THE TRACHEID VESSEL ELEMENT TRANSITION IN

Transcription

1 American Journal of Botany 89(2): THE TRACHEID VESSEL ELEMENT TRANSITION IN ANGIOSPERMS INVOLVES MULTIPLE INDEPENDENT FEATURES: CLADISTIC CONSEQUENCES 1 SHERWIN CARLQUIST AND EDWARD L. SCHNEIDER 2 Santa Barbara Botanic Garden, 1212 Mission Canyon Road, Santa Barbara, California USA Current definitions of tracheids and vessel elements are overly simple. These definitions are based on light microscope studies and have not incorporated information gained with scanning electron microscopy (SEM) or transmission electron microscopy (TEM). Current definitions are based primarily on angiosperms, especially eudicots, and were devised before many basal angiosperms were carefully studied. When all sources of information are taken into account, one can recognize changes in six characters in the evolution of tracheids into vessel elements in angiosperms (or vice versa) as well as in other groups of vascular plants. There is an appreciable number of taxa in which all criteria for vessel origin are not met, and thus incipient vessels are present. At the very least, vessel presence or absence should not be treated as a single binary character state change in construction of cladistic matrices. Increase in conductive area of an end wall by means of lysis of progressively greater areas of pit membrane and increase in pit area on the end wall (as compared to pit area on equivalent portions of lateral walls) are considered the most important usable criteria for recognizing intermediacy between tracheids and vessel elements. Primitive character states in vessel elements are briefly discussed to differentiate them from changes in character states that can be regarded as intermediate between tracheids and vessel elements. Key words: cladistics; evolution in primitive woods; origin of vessels; pit membranes; tracheid vessel element transition ; vessel elements; xylem. Definitions of vessels in pertinent literature and in glossaries provide insufficiently detailed diagnostic criteria. For example, the IAWA Committee on Nomenclature (1964) defines a vessel as an axial series of cells that have coalesced to form an articulated tube-like structure of indeterminate length; the pits to congeneric elements are bordered. How the coalescence of vessel elements occurred is not explained, and the first half of the definition could apply equally well to articulated laticifers. The second half of the definition would be better if it made a distinction between the presence of bordered pits on lateral walls and perforation plates on end walls of vessel elements. Commonly vessel elements are considered perforate cells, whereas tracheids are considered a type of imperforate tracheary element in dicotyledons (along with libriform fibers and fiber-tracheids). The perforation plate is an end wall in which pit membranes are lacking in this commonly used or assumed definition. This common-sense definition probably is better than that of the IAWA Committee on Nomenclature (1964) definition. Often omitted is the important distinction between vessel element (also known as vessel segment or vessel member) and a vessel, which is a series of vessel elements in which contiguous end walls are in contact. For purposes of cladistics, binary definitions are generally preferred. Where definitions of vessel presence or absence are concerned, such a usage can be traced to Young (1981), who hypothesized that vesselless dicotyledons are secondarily vesselless, in contrast with the traditional idea that vessellessness is primitive in dicotyledons (e.g., Bailey, 1944). Workers interested in developing phylogenetic schemes for the basal dicotyledonous angiosperms, therefore, have taken an interest in 1 Manuscript received 8 May 2001; revision accepted 28 August Appreciation is expressed to the xylaria of Harvard University, the U.S. Forest Products Laboratory, and the U.S. National Museum of Natural History (Smithsonian Institution) for wood samples. Dr. Douglas E. Soltis and Dr. Pamela S. Soltis offered helpful cladistic perspectives. Improvements to the drawing were kindly provided by Mr. Don Matsumoto. 2 Author for reprint requests. 185 the phyletic status of vessellessness. Herendeen and Miller (2000) deal with this situation by recognizing the feature wood vesselless and say that the feature is accounted for under perforation plate structure (i.e., perforation plates absent). Under perforation plates, they describe the character states as simple; scalariform; reticulate, foraminate, and/or other types of multiple perforation plates; categories for number of bars per plate, therefore recognizing variations in easily recognizable perforation plates but not degrees of intermediacy between tracheids and vessel elements. The definition of vessellessness assumes that the presence or absence of perforation plates can be readily recognized or defined, but as we will show, that is not always the case. Vessel presence in transections of axial xylem is often used pragmatically by wood anatomists who expect vessels ( pores in dendrological literature) to be wider than imperforate tracheary. Note should be taken that in a relatively small number of woods (usually highly specialized), vessels may have diameters similar to those of imperforate tracheary elements. Clearly, there is more than one criterion for declaring that vessel elements are present or absent or for defining a vessel element as different from an imperforate tracheary element. The importance of vessel evolution in angiosperms is considerable, and one can honor the desire of cladists to incorporate this feature into data matrices and to interpret the phyletic status of vessellessness based on the trees that are produced. Although tracheary elements in some dicotyledon groups are transitional (as noted in the title of at least one, which includes the wording, tracheid vessel element transition : Carlquist, 1992a), the problems created by this transition have not been fully incorporated into cladistic practice. The nature of this transition stems in part from the recent intensified study of xylem of basal angiosperms (Magnoliales, Illiciales, Laurales, Nymphaeales, Piperales, Ceratophyllaceae, Amborellaceae, and monocotyledons), Gnetales, and ferns with the aid of scanning electron microscopy (SEM) and transmission electron mi-

2 186 AMERICAN JOURNAL OF BOTANY [Vol. 89 croscopy (TEM). Changes in character states of several characters can be hypothesized in the evolution of a vessel from a tracheid. This analysis is presented in the form of generalizations that have typically been used or assumed in defining vessel elements. Exceptions to each of these generalizations are then cited. The exceptions are cited because they give details of tracheary elements that are intermediate between tracheids and vessel elements and therefore document that changes in character states of more than one character can and have taken place independently. Given the validity of this analysis, one must concede that the tracheary elements described under these exceptions do not fit a binary system. If that is true, new cladistic solutions and usages are advisable. Because dicotyledons ( eudicots and basal angiosperms) are either vesselless or, in the vessel-bearing species, can have tracheids (or other imperforate tracheary elements) associated with primitive vessel elements, the comments in this paper are concerned with dicotyledons. Monocotyledons can have vessels that are not associated with tracheids; fiber-tracheids and libriform fibers are not to the best of our knowledge known to exist in monocotyledons, however. Plant anatomy texts universally give this interpretation, although we know of no statement on this situation other than the one we give above. Some tracheid-like features can be found in vessels of primitive moncotyledons and will be mentioned in appropriate places below. In fact, the first two (possibly three) characters discussed below pertain to monocotyledons as well as dicotyledons and could be used in cladistic analyses of monocotyledons. MATERIALS AND METHODS Collection data, methods, and author citations are the same as those in Carlquist (1990, 1992a, b). The SEM equipment available to us is, unfortunately, old and inexpensive and does not yield images of optimal sharpness at higher magnifications. Nevertheless, we are confident that the structures shown and cited are entirely real, even if sometimes rendered with suboptimal resolution. The pores in pit membranes and the various forms of pit membrane remnants are in agreement with those figured by Meylan and Butterfield (1978) and Butterfield and Meylan (1980), whose equipment and technique produce exceptionally fine results. We believe that our figures contain a minimum of artifacts. We have offered a separate study in which pit membrane remnants are alike using various microtechnical methods (Schneider and Carlquist, in press). Our figures of pit membrane remnants reflect mature vessel elements, not those adjacent to cambium; we viewed the entire width of radial sections in order to establish the mature vessel element condition. Our analyses of tracheids and pit membranes do not apply to conifers and Ginkgo, in which the pit membrane structure represents distinctive expressions. The tracheids of conifers differ from those of angiosperms in structure and also physiology (e.g., pit aspiration that prevents functioning of margo pores). RESULTS Intermediacy between tracheids and vessel elements Listed below are six characters in which evolution from tracheids to vessel elements (or else the reverse phylesis) has occurred. The boldface headings are, therefore, characters that represent portions of the definition of a vessel element as typified by definitions in the literature. Each of the six headings is followed by exceptions. In these exceptions are cited examples of tracheary elements that are not usually considered to meet that criterion of vessel element presence, but represent departure from the all-tracheid state toward vessel presence (or conceivably the reverse). Thus, we are stating stages in the tracheid vessel element transition as exceptions to vessels that have acquired the character states usually presented as defining vessel elements. 1. Pit membranes are absent in perforations of mature vessel elements (Figs. 1 12). A perforation plate is assumed to be one or a series of modified pits ( perforations) in the end wall of a tracheary element; these modified pits are assumed in most sources to lack pit membranes entirely. The absence of a pit membrane is the result of lysis of the pit membrane as the vessel element matures; the dislodged pit membrane disintegrates and disappears (Esau and Hewitt, 1940; drawings reproduced in Fig in Esau, 1965). Although Esau and Hewitt (1940) studied this process in a simple perforation plate, the same process undoubtedly occurs in each perforation of perforation plates with more than one perforation (i.e., scalariform, multiperforate, and foraminate plates). Exceptions. Using SEM, Meylan and Butterfield (1978) figured microfibrillar webs that persist in perforations of mature vessel elements. The instances they illustrate show strand or weblike pit membrane portions that occlude much less than half of each perforation (in scalariform perforation plates). Most of the instances they figured do not constitute a serious exception to the principle that pit membranes are removed from perforations during maturation of vessel elements. However, in a review of presence of pit membrane remnants in scalariform perforation plates as seen with SEM (Carlquist, 1992b), extensive porose or weblike pit membranes that represent blockage of 50% of the areas of perforations are figured for species from the genera Ascarina (Fig. 6), Heliamphora, Illicium, and Paracryphia (Fig. 5). Some other species of Illicium figured in that paper show blockage of 50% of the membrane area. Although numerous genera with pit membrane remnants are listed in this review (Carlquist, 1992b), more such genera with various degrees of pit membrane lysis will likely be found, because relatively few woods have been examined with SEM in this respect. Theoretically, a vesselless wood should contain nonporose end wall pits at the level of resolution shown in Fig. 1 (plasmodesmata are not included in the present discussion and would not be resolved by our equipment). It is only a short step from the porose pit membranes of tracheary element end walls of Tetracentron (Figs. 3, 4; see also Carlquist, 1988a, 1992b, 1996a, 2001) to the porose pit membrane remnants in perforations of the genera listed above. The tracheary elements of Tetracentron have never been designated as anything but tracheids, although the pits of the lateral walls are smaller than those of the end walls and, in addition, lack porosities (see illustrations in Carlquist, 1996a). A similar instance has been reported by Takahashi (1988), who, using TEM, found small pores in pit membranes of stem tracheids of Sarcandra (Chloranthaceae), a genus once thought to be vesselless, but the pores are the only reason why the stem tracheary elements can be said to be incipient vessels (in contrast with the root vessels of Sarcandra that have few pit mebrane remnants, see Carlquist, 1987a). Porose pit membranes in end-wall pits combined with nonporose pit membranes were observed in Bubbia of the Winteraceae (Carlquist, 1983a) and in Amborella of the Amborellaceae (Carlquist and Schneider, 2001b). Feild et al. (2000), on the contrary, report absence of pit membranes (perhaps an artifact) in end walls of some tracheids but not in others, and claim on the basis of low-pressure air injection measurements that vessels averaging two cells in length are present. Porose pit membranes in end walls, combined with nonporose pits on lateral walls, were

3 February 2002] CARLQUIST AND SCHNEIDER TRACHEID VESSEL ELEMENT TRANSITION 187 Figs Scanning electron microscopy photographs of tracheids of vesselless dicotyledons. Figs Trochodendron aralioides. 1. End wall, nonporose pit membrane. 2. Lateral wall pitting, showing smaller area compared to end wall pits. Figs Tetracentron sinense, pits on end walls of tracheids. 3. Pit membrane with relatively few porosities. 4. Highly porose pit membrane (resolution unfortunately suboptimal because pit membrane is recessed below the protruding pit borders). Scale bars in each figure 2 m.

4 188 AMERICAN JOURNAL OF BOTANY [Vol. 89 Figs Scanning electron microscopy photographs of perforation plates of vessel elements with extensive pit membrane remnants in perforations. 5. Paracryphia alticola, pit membranes nearly intact (a few tears and extraneous deposits present) with numerous porosities of various sizes. 6. Ascarina philippinense, pit membranes of perforations intact but with porosities ranging widely in diameter. 7. Berzelia cordifolia, pit membrane remnants mostly threadlike, but a pit with intact pit membrane at bottom of the perforation plate. 8. Euptelea polyandra, flakelike and strandlike pit membrane remnants in perforations. Scale bars in each figure 2 m.

5 February 2002] CARLQUIST AND SCHNEIDER TRACHEID VESSEL ELEMENT TRANSITION 189 Figs Scanning electron microscopy photographs of perforation plates of dicotyledons with moderate to minimal presence of pit membrane remnants. 9. Ascarina maheshwarii, porose pit membrane remnants at ends of perforations. 10. Cliftonia monophylla, strandlike or porose pit membrane remnants at ends of perforations. 11. Ascarina solmsii, threadlike pit membrane remnants at ends of perforations. 12. Ascarina brenesii, pit membrane remnants present only as a series of warts along the bars of the perforation plate. Scale bars in each figure 2 m.

6 190 AMERICAN JOURNAL OF BOTANY [Vol. 89 Fig. 13. Drawing that summarizes changes that occur when tracheids of a vesselless angiosperm (left) evolve into vessel elements and associated tracheids of a vessel-bearing angiosperm (right). Modified from Carlquist (1988). Although generalized, the tracheids (latewood at left, earlywood at right) like those shown can be found in Amborella, Tetracentron, and Trochodendron, and the wood cells of a vessel-bearing woody angiosperm (vessel element at left, associated tracheid at right) are like those of Chloranthaceae, Illicium, Euptelea, etc. The six changes illustrated here are as follows: (1) Pit membranes of end walls of tracheids are nonporose (not shown), then porose (drawing to right of tracheid end wall), with various degrees of pit membrane remnants in perforations as vessels originate (drawing to left of perforation plate of vessel element); meanwhile, pit membranes of lateral walls of vessels remain intact and nonporose, just as they are on lateral walls of tracheids (drawing in center, middle). (2) Conductive area of the end wall of the (wider) earlywood tracheid (at right in the pair at left) is greater than that of pits of tracheid lateral walls because the pits are larger on the end walls; this is interpreted as an incipient form of vessellike structure. As vessels originate, the differences in size of perforations as compared to lateral wall pits are more pronounced. (3) There are changes in morphology

7 February 2002] CARLQUIST AND SCHNEIDER TRACHEID VESSEL ELEMENT TRANSITION 191 observed in vessels of dicotyledons with no secondary xylem: this includes all genera of Nymphaeales except for Euryale and Ondinea (Schneider and Carlquist, 1995a, b, 1996a; Schneider et al., 1995). Porose pit membranes in end walls also occur in Nelumbo of Nelumbonaceae (Schneider and Carlquist, 1996b). Similarly, most ferns in which vessels have been reported have end walls in which pit membrane presence ranges from membranes with small pores to weblike pit membrane remnants to apparent absence of pit membranes (Carlquist and Schneider, 2001a, and the papers cited therein). Only a fraction of fern genera fall into the latter category. Another interesting exception to presence of pit membranes in perforations of vessels is mature wood of Myrothamnus, in which vessels with perforations, all of which are covered by unbroken and nonporose pit membranes, are intermixed with vessels that have no pit membranes at all (Carlquist, 1988a, p. 69). In the majority of instances in which pit membrane remnants are present, they cover 50% of the perforation area; the remnants may be weblike (Fig. 7), threadlike, or flakelike (Fig. 8), cover only ends of perforations (Figs. 9 and 10), or may be present as a few residual threads (Fig. 11) or warts (Fig. 12). Vessel elements with pit membrane remnants, some threadlike, some weblike, some composed of porose pit mebranes, can be found in monocotyledons (e.g., Carlquist and Schneider, 1998; Schneider and Carlquist, 1998). 2. Conductive area of end walls (perforation plate) is greater than that of an equivalent area of lateral wall pitting for vessel elements because pits (perforations) are larger and/or more densely placed on the end wall of vessel elements. In tracheids, end walls resemble lateral walls in pit size (and therefore conductive area per unit wall area). This definition is universally accepted (although rarely explicitly stated) with respect to vessel elements, not only in angiosperms (see note below), but in Ephedra and Welwitschia of the Gnetales (Carlquist, 1996b). This feature permits us to identify vessel elements (at least in the currently used version of that term) in all but a few vessel-bearing woods, and exceptions to this part of the definition are not cited here because woods with vessels that simulate imperforate tracheary elements in diameter are quite infrequent, are mostly in derived phylads of dicotyledons (e.g., Loricaria of Asteraceae; Carlquist, 1961), and can readily be shown to have vessel elements on the basis of other criteria. Exceptions. Tracheary elements with pits larger on end walls than on lateral walls have been observed in Tetracentron (Carlquist, 1996a), Trochodendron (compare Figs. 1 and 2) and Amborella (Carlquist and Schneider, 2001b), despite the fact that pit membranes are present in end walls as well as lateral walls. The pit area of end walls in tracheids of six of the eight genera of Winteraceae is larger than than of the lateral walls (see below). 3. Morphology of the perforation plate is perceptibly different from that of lateral-wall pitting in vessel elements, whereas end walls are essentially like lateral walls in tracheids with respect to morphology. Perforation plates of vessels that differ in morphology from lateral walls (e.g., a scalariform perforation plate on a vessel element that has alternate lateral-wall pitting) are familiar (one such difference, narrower borders on perforations than on lateral walls, is shown in the vessel element in Fig. 13). However, one component of this difference that has been taken for granted is the absence of pit membranes in the end walls that look morphologically different from the lateral walls. This is assumed to be true of all vessel elements, but tracheid end walls are thought to be comparable to lateral walls in tracheids in pit morphology. Exceptions. At least in some species in Belliolum, Bubbia, Drimys, Tahktajania, Tasmannia, and Zygogynum, there are scalariform pits in end walls of tracheids, whereas lateral walls have, with few exceptions, alternate circular pits (Carlquist, 1981, 1983a, b, 1988b, 1989, 2000). In two genera of Winteraceae, Exospermum (Carlquist, 1982) and Pseudowintera (Meylan and Butterfield, 1978), the circular bordered pits typical of lateral walls of tracheids in the family are also found on end walls, and thus pit morphology of end walls (as well as pit membrane conductive area per unit area of end walls) is about the same as that of the lateral walls. Note should be taken that differences between end walls and lateral walls in pitting occur in radial walls of tracheids of the above genera and therefore must be studied on radial wood sections if SEM is used. 4. Vessels have greater diameter than tracheids with which they are associated. This distinction, as mentioned above, is seen most readily in transections (some narrow vessels the same diameter as an imperforate tracheary element may be seen in transections of some wood samples, but since these narrow vessels have all of the other criteria of vessels and tend to occur in specialized phylads, the usefulness of this feature is not negated. The greater diameter of vessel elements than of tracheids in primitive phylads of dicotyledons can be considered a byproduct of division of labor. The expansion of the end wall, its perforation plate, and therefore the enhancement of conductive abilities of the perforation plate are so much greater than the conductive abilities of a tracheid end wall that there is little selective value for randomly distributed tracheids of various diameters in a wood (note the distinction made here between differences in latewood and earlywood radial tracheid diameters and the difference between the diameter of a vessel and the diameter of the tracheids that surround that vessel in such a wood as Quercus or Rosa). In vesselbearing dicotyledons that have tracheids and in Gnetales, tangential diameter of tracheids is relatively uniform in any given portion of a wood transection. Although tracheids undoubtedly do function as a subsidiary conductive system in a vesselbearing dicotyledon (Braun, 1970), their uniform narrow diameter supports the probability that mechanical strength is their main function. The dimorphism in diameter between ves- of perforations of end walls as compared to lateral walls as vessels originate. Narrower borders on perforations of the vessel, as compared to borders of lateral wall pits, is shown here, but other differences in morphology of lateral wall pits (e.g., transitional, circular pits) occur during the process of vessel origin. (4) Vessel elements become wider than the tracheids they accompany. The vessel element, at right, is wider than the tracheid associated with it. (5) Vessel elements are shorter than the tracheids they accompany. Although tracheids are monomorphic in length in a vesselless wood (left), vessel-bearing woods have vessel elements shorter than the imperforate tracheary elements they accompany (right). (6) Tracheids in a vesselless wood (left) are longer than vessel elements or imperforate tracheary elements in a vessel-bearing wood (right). In any given phylad, origin of vessels results in shortening not merely of vessel elements, but of the accompanying imperforate tracheary elements as well.

8 192 AMERICAN JOURNAL OF BOTANY [Vol. 89 sels and tracheids represents a release effected by division of labor. When one notes that conductive ability is calculated by the fourth power of the diameter of a capillary (the Hagen- Poiseuille equation; Zimmermann, 1983), the degree to which conduction is enhanced by having a vessel lumen as little as 50% wider than the lumen of an accompanying tracheid becomes evident. Vessels of primitive dicotyledons tend to be narrower than those of more specialized ones. The process of phylogenetic widening of vessels while imperforate tracheary elements remain narrow in a wood during the process of vessel origin in a phylad has not been documented directly, but can be demonstrated indirectly by the fact that vessels of primitive dicotyledons tend to be more angular in transection (Frost, 1930). The angularity reflects contact with fewer fibriform wood cells; in specialized woods, vessels are more frequently round in transection because the vessel is in contact with many more fibriform wood cells. One could think of narrowness of vessels in primitive dicotyledons as a vestige of the origin of vessels from tracheids, but in a vessel-bearing dicotyledon wood that contains tracheids, there is usually little difficulty in distinguishing vessels from the tracheids (Fig. 13, right). The diameters of vessels even in wood of such primitive vessel-bearing dicotyledons do not overlap appreciably with diameters of the tracheids. Exceptions. Vessels were suspected in roots of Sarcandra (Carlquist, 1987a) because in transections, the putative vessel elements were wider in diameter than the associated tracheids. The identity of these cells as vessel elements was confirmed by study of radial sections with SEM. In addition to being wider than tracheids, the putative vessels had pit membrane remnants of various extents in the perforations, but in pits of lateral walls, unaltered pit membranes were present; perforation plates also proved to have somewhat greater conductive area per unit wall area than the lateral-wall pitting. Thus, three criteria for recognition of vessel presence were realized. However, Takahashi (1988) studied the stems of Sarcandra, which are canelike, last only a few years, and do not develop as much secondary xylem as do the roots of Sarcandra. Takahashi s TEM sections revealed that in stems of Sarcandra, pores are present in pit membranes of end walls of tracheids. No other characteristics of vessels occurred in the tracheids he studied. This accounts for the contention by Swamy and Bailey (1950), who used only light microscopy, that Sarcandra was vesselless: they studied only stems from herbarium specimens (few specimens bearing roots are available). However, the tracheids of Amborellaceae, Trochodendraceae (including Tetracentron), and Winteraceae are uniform in tangential diameter. The wood of these three families has been universally regarded as vesselless. However, the presence of porose pit membranes in end walls (whereas lateral walls have nonporose pit membranes as far as is known) and the difference in conductive area (per unit area of wall) between end walls and lateral walls, and, in all but two genera of Winteraceae, difference in pitting morphology between end walls (scalariform) and lateral walls (circular pits) represent three character state departures from strictly tracheid expressions in the direction of vessel identity. One could make a case that tracheids in the abovementioned vesselless genera are actually not tracheids, but tracheary elements intermediate between tracheids and vessel elements. An entirely different case is represented by vessels that have the same diameter as tracheids: the very narrow vessels sometimes found in latewood and in diagonal bands of vessels (as seen in wood transections) in certain relatively specialized dicotyledons (Carlquist, 1987b). These narrow vessels may be identical to tracheids as seen in transection, but in macerations, they can readily be shown to have perforation plates. More importantly, the narrow diameter of these vessels is related to adaptation by wood to ecological conditions and is not a vestige of intermediacy between tracheids and vessel. In all respects other than diameter, these narrow vessels clearly qualify as vessels. Large numbers of vessels about as narrow as the imperforate tracheary elements they accompany may be seen in such specialized dicotyledons as Loricaria of the Asteraceae (Carlquist, 1961). 5. In vessel-bearing woods that also contain tracheids, vessel elements are shorter than the tracheids in any given sample, whereas in a vesselless wood, lengths of tracheids in any given sample would form a curve with a single mode. This principle is clearly evident from the data of Bailey and Tupper (1918), who calculated mean lengths for vessel elements and for imperforate tracheary elements of vesselbearing as well as for tracheids of vesselless dicotyledons, gymnosperms, and vascular cryptogams. The phenomenon of modally different lengths for vessel elements and imperforate tracheary elements in any given wood sample of vessel-bearing dicotyledons was summarized in a bar graph (Carlquist, 1975, p. 141). This phenomenon has not received appreciable comment in discussions of evolution of tracheids to vessel elements. The shorter length of vessel elements as compared to tracheids in a vessel-bearing wood (Fig. 13, right) can be ascribed to the value of shorter cells in confining air embolisms (Carlquist, 1988a, 2001). There is a value in greater length of imperforate tracheary elements, on the contrary, in providing mechanical strength, since greater length provides greater strength (Carlquist 1975, 1988a, 2001, and references cited therein). Exceptions. One can find instances of relatively primitive woods (featuring scalariform perforation plates) in which vessel elements have a longer mean length than the tracheids they accompany (Myrothamnus, Carlquist, 1976; Grubbia, Carlquist, 1977). More extensive sampling might reveal that the mean lengths of tracheids and vessel elements were about the same in these two genera, respectively. In any case, this instance is mentioned only to stress that these two genera do not represent presence of cells intermediate between tracheids and vessel elements. Instead, the relatively long vessel elements probably represent an exceptional degree of intrusiveness permitted by narrowness of the vessel elements in these genera. There is no question that in woods of very primitive vessel-bearing dicotyledons, the tracheid length averages only a little longer than vessel element length (e.g., times longer in Hedyosmum, Carlquist, 1992a). The closeness in lengths of the two cell types in this genus very likely does represent a vestige of origin of vessel elements from tracheids in Chloranthaceae. However, the vessel elements of Hedyosmum are shorter than the tracheids and also qualify for the definitions of vessel elements and the tracheids that accompany them given in the preceding four sections. 6. Tracheids of vesselless dicotyledons are much longer than either tracheids or vessel elements of vessel-bearing dicotyledons. Following origin of vessels from tracheids, the mean length of tracheary elements drops considerably. This statement is easy to demonstrate by using samplings of vesselless and vessel-bearing dicotyledons (Carlquist, 1975, p. 141). The explanation of this shift is that tracheids, to be conductively equivalent to vessel elements, must be relatively

9 February 2002] CARLQUIST AND SCHNEIDER TRACHEID VESSEL ELEMENT TRANSITION 193 long in order to have long end walls capable of serving for cell-to-cell conduction. Another consequence of a long conductive cell is the presence of fewer cross-walls, which would slow flow between superposed cells. As mentioned above, shorter vessel elements are superior to long vessel elements in containing air embolisms, so having vessel elements shorter than the tracheids of vesselless dicotyledons is understandable. However, there is no known family of dicotyledons in which both vesselless and vessel-bearing species occur, and which therefore could demonstrate this phylesis, although Sarcandra comes close to illustrating it. Origin of vessels in monocotyledons and ferns may represent an exception to the principle stated above in that the resultant vessel elements may not be shorter than tracheids of the ancestor, as tracheary element length in ferns is related primarily to the degree of elongation of the organ in which they are located (Carlquist, 1975). The possibility that vessel element length remains the same during phyletic origin of vessels in ferns and monocotyledons is an intriguing possibility that deserves further investigation. This possibility shows that this consequence of vessel origin is not a corollary of the fifth criterion above (dimorphism in length), because a monocotyledon organ in which vessels originate very likely does not contain associated tracheids after vessel origin has occurred. If a phylad loses vessels as hypothesized for several by Young (1981), the selective value of conversion from dimorphism in length of the vessel-bearing state to monomorphism in length is not obvious unless one assumes that an end wall length optimal for conduction or other reasons is thereby accomplished. Character state changes following vessel origin Any change in a character state after vessels have originated (specialization of vessel elements) is not of concern in this discussion. Because character state changes subsequent to vessel origin can represent vestiges of tracheid characteristics, they may be confusing to those unfamiliar with wood anatomy. For example, in vessels of primitive dicotyledons, borders are present on the bars of scalariform perforation plates. Borders are also present on the foraminate perforations of Gnetales (Carlquist, 1996b). The borders on perforations undoubtedly represent vestiges of the similarity between pits on lateral walls and on end walls of tracheids that preceded vessel elements in a phylad. However, all known instances in which borders are present on perforations of end walls of vessels have all known characteristics of vessel elements and therefore follow vessel origin and are not applicable to our examination of the character state changes during origin of vessel elements from tracheids. DISCUSSION With respect to cladistics, the question of primary interest is whether the six criteria of vessel origin given above overlap or can theoretically be achieved independently (Fig. 13). As the examples cited in the exceptions indicate, examples can be cited of various genera in which one, two, or three of these features are characteristic, respectively. For example, some ferns and the Sarcandra stem tracheids have end wall pitting like that of lateral walls but pit membranes on end walls contain porosities sometimes large ones. In six genera of Winteraceae, end walls of tracheids differ in pit morphology from that of lateral walls, pit area of end wall pits is greater than that of lateral wall pits, and presence of pores in pit membranes of end wall pits contrasts with lack of pores in pit membranes of lateral walls. Indeed, if these three features constitute criteria of vessel presence, one may ask whether tracheids of Winteraceae should be considered an intermediate type of vessel element. If such winteraceous tracheids are considered an intermediate type of vessel element, are Winteraceae primarily vesselless or secondarily vesselless? The question may be moot if tracheids of Winteraceae are not really referable to either category. More importantly, the contrast between vesselless and vessel-bearing can no longer be considered two character states of a single feature. Vessel presence is a syndrome of a series of changes, probably governed by independent genes (although functionally related in effect in some cases). Each of the six changes involved in this syndrome might be considered as constituting an independent character. With respect to cladistic treatment, one might consider that at least the first three of the criteria discussed above are worthy of treatment (the most tracheidlike state 0 in the below designations). I. Pit membranes of end walls: 0, lacking in porosities (Fig. 1); 1, with minute pores (whereas membranes of lateral wall pits lack pores) (Figs. 3, 4, and 5); 2, pit membrane 50% intact (Fig. 6); 3, pit membrane of remnants covering 50% of pit cavity (Figs. 7 11); 4, pit membrane lacking because of natural lysis (Fig. 12). Although five character states are justified in terms of pit membrane conditions reported to date in the literature, analysis may not always be easy, and virtually all species in which pit-membrane remnants have been reported are insufficiently known. We do not underestimate the difficulty of attempting to designate which of the above character states apply in any particular species or even sample. The situation is easiest for genera in which no pit membrane remnants (state 4) are present. There are also genera, such as Carpenteria, Eupomatia, Sphenostemon, and Symplocos, in which pit membrane remnants are few and therefore not very variable (S. Carlquist, unpublished data). On the other end, few genera are likely to show, with any uniformity, the presence of no pores or of minute pores in pit membranes of tracheid end walls, and these expressions seem likely to be relatively uniform for a genus. Thus, the designation of states 0 and 1 should provide little problem. However, in genera in which character states 2 and 3 can be observed, a range of expressions is likely to be expressed, but we see little difficulty in designating what is typical for that sample. For example, in Illicium floridanum (Schneider and Carlquist, in press), we have attempted to document the range in pit membrane presence. In that species, we were able to find all character states from 0 to 4, although the extremes were infrequent and the vast majority of perforation plates had pit membranes referable to character state 3. If features of primitive vessel elements represent degrees of intermediacy between tracheids (narrowly defined, with no vessel-like features) and vessels (as traditionally defined), that is precisely what one would expect in an evolutionary process in progress. Moreover, this situation poses interesting problems with respect to physiology. Why, in a wood like Illicium floridanum, do vessel elements vary with respect to pit membrane presence? Is selective value for clear perforations so low that except on unusual days, low conductive rates suffice? Naturally wilting plants of Illicium cambodianum on an unusually hot day (Carlquist, 1975; Plate 9) are noteworthy; this occurrence was recorded before the extensive presence of pit membrane remnants was recorded for Illicium

10 194 AMERICAN JOURNAL OF BOTANY [Vol. 89 (Carlquist, 1992b). Does the conductive stream play a role in enlarging pores or in removing pit membrane portions in genera such as Illicium? Are some genera with pit membrane remnants enzymatically less efficient in pit membrane removal? Do genera with more numerous bars per perforation plates tend to have more pit membrane remnants than those with fewer bars per perforation plates, and if so, are the wider expanses of pit membranes in the latter species more easily removed? Evolutionary questions such as these are more interesting that the problem of assigning character state numbers to a specimen, desirable though that activity may be. II. Pits of end walls: 0, with the same conductive area per unit wall area than pits of lateral walls (Fig. 13, earlywood tracheid at extreme left); 1, with more conductive area per unit wall area than pits of lateral walls (compare Figs. 1 and 2). III. Pit morphology of end walls: 0, identical with pitting of lateral walls; 1, different from that of end walls (e.g., narrower borders on perforations than on lateral walls; e.g., vessel element, Fig. 13, right). IV. Tracheary element diameter: 0, uniform, radial diameter of all tracheary elements alike; 1, bimodal, vessel elements wider than the imperforate tracheary elements they accompany in a wood (Fig. 13, right). Character state designations could be made for vessel criteria 5 and 6, but those do not seem readily applied to what is known of observed xylary characteristics. One could attempt to condense the above information into a single trinary expression (0, tracheids present; 1, tracheary elements intermediate between tracheids and vessel elements; 2, vessel elements present). The potential convenience of that treatment should be tested against the possibility that an intermediate category would probably contain a series of different phenomena and character expressions. Note should be taken that the five character states of criterion I (above) require information from SEM or TEM (preferably SEM) with methods likely to produce minimal artifact formation. We believe that our comparison of methods used in Illicium floridanum amd the results of Butterfield and Meylan (1980) and Meylan and Butterfield (1978), whose work has not received challenge, validate our belief that artifact formation in the figures of the present paper is minimal. The use of a character that can only be seen with an electron microscope may seem objectionable to those who construct data matrices only on the basis of what can be seen without magnification or with the aid of a light microscope. However, to neglect the nature of pit membranes is to ignore the physiological and evolutionary significance of pit membrane morphology in vascular plants. Undoubtedly, the greater the area of a pit cavity free from pit membrane or pit membrane remnants, the less the impediment to the conductive stream and for large pores, the less the tendency to confine embolisms to a single cell. If these considerations are rejected, the resultant cladogram may be less informative, and the designation of vessel absence as primary or secondary may be less accurately rendered. As suggested earlier (Carlquist, 1975), persistence of tracheidlike characteristics in vessel elements probably indicates unbroken occupany of mesic habitats by particular phylads. Entry into more seasonal regimes has sparked specializations of vessel elements (e.g., the simple perforation plate). Although attention needs to be focused on the basal angiosperms, there are notably primitive vessel elements (i.e., with extensive pit membrane remnants) in eudicot families as well, e.g., Clethraceae, Dilleniaceae, and Sarraceniaceae. For that reason, we are investigating vessel morphology (especially as seen with SEM) in more basal families of the major clades of dicotyledons. LITERATURE CITED BAILEY, I. W The development of vessels in angiosperms in morphological research. American Journal of Botany 31: BAILEY, I. W., AND W. W. TUPPER Size variation in tracheary cells. I. A comparison between the secondary xylems of vascular cryptogams, gymnosperms, and angiosperms. Proceedings of the American Academy of Arts and Sciences 54: BRAUN, H. J Funktionelle Histologie der sekundären Sprossachse. I. Das Holz. Handbuch der Pflanzenanatomie 9(1): BUTTERFIELD, B. G., AND B. A. MEYLAN Cell wall hydrolysis in the tracheary elements of the secondary xylem. In P. Baas [ed.], New perspectives in wood anatomy, Martinus Nijhoff/Dr. W. Junk, The Hague, Netherlands. CARLQUIST, S Wood anatomy of Inuleae (Compositae). Aliso 5: CARLQUIST, S Ecological strategies of xylem evolution. University of California Press, Berkeley, California, USA. CARLQUIST, S Wood anatomy of Myrothamnus flabellifolia (Myrothamnaceae) and the problem of multiperforate perforation plates. Journal of the Arnold Arboretum 57: CARLQUIST, S Wood anatomy of Grubbiaceae. Journal of South African Botany 43: CARLQUIST, S Wood anatomy of Zygogynum (Winteraceae); field observations. Adansonia n. s. 3: CARLQUIST, S Exospermum stipitatum (Winteraceae): observations on wood, leaves, flowers, pollen, and fruit. Aliso 10: CARLQUIST, S. 1983a. Wood anatomy of Bubbia (Winteraceae), with comments on origin of vessels in dicotyledons. American Journal of Botany 70: CARLQUIST, S. 1983b. Wood anatomy on Belliolum; a note on floral opening. Journal of the Arnold Arboretum 64: CARLQUIST, S. 1987a. Presence of vessels in Sarcandra (Chloranthaceae): comments on origin of vessels in dicotyledons. American Journal of Botany 70: CARLQUIST, S. 1987b. Diagonal and tangential vessel aggregations in wood: function and relationship to vasicentric tracheids. Aliso 11: CARLQUIST, S. 1988a. Comparative wood anatomy. Springer Verlag, Berlin, Germany. CARLQUIST, S. 1988b. Wood anatomy of Drimys s. str. (Winteraceae). Aliso 12: CARLQUIST, S Wood anatomy of Tasmannia; summary of wood anatomy of Winteraceae. Aliso 12: CARLQUIST, S Wood anatomy of Ascarina (Chloranthaceae) and the tracheid vessel element transition. Aliso 12: CARLQUIST, S. 1992a. Wood anatomy of Hedyosmum (Chloranthaceae) and the tracheid vessel element transition. Aliso 13: CARLQUIST, S. 1992b. Pit membrane remnants in perforation plates of primitive dicotyledons and their significance. American Journal of Botany 79: CARLQUIST, S. 1996a. Wood anatomy of primitive angiosperms: new perspectives and syntheses. In D. W. Taylor and L. J. Hickey [eds.], Flowering plant origin, evolution, and phylogeny, Chapman and Hall, New York, New York, USA. CARLQUIST, S. 1996b. Wood, bark, and stem anatomy of Gnetales: a summary. International Journal of Plant Sciences 157(6 suppl.): S58 S76. CARLQUIST, S Wood and bark anatomy of Takhtajania (Winteraceae): phylogenetic and ecological implications. Annals of the Missouri Botanical Garden 87: CARLQUIST, S Comparative wood anatomy, 2nd ed. Springer Verlag, Berlin, Germany. CARLQUIST, S., AND E. L. SCHNEIDER Origins and nature of vessels in monocotyledons. 5. Araceae subfamily Colocasioideae. Botanical Journal of the Linnean Society 128: CARLQUIST, S., AND E. L. SCHNEIDER. 2001a. Vessels in ferns: structural, ecological, and evolutionary significance. American Journal of Botany 88: CARLQUIST, S., AND E. L. SCHNEIDER. 2001b. Vegetative anatomy of the

11 February 2002] CARLQUIST AND SCHNEIDER TRACHEID VESSEL ELEMENT TRANSITION 195 New Caledonia endemic Amborella trichopoda: relationships with the Illiciales and implications for vessel origin. Pacific Science 55: ESAU, K Plant anatomy, 2nd ed. John Wiley, New York, New York, USA. ESAU, K., AND W. B. HEWITT Structure of end walls in differentiating vessels. Hilgardia 13: FEILD, T. S., M. A. ZWEINIECKI, T.BRODRIBB, T.JAFFRÉ, M.J.DONOGHUE, AND B. M. HOLBROOK Structure and function of tracheary elements in Amborella trichopoda. International Journal of Plant Sciences 161: FROST, F. H Specialization in secondary xylem in dicotyledons. I. Origin of vessel. Botanical Gazette 89: HERENDEEN, P. S., AND R. B. MILLER Utility of wood anatomical characters in cladistic analyses. IAWA Journal 21: IAWA COMMITTEE ON NOMENCLATURE Multilingual glossary of terms used in describing wood anatomy. Verlagsbuchanstalt Konkordia, Winterthur, Switzerland. MEYLAN, B., AND B. G. BUTTERFIELD The structure of New Zealand woods. DSIR Bulletin 222, Science Information Service, DSIR, Wellington, New Zealand. SCHNEIDER, E. L., AND S. CARLQUIST. 1995a. Vessels in the roots of Barclaya rotundifolia (Nymphaeaceae). American Journal of Botany 82: SCHNEIDER, E. L., AND S. CARLQUIST. 1995b. Vessel origins in Nymphaeaceae: Euryale and Victoria. Botanical Journal of the Linnean Society 119: SCHNEIDER, E. L., AND S. CARLQUIST. 1996a. Vessels in Brasenia (Cabombaceae): new perspectives on vessel origin in primary xylem of angiosperms. American Journal of Botany 83: SCHNEIDER, E. L., AND S. CARLQUIST. 1996b. Vessels in Nelumbo (Nelumbonaceae). American Journal of Botany 83: SCHNEIDER, E. L., AND S. CARLQUIST Origins and nature of vessels in monocotyledons. 4. Araceae subfamily Philodendroideae. Journal of the Torrey Botanical Society 125: SCHNEIDER, E. L., AND S. CARLQUIST. In press. Pit membrane remnants in perforation plates of Illicium floridanum. Sida. SCHNEIDER, E. L., S. CARLQUIST, K. BEAMER, AND A. KOHN Vessels in Nymphaeaceae: Nuphar, Nymphaea, and Ondinea. International Journal of Plant Sciences 156: SWAMY, B. G. L., AND I. W. BAILEY Sarcandra, a vesselless genus of the Chloranthaceae. Journal of the Arnold Arboretum 31: TAKAHASHI, A Morphology and ontogeny of stem xylem elements of Sarcandra glabra (Thunb.) Nakai (Chloranthaceae). Botanical Magazine (Tokyo) 101: YOUNG, D. A Are the angiosperms primitively vesselless? Systematic Botany 6: ZIMMERMANN, M. H Xylem structure and the ascent of sap. Springer Verlag, Berlin, Germany.