than moist understories (although some species can be found in these habitats). Piper has a sympodial growth form with xylary redesign (perforation plates are simple in Piper) that seems basic to success of this genus, which has more species than any other genus of basal angiosperms outside the monocots. Living (septate) fibres may support maintenance of the water columns in other basal angiosperms such as Chloranthaceae, which have wood much like that of Piperaceae, but scalariform perforation plates in vessels that may be related to the relatively narrow ecological range of Chloranthaceae. Development of fibres that help to resist negative xylem pressures (Jacobsen et al., 2005) may be a distinctive feature of angiosperm wood design that resists cavitation formation. Although this wood formula has been tested in only a few plants, its significance may be widespread.

Tracheids as a safety system

Tracheids as a subsidiary conducting system in a vessel-bearing wood do provide enormous safety, as the amazing number of vessel-bearing woody angiosperms with either tracheids or vasicentric tracheids (Carlquist, 1985) demonstrates. Realization of the physiological and evolutionary importance of these cells has not been helped by the failure of the IAWA Committee (1989) to define tracheids inclusively or recognize the pervasive nature of vasicentric tracheids (see Rosell et al., 2007). Because these tracheids and vessels are conductive elements in woods (whereas fibre-tracheids and libriform fibres are not), the importance of parenchyma in supporting conduction maintenance not just in vessels but also in tracheids has not been appreciated. To date, the idea of Braun (1970), supported by experiments, that axial parenchyma supports maintenance of water columns in vessels has not been extended to tracheids. There is every reason to believe that axial parenchyma functions with relationship to tracheids as well as to vessels. In tracheid-bearing woods, axial parenchyma is overwhelmingly in the form of diffuse (or diffuse-in-aggregates) axial parenchyma distribution. Neither vessels nor tracheids need to be ‘sealed’ by a complete sheath of parenchyma for axial parenchyma to support the maintenance of conduction in a wood as the legume examples sometimes mentioned suggest. Three-dimensionally, probably all vessels and tracheids in a vessel-bearing wood are in contact with axial parenchyma (mostly diffuse) or upright ray cells. This may help explain the abundance of upright ray cells in woods of early angiosperms, as well as the persistence in so many angiosperms of diffuse or diffuse-in-aggregate axial parenchyma. The torus-margo construction as seen in conifer tracheids with its attendant advantages (see below) is essentially lacking in angiosperms, so that an axial parenchyma system becomes potentially more valuable as an osmotic support system for angiosperm tracheids.

Axial parenchyma function

Bailey (1957) noted that at least 90% of woods with primitive vessels (narrow, with scalariform perforation plates) have diffuse (or diffuse-in-aggregates) axial parenchyma. These same angiosperms mostly have tracheids (= conductive cells) rather than fibre-tracheids or libriform fibres as the imperforate tracheary element type. This suggests that spread of axial parenchyma among numerous conductive cells, perforate and imperforate, is necessary to support the conductive system.

Design of rays with numerous upright cells correlates with this as a way of forming a living cell system interactive with the axial parenchyma. Redesign of grouped axial parenchyma cells that better aid (through osmotic activity) conduction in wider, more conductively efficient vessels is probably basic to improving the vessel as a conducting cell. In turn, wider rays with more procumbent cells enable woody species to use rays as photosynthate conduits for thicker trunks. In lianas, which have relatively abundant wide vessels, abundant procumbent cells in rays are a strategy for rapid shunting of photosynthates into axial parenchyma as a way of osmotically supporting the conductive system. This might help explain why primitive vessels were marginal improvements over tracheids. Woods with primitive vessels mostly have tracheids, so axial parenchyma might be a support system to conduction and maintenance of water columns in both vessels and tracheids. Because axial parenchyma must be distributed in a diffuse pattern to support (through osmotic functions) the vessel/tracheid conductive system, rays would have to be co-designed to be numerous and narrow and to link with the axial parenchyma by upright cells. Such a design would not serve for rapid shunting of photosynthates throughout the secondary xylem. Vessel origin and design modification required simultaneous changes in parenchyma and thus an overhaul of the physiological nature of how the wood works. Considerations such as these may explain why, if Sperry et al. (2007) are correct, wood of ‘primitive’ vessel-bearing angiosperms is only marginally more efficient than vesselless wood. Vesselless woody angiosperms have only small amounts of axial parenchyma (always diffuse or diffuse-in aggregates) in secondary xylem and the restrictions of such a hydraulic support system offered by the parenchyma to the tracheids may be a feature that has limited radiation of vesselless groups of angiosperms.

Margo-torus pits of conifers vs. angiosperm tracheary element pit membranes

Conifers have a vesselless system superior to that of vesselless angiosperms because of the torus-margo construction of tracheid pit membranes (Pittermann et al., 2005). The large margo pores enhance flow; the torus, when displaced owing to pressure differential among tracheids, offers excellent closure of pores to enhance safety. Although there is no direct evidence that the earliest angiosperms lacked a torus-margo pit membrane construction, all available evidence, especially the tracheary element structure found in basal angiosperms, suggests that basal angiosperms did not have a torus-margo pit structure in tracheids (or vessel elements). There is a simple topological circumstance that has not been appreciated to date: the torus-margo scheme works well when a pit is circular: the margo strands are equal in length and therefore tension on all of them is similar, so that deflection of the torus is mechanically feasible. If early angiosperms had scalariform pitting in tracheids, especially wider tracheids (as in Amborella, Tetracentron, Trochodendron Siebold & Zucc. and some Winteraceae), the torus-margo system would have been unworkable for topological and mechanical reasons. An elliptical torus would have been conceivable, perhaps, but there would have been differential tension on margo strands attached to the elliptical torus that would have prevented closure (more tension on some strands than on others in order for closure to be achieved). A circular valve is mechanically better than an elliptical one. Thus, the early angiosperms were faced not merely with the invention of the vessel element in order to achieve conductive competency, but a vessel element designed, together with its histological ‘support system’ (parenchyma, tracheids, etc.) in wood that was at least as good as the torus-margo system of tracheid design in conifers. Sympodial plant construction, juvenilized xylem and occupation of moist habitats permitted the early angiosperms to ‘buy time’, during which the disadvantages of early xylem structure could be countered. Paedomorphosis in xylem evidently was basic to survival of early angiosperms and critical to the explosive diversification of angiosperms once competence of the vessel system had been achieved.

Parenchyma cell longevity and abundance: new modes

Axial parenchyma is mostly diffuse and mostly not abundant in early-divergent woody angiosperms; rays are narrow and numerous. These features are characteristic of conifers (and are not indicative of relationship with conifers). The change to more abundant axial parenchyma, often in groupings around vessels (paratracheal) is characteristic of later-diverging (= most) eudicots. As noted above, this configuration may be associated with controlling conduction in vessels and refilling them after air bubble or embolism formation, but, in addition, statistically it is associated with wider ray size (Carlquist, 1988a). Wider rays, which usually feature more procumbent ray cells than do narrower rays, are associated with radial transport and, more significantly, storage of photosynthates (Carlquist, 2007a).

The functions just cited for parenchyma in angiosperm woods are related to longevity as well as abundance and grouping. Unfortunately, most wood anatomy studies have been based on dried specimens, from which presence of nuclei cannot be discerned. In fact, longevity of both axial parenchyma and ray cells is considerable (Frey-Wyssling & Bosshard, 1959; Nair & Chavan, 1983; Spicer, 2005; Spicer & Holbrook, 2007). Extent of living axial parenchyma and ray cells is, according to these authors, approximately the same as the extent of sapwood (as discerned by other methods). The relatively large volumes of parenchyma in sapwood, their organization and their longevity must be taken into account in constructing a picture of the evolution of angiosperms. Certainly, change in configurations and longevity of axial and ray parenchyma accompanies the shift from paedomorphic patterns to adult (progenic) patterns.

Relative roles of root pressure and the cohesion-tension system

Fisher et al. (1997) studied root pressure in a relatively large assemblage of tropical species and their conclusion is very pertinent to the present considerations: ‘Although root pressure alone could not refill embolized vessels in tall trees and vines, the reduction of xylem sap tension would facilitate the dissolving of gas bubbles and the reestablishment of vessel function, at least in the lower parts of the plant or throughout shorter plants’. A conductive system based largely on root pressure is, therefore, quite conceivable in the early angiosperms. The difficulty in evolution of tree forms in angiosperms may have been the difficulty in establishing cohesion-tension conduction and the design of wood (and foliage) suited for that. Further studies of the roles of the two respective modes of conduction, how they interact, in which plants and how wood structure is involved, are very much needed. Certainly, the idea of Sperry et al. (2007) that primitive vessels conferred less conductive advantage than once thought is related to a shift in conductive modes.

Monopodial woody angiosperms: heterochrony and flexibility in wood design

If, as Sperry, Hacke & Pittermann (2006) and Sperry et al. (2007) contended, design of a successful vessel-bearing system was fitful at first, vessel eficiency and safety became requisites for tree wood design. Key to this design is development of conduits that had low resistance. Large vessel diameter is the main mechanism. No shrub in the flora of southern California has a vessel diameter that reaches 100 µm, whereas vessels that wide are common in wet tropical and temperate forests (Carlquist & Hoekman, 1985). Vessel elements with few or no bars per perforation plate characterize tropical rain forest trees (Carlquist, 1975). Maintenance of such vessels with systems of axial parenchyma is also a requisite, so that axial parenchyma that partly or wholly ensheathes vessels and is sufficient for contact with ray cells in which photosynthate conduction is actively occurring is necessary. Sugar release into vessels to maintain osmotic status during flux in conduction rates seems an important function (the extent of which in angiosperm woods remains unstudied). Complete division of labour between a conductive system of vessels and a mechanical system of fibres provides strength for support of a large foliage crown and also fibre sheathing of vessels, protecting them from deformation (Jacobsen et al., 2005; Carlquist, 2009). Abundance of procumbent cells in rays provides for input of photosynthates to living cells at various distances away from the cambium and rapid mobilization of photosynthates for abrupt massive flowering and growth flushes. The evidence from comparative anatomy shows that the above characteristics can be found throughout large angiosperm forest trees. These fibro-vascular characteristics explain how in tropical forests angiosperms have long been able to outcompete conifers, which are adapted to a steadier system of conduction.

There is a downside to these systems in that huge wood accumulations must be protected from fungal attack, roots must be able to access an increasing or steady amount of water and foliage and wood must be able to cope with moderate fluctuation in conductive rate. The conductive systems of arborescent angiosperms, however good at conduction, need considerable evolutionary modification for safety (deterrence of embolism formation and resorption of air bubbles) when a particular clade is faced with adaptation to dryer and colder conditions. Such mechanisms include narrowing and shortening of vessel elements, presence of vasicentric tracheids, development of helical thickenings in vessels (Carlquist & Hoekman, 1985), various non-random groupings of vessels (Carlquist, 2009), development of growth rings (with latewood of sufficient safety for conduction), thickening of vessel walls and development of suitable vessel length and membrane porousness in lateral vessel walls (Sperry & Hacke, 2004; Wheeler et al., 2005; Sperry et al., 2007; Jansen, Choat & Pletsers, 2009). Most importantly, yet unappreciated, is the ability to redesign woods for particular conditions by shift in quantities, positioning and morphology of particular cell types. This flexibility in wood composition and design is a key to the success of angiosperms. Such flexibility is notably absent in conifers, in which shifts from a basic design pattern, seen by means of comparative wood anatomy, are minor compared with the manifold redesigning attained by angiosperms.

Gene action and paedomorphosis

Paedomorphosis is an important component of much design flexibility in angiosperm woods (Figs 1, 2) because the spectrum of character diversity it represents can apparently be accessed by relatively minor changes in gene regulation and action rather than by extensive changes of genomes (Knox, Downie & Palmer, 1993; Panero et al., 1999; Olson, 2003; Rowe & Speck, 2004; Lahaye et al., 2005; Melzer et al., 2008). Paedomorphosis is not a single invention; it is a series of inventions with multiple origins.

Non-woody sympodiality: monocot designs and their heterochronic diversity

The idea that monocots may represent not merely one but numerous forms of paedomorphosis may seem illogical, but the systematic and organographic distribution of xylem types in monocots is explainable in these terms. Loss of a cambium in monocots is an extreme heterochronic event, certainly, and thus paedomorphosis is basic to their evolution. However, within monocots, there have been various types of vessel shifts: roots, stems, leaves and inflorescence axes are developmentally not synchronous and different xylem types may be found in these organs.

Loss of a cambium at the point of origin of monocots, far from being a retrograde step, has led to diversification in xylary tissue and growth forms. In secondary growth of a woody dicotyledon, vessel initiation by the vascular cambium extends all the way from branches through the trunk to the roots synchronously. This may seem like an advantage and it is under many circumstances. However, such a vessel system, to cite merely two examples, is an invitation for spread of pathogens throughout a plant and requires a long-term reliable water supply at subsurface levels. Correlated with lack of a vascular cambium is the sympodial design of monocots. Because sympodial growth forms are a basic pattern in basal angiosperms, this is not surprising, but the sympodial habit and lack of a cambium are strongly linked. Formation of a ‘monocot’ cambium in Aloe, Dracaena, etc. characterizes only a few genera and is an apomorphic feature in monocots and the xylary systems of palms (which have no monocot cambium) are special topics outside of the scope of the present essay. Relevant to this discussion, however, are non-monocot families that have vascular bundles without cambial activity: Nymphaeaceae, Nelumbonaceae, Gunneraceae, some Apiaceae and Piperaceae (Hydatellaceae have stems in which a vascular column is intermixed with parenchyma cells). Features of the monocot plan relevant to sympodiality and organographic paedomorphosis include:

  • 1Occupancy of wider areas at ground level by means of stolons and rhizomes.
  • 2Cable construction: vascular bundles separated from each other, sheathed in fibrous or sclerenchymatous tissue in various fashions so that protection of strands of water columns in vessels from the effects of mechanical torsion is possible. In woody angiosperms, the location of fibres is tied to secondary phloem and secondary xylem. In the monocot plan, fibres and sclerenchyma can be located in any conceivable pattern with respect to the vascular tissues.
  • 3Design of the roots as a water-absorption organ with a temporal duration different from that of stems, with more tracheary elements less resistant to flow (i.e. simple perforation plates) than those of the stem. Such roots permit rapid water input (especially valuable if roots are few) which can be achieved during a short wet season.
  • 4Maximal efficiency in use of water sources near the ground surface, avoiding dependence on deeper groundwater sources.
  • 5Design of roots, stems and leaves as organs independent of each other where perennation, storage, frost resistance and drought resistance are concerned. The fact that all roots (except for the ephemeral first root) are adventitious permits a conductive disjunction between the organs of a plant and therefore functional capacities of each of the three main organs. In a typical bulb (e.g. Allium L.), for example, roots are ephemeral (and have vessels with simple perforation plates, which enhance conduction), whereas stems and leaves have tracheids only. Because tracheids are more characteristic of protoxylem, whereas vessels occur more commonly in metaxylem, a form of paedomorphosis can be accessed to achieve predominance of a particular tracheary element type (e.g. elements with helical thickenings in organs in which prolonged elongation occurs).
  • 6Escape from infection of an entire plant from terrestrial fungi such as Armillaria because adventitious roots are more ephemeral than stems in most monocots. Better survival from underground predators.
  • 7Organographic differences that are related to different types of tracheary tissue in terms of tracheary element elongation. This permits basal meristems in leaves because of the elongation capabilities of helical elements, controlling degree and timing of leaf elonation. Although metaxylem elements are typically pitted in, for example, the rhizomes of Canna L., the inflorescence axes, which are subject to rapid elongation related to flowering, have helical elements and no pitted elements (original data). The characteristics of protoxylem (helical thickenings) and metaxylem (pitted elements) can thus be distributed differently among organs, a pattern that should be regarded as a form of heterochrony.
  • 8By virtue of scattered and roughly equidistant bundles or vascular strands, input of photosynthates into a parenchymatous background is facilitated. Likewise, this plan is ideal for water storage in stems or leaves (e.g. orchids).

Cheadle (1942, 1943a, b, 1944) described family-by-family distributions of vessels in monocots and degrees of specialization in them. Cheadle had the idea that there is a progressive origin and specialization of vessels phylogenetically and, within a plant, organographically. These vessel distributions were mapped in terms of organographic and ecological differences (Carlquist, 1975). When one does this, it can be seen that few genera and families have vessels with types of perforation plates other than scalariform throughout the plant body. It can also be seen that an appreciable number of plants have only tracheids in stems and leaves, a marked disjunction rather than a progressive specialization. The families and genera with vessels in roots but with only tracheids in the above-ground organs include aquatics, but also many plants with distinctive non-aquatic habits, notably bulbs in Asparagales and Liliales, and orchids and bromeliads. Yet all of these have the genetic information to make vessels. Cheadle's underlying assumption is that monocots began with a vesselless condition and have experienced increasing presence of vessels and increasing specialization (scalariform perforations plates phylogenetically yielding to simple ones). However, can there be any reversibility or is such a progression inexorable, with clades unable to retreat to the safety of a tracheid-only system for leaves and stems? The evidence from orchids (Carlquist & Schneider, 2006) suggests that reversibility is possible, although reversion from vessels to tracheids is not demonstrated in that paper (such a reversion would be difficult to demonstrate). Heterochrony offers one mechanism, but there are others. Acorus L., now considered sister to the remainder of monocots (e.g. Soltis et al., 2000), lacks vessels in stems and leaves and may also lack them in roots (the small pores figured in end walls of root tracheary elements of Acorus by Carlquist & Schneider (1997) probably do not qualify those cells as vessel elements). Aponogeton L.f., a member of the early-diverging, largely aquatic clade Alismatales, is apparently vesselless (Cheadle, quoted in Carlquist, 1975). Ancestrally, monocots might have been vesselless or nearly so. There is no line between tracheids with porose pit membranes on end walls and vessels with holes only marginally larger in pit membranes of end walls and the binary definition of tracheids and vessel elements is untenable in stems and even roots of a number of monocotyledons. One could say that a number of orchids, especially terrestrial ones, are vesselless, depending on one's definitions, which in turn must be based on SEM observations (Carlquist & Schneider, 2006), not a type of evidence that is accessible to most phylogeneticists. Although secondary vessellessness in Winteraceae and Trochodendraceae has received extensive comment by phylogeneticists, the possibility that some monocots might have experienced evolutionary loss of vessels in the tracheary tissue of a particular organ has not received comment. Certainly one could hypothesize a positive selective value for tracheids, which can prevent spread of air embolisms throughout a conductive system in stems and leaves of epiphytic orchids and bromeliads, as well as in other monocots. In view of the evolutionary diversity and taxonomic size of Orchidaceae and Bromeliaceae, inability to evolve vessel elements in stems and leaves may be an idea more difficult to envisage than a kind of gene action that shuts off vessel formation in organs where rapidity of water conduction is not vital but protection against embolism spread is (e.g. orchid pseudobulbs).

All Nymphaeaceae could easily be considered to be vesselless. They could just as easily be considered to have tracheids in stems but tracheids somewhat transitional to vessel elements in roots (Carlquist & Schneider, 2009; Carlquist, Schneider & Hellquist, 2009; Schneider, Carlquist & Hellquist, 2009).

Limits and extent of heterochrony in angiosperms

Systematic distribution

A surprising number of the larger families of non-basal angiosperms have not only a remarkable range in habit from woody to non-woody, but considerable range even within genera. Such families include Acanthaceae (3510 species) Asteraceae (23 000), Brassicaceae (3850), Euphorbiaceae (7000), Lamiaceae (6700), Malvaceae (2967) and Melastomaceae (4830) (species numbers from Soltis et al., 2007). These families all have a wealth of derived wood characters (e.g. simple perforation plates). Differences in degree of woodiness are probably easily achieved, perhaps by such mechanisms as those described by Nilsson et al. (2008). The difference between herb and non-herb can be described merely as the number of cell layers produced by the cambium, but certainly not as presence or absence of cambial activity. Brassica oleracea L. can be an annual or a small tree, depending on the cultivar, and plants of this species as a whole have cambial activity. The ease of transformation of growth forms in many genera, especially of the families mentioned above, must be attributed to heterochrony in the form of gene action dictating more cambial activity or less of it.

Geographical and ecological distribution

Secondary woodiness on islands is a common phenomenon and molecular analyses of genera involved in this phenomenon permit recognition of increase in woodiness on islands as compared with closely related taxa elsewhere. Although islands were early cited as home to striking instances of secondary woodiness (Carlquist, 1974), there are many areas where one can find secondary woodiness. Notable among these are equatorial ‘sky islands’ (East Africa and the Andes of South America most prominently), but the phenomenon is more common than one might think. For example, the woody species of Asteraceae, Hydrophyllaceae, Lamiaceae, Polygonaceae and Scrophulariaceae in southern California contain good examples: the frost-free or nearly frost-free nature of this area in winter is basic to evolution of secondarily woody species.

Heterochrony: limits and kinds

Heterochrony as envisioned here goes well beyond degree of woodiness. Criteria for paedomorphosis are given in the body of the paper (descending age-on-length curves for vessel element length, predominance of upright cells in rays and wide rays that extend with little alteration from pith into secondary xylem, etc.). However, a phenomenon such as raylessness also exemplifies the tendency for ontogenetic change in wood histology related to retention of a juvenile condition. All features regarded as juvenile here should be considered functional at the present time, rather than relictual. For example, procumbent ray cells should be considered to function in radial translocation of solutes, as hypothesized by Braun (1970): in thick stems, procumbent ray cells permit shunting of solutes into and out of three-dimensionally large starch storage systems. Thus, narrower stem diameter is related to upright ray cell predominance: photosynthate translocation is predominantly vertical in stems with a greater number of juvenile characters. Axial and ray parenchyma systems are in contact and definitely function conjunctively. Axial parenchyma functions in releasing sugars into conductive xylem cells. Some axial parenchyma types (diffuse, diffuse-in-aggregates) relate to whether tracheids are present and how they are disposed, for example, and tracheid presence in wood is not a feature related to degree of juvenilism or adulthood, but rather to other factors (Carlquist, 1985).

The idea that numerous angiosperms show some degree of paedomorphosis in xylary structure rather than the few mentioned earlier (Carlquist, 1962), and thereby can be ranked along a continuum (Fig. 2), may seem novel, for this is a huge expansion of this concept in plants. Conformations in wood structure with paedomorphosis are readily seen exemplars of heterochrony. However, rapid changes in the cambium that result in ‘typical woody’ patterns are also exemplars: both represent changes from primary xylem, differing in degree and kind; ontogenetic changes take place in all angiosperm xylem To omit the adult patterns of ‘woody species’ from consideration would be as mistaken as to think of paedomorphosis in woods as non-existent. The age-on-length curves for woody species presented by Bailey & Tupper (1918) show ontogenetic changes in woody species such as Liriodendron tulipifera well. However, Bailey & Tupper (1918) did not include in those curves types of secondary xylem other than those that are ‘typically woody’ and thus implications of their curves have waited for nearly a century to be put into the context of angiosperms as a whole.

Changes in histology within primary xylem were presented in careful detail by Bierhorst & Zamora (1965). They pioneered the idea that presence of tracheids and vessels with various types of perforation plates must be understood within an ontogenetic context. The ontogenetic changes within primary xylem that they showed are by no means uniform within angiosperms and are thus indicators of heterochrony. The concept of organographic heterochrony in monocots may seem novel, but organs are, of course, developed in an ontogenetic sequence in monocots. More importantly, as monocots lack vascular cambia, they have developed another type (or form) of diversification of xylem histology with its own kinds of timing.

If all ontogenetic changes described here are related to functions, how can heterochrony be indicative of angiosperm origins? The sympodial nature of most early angiosperms, cited occasionally (Carlquist, 1992b), has not been generally appreciated. Early-diverging angiosperms are pervasively and apparently ancestrally sympodial, with ‘escapes’ into monopodiality (woody trees and growth forms leading up to them) occurring only in a few taxonomic groups (Laurales, Magnoliales). These ‘escapes’ in turn are probably the result of progressively better vessel function. As noted by Sperry et al. (2007), vessels with primitive morphology, along with other suboptimal features of xylem organization, made early vessels of marginal value. Such suboptimal conductive solutions are still surviving in some basal angiosperms because of compromises that have permitted their survival. Sympodiality in early angiosperms essentially equates with degrees of paedomorphosis, because shoots are limited in duration and roots are adventitious. Thus, we find Amborella to be a plant with sympodiality yielding to monopodiality to a moderate degree, but still notably juvenile in its ray structure in accordance with the finite duration of its stems: a survivor in wet understory on an island habitat that is a refuge for angiosperm relicts. The next split on the phylogenetic tree of angiosperms, Nymphaeales, consists of sympodial plants with no cambial activity, surviving in habitats successfully entered by relatively few other angiosperm clades: relatively anoxic streams and ponds requiring special gas exchange systems within the plant. Subsequent splits on the tree of angiosperms have other distinctive features relating to sympodiality and management of water. A derived state of vessel morphology (simple perforation plates) is found in Piperaceae. We can hypothesize that the enormous radiation of Piper (about 2000 species: Jaramillo et al., 2008) relates to an early solution of vessel design by a family that has retained its sympodiality.

Broader vistas; gymnosperms

Does heterochrony extend to seed plants other than angiosperms and have a relationship to their success? Conifers offer an interesting example. Where xylem is concerned, conifer wood ontogeny corresponds to the ‘adult’ model in the age-on-length curve offered by Bailey & Tupper (1918). Conifers (not counting Gnetales) have in the torus-margo pit structure of tracheids an effective alternative to angiosperm conductive systems in lower resistance (margo) to movement of water columns combined with maximal safety (displacement or aspiration of the torus). This system has evidently been so successful that alterations to it have been minor compared with the tremendous diversity in structure of angiosperm woods. Conifers have mostly uniseriate rays, composed of procumbent cells, and modest quantities of axial parenchyma: the parenchyma systems do not have the functional and anatomical diversity of axial parenchyma and ray parenchyma in angiosperms. Conifer woods are relatively stereotyped and also minimal in degree of paedomorphosis. Diversification in parenchyma and restructuring both for storage (rays) and osmotic support of the vessels and tracheids of xylem were probably important parts of the changes required in angiosperm wood before angiosperms became competitive with conifers for a wide range of habitats. The innovations in angiosperm xylem of parenchyma types and their strong interlinkage and the diversification of angiosperm tracheary elements offer keys to angiosperm diversification. Ontogenetic diversification and thus paedomorphosis in xylem is an inevitable part of the many reconfigurations of angiosperm stems and roots for various kinds of storage, photosynthate travel paths and rates and even stem elongation. One notes the absence of lianas among conifers, for example. The distance between paedomophosis and progenesis is constricted within conifers.

Cycadaceae have relatively stereotyped growth forms, suggesting, in contrast, a sort of permanent juvenilism. Primary xylem of cycads contains scalariform pitting of tracheids (Greguss, 1968). Presumably this pattern, common to all cycads, is evidence of the refugium effect and is an ancestral feature of cycads. However, scalariform pitting is extended indefinitely in the genera Bowenia Hook., Stangeria T. Moore and Zamia L., which have rhizomes (Stangeria) or condensed succulent stems, in contrast to the upright trunks of the other genera. This is a clear indication of paedomorphosis. In the cycad genera other than Bowenia, Stangeria and Zamia, such as Cycas L. or Encephalartos Lehm., scalariform tracheid pitting yields to circular (or polygonal) pits at various times (depending on the species and the site sampled within the plant), so that juvenile patterns do eventually yield to an adult pattern (Greguss, 1968). Unlike pinalean conifers, cycads may be thought to be locked into a number of juvenile features (thick condensed trunk devoted mostly to starch storage) that give cycads excellent survival against drought, herbivores and fire.

Key to the many expressions of heterochrony in xylem are types of gene action. Study of these has just begun (see literature cited above). There are various expressions of heterochrony related to xylary ontogeny. Some types of gene action may merely involve changes in levels of a single chemical, but others will be complex. We must understand the physiological bases of different xylary types and their histologically diverse ontogeny in order to direct the tasks involved in discerning modes of gene action.

ACKNOWLEDGEMENTS

Observations that lead to evolutionary syntheses come as a result of optimal career length and thereby those who have believed in me have made a paper such as the above possible. Among those individuals are Bruce G. Baldwin, Dana Campagna, Lincoln Constance, Thomas S. Elias, Karl Niklas, Regis B. Miller, Philip A. Munz, Mark E. Olson, Peter H. Raven, Edward L. Schneider, William L. Stern, Warren L. Wagner and Scott Zona. There have been others, but memory is sometimes faulty. Thanks are also extended to lenient editors I have known and to some kindly reviewers. Without the help of such individuals at important moments, a career falters. Mark W. Chase and Edward L. Schneider offered many suggestions that substantially aided the manuscript.

Ancillary