Angiosperm Wood Evolution and the Potential Contribution of Paleontological Data.
Angiosperm wood anatomy has received considerable attention from both neobotanists and paleobotanists for well over 100 years, and a considerable body of literature is now available that describes wood anatomy in living and fossil angiosperms. In addition to this descriptive literature, significant efforts have been directed at developing evolutionary interpretations of patterns observed in the wood anatomical data. However, most of these studies have employed phenetic methodologies, or the evolutionary explanations have been based on intuitive interpretations of the relative adaptive values of "primitive" versus "advanced" anatomical features. Relatively few cladistic studies have included wood anatomical characters (e.g., Dahlgren & Bremer, 1985; Loconte & Stevenson, 1991; Hufford, 1992). As a consequence, wood anatomy of both living and fossil angiosperms has had relatively little influence on modern cladistic phylogenetic studies of angiosperm relationships and evolution.
In our view, evolutionary interpretations of wood anatomical patterns are best framed in an explicit phylogenetic context. In this paper we: 1) summarize comparative wood anatomical studies based on extant taxa and the relevant data from the fossil record; 2) discuss the issue of "evolutionary trends" derived from broad correlation studies versus interpretations of character evolution based on specific explicit phylogenetic hypotheses; 3) discuss practical issues relative to utilizing wood anatomical characters in cladistic analyses, especially variability within taxa and quantitative versus qualitative characters; 4) present wood anatomical data for Magnoliidae and "lower" Hamamelididae in a cladistic context to evaluate traditional evolutionary interpretations of character evolution; and 5) outline our view of how fossil woods can make the greatest contribution to hypotheses of angiosperm evolution and phylogeny.
III. Comparative Wood Anatomy and the Major Trends of Xylem Evolution
Features of angiosperm wood anatomy have long been viewed as a potential source of systematically informative characters that can be employed to provide an independent assessment of hypothesized evolutionary relationships for a particular family, genus, or species of flowering plant. For example, wood anatomy of Ticodendron was recently used by Cariquist (1991) in evaluating the systematic relationships of this new genus and family (Gomez-Laurito & Gomez P., 1989).
This view of the potential utility of angiosperm wood anatomical data arose out of the great body of comparative anatomical data that has been amassed by numerous wood anatomists, most recently Sherwin Carlquist, who has made numerous contributions, particularly for many small but systematically significant families of magnoliid dicotyledons (see Carlquist 1988b, 1996 for extensive bibliographies). In the early and mid-1900s I. W. Bailey, Tupper, Frost, Kribs, Cheadle, and others compiled anatomical data for many angiosperm taxa, and analyses of these data documented correlations between diverse wood anatomical characters, from which they hypothesized evolutionary explanations for these correlated features (Bailey & Tupper, 1918; Frost, 1930a, 1930b, 1931; Barghoom, 1940, 1941a, 1941b; Bailey, 1944). These are often referred to as the "salient trends" or "major trends of specialization" in angiosperm woods (Bailey, 1944; Carlquist, 1988b). based on functional and ecological arguments, many of these trends are viewed as irreversible evolutionary processes (Bailey, 1944: 424; Carlquist, 1988b). The correlated features include the following: vessel element length with perforation plate structure, vessel element length with lateral pitting arrangement, perforation plate structure or vessel element length with axial parenchyma distribution, and perforation plate structure with ray structure (Frost, 1930a, 1930b, 1931; Kribs, 1935, 1937; Barghoorn, 1940, 1941a, 1941b; see Carlquist, 1988b for details).
Of the wide array of wood features, the characters that are most often employed in assessments of systematic relationships include: perforation plate structure, vessel element length, intervascular pitting, ray composition, axial parenchyma distribution, and pit structure of imperforate elements (tracheids, fibers, etc.). The primary rationale for employing these particular characters has been the prevailing view that the ancestor of the angiosperms had wood with long tracheary elements and that the pattern of evolution within angiosperms is irreversible. Bailey (1957) cautioned that conclusions regarding relationships should be based on the totality of evidence rather than just on wood anatomy. He wrote that these evolutionary trends, "when considered by themselves, have been most reliable and significant in negations" of hypothesized relationships (Bailey, 1957: 250). If wood anatomical data are to make a significant contribution in developing hypotheses of phylogenetic relationships among flowering plants, they need to be utilized in a cladistic context. Until recently little effort had been devoted to documenting actual historical patterns for these classical wood anatomical patterns or to evaluating these characters in an explicit phylogenetic context.
IV. Wood Anatomical Patterns in the Fossil Record
Data from fossil angiosperm woods have been compiled and evaluated by Wheeler and Baas (1991, 1993) to determine whether the "major trends" are apparent in the fossil record. Wheeler and Baas compiled data for fossil dicotyledonous woods from the Cretaceous and Tertiary on a global scale (approx. 1200 records) and examined chronological patterns for the incidence of such features as vessel element length, scalariform perforation plates versus simple plates, opposite and scalariform versus alternate intervessel pitting, imperforate element pit structure, heterocellular versus homocellular rays, and axial parenchyma distribution patterns [ILLUSTRATION FOR FIGURE 1 A-D OMITTED]. Their study documents that, in general, the fossil record provides chronological corroboration of many of these hypothesized evolutionary trends. However, woods with exclusively simple perforation plates and alternate intervessel pitting are among the earliest fossil woods known, and the incidence of several features - for example, scalariform and simple perforation plates [ILLUSTRATION FOR FIGURE 1A OMITTED], intervessel pitting [ILLUSTRATION FOR FIGURE 1B OMITTED], and imperforate element pit structure [ILLUSTRATION FOR FIGURE 1C OMITTED] - remains essentially unchanged throughout most of the Tertiary. Did these anatomical characters attain their modern patterns of distribution and incidence by the Early Tertiary and subsequently remain unchanged, or are there essentially offsetting rates of "forward" and "reversed" changes in these features during much of the Tertiary? The fossil record provides general support for the "major trends" of xylem evolution but also raises additional questions.
V. Major Trends versus Details of Character Evolution
Neither interpretations of character distribution patterns based on correlated features across a broad range of taxa nor character incidence frequencies derived from paleontological data can reveal details of character evolution, either through time or within clades. In the absence of phylogenetic context, it is impossible to determine from these data whether character evolution has proceeded strictly according to the hypothesized major trends or whether there have been evolutionary reversals or other changes not in accord with these trends in some angiosperm clades.
Most of the difficulties arise because many wood anatomical characters are variable within flowering plants at many levels. Such variability considerably complicates the interpretation of simple general trends. For example, perforation plate structure has figured prominently in earlier discussions of wood evolution, but this feature varies considerably, both within and among many angiosperm families. Families in which both scalariform and simple perforation plates have been documented in the same species, among different species, or both, are listed in Table I. Similarly, Table II lists families in which variation in imperforate elements occurs within or among species in a family. Such variability raises several important questions. For instance, has the evolution of perforation plate structure in all of these families conformed to the Baileyan pattern of long scalariform plates evolving to shorter scalariform plates to simple perforation plates, or have reversals in this sequence occurred in some of these families? For imperforate elements, is the Baileyan pattern of true tracheids transforming to fiber-tracheids and then to libriform fibers the only transformation series, or is there evidence of other evolutionary patterns?
VI. Testing the Baileyan Trends: Wood Anatomical Characters in Cladistic Analyses
To date, wood anatomical characters have been used in relatively few cladistic analyses (e.g., Young, 1981; Dahlgren & Bremer, 1985; Baas et al., 1988; Donoghue & Doyle, 1989; Anderberg, 1992; Hufford, 1992; Hufford & Dickison, 1992; Zhang, 1992; Terrazas, 1994; Keller et al., 1996; Noshiro & Baas, 1998; Klaassen, 1999). A manuscript specifically addressing the utility of wood anatomical characters in phylogenetic analyses will be presented elsewhere (Herendeen, in prep.). Perhaps the most notable cladistic study that addressed wood evolution was that of Young (1981). Contrary to the traditional view that flowering plants are primitively vesselless, based on a simple parsimony argument Young concluded that the absence of vessels in five lineages of flowering plants is a derived feature. Although this study has been criticized from various perspectives (Meeuse, 1982; Carlquist, 1983b, 1987; Riggins & Farris, 1983), and the demonstration of vessels in Sarcandra (Chloranthaceae; Carlquist, 1987) reduces the number of multiple vessel losses or origins that must be hypothesized, Young's conclusion that vessels have been lost in these groups has been supported by other studies (Dahlgren & Bremer, 1985; Donoghue & Doyle, 1989; Loconte & Stevenson, 1991; Doyle & Donoghue, 1993).
Arguments in favor of the less parsimonious interpretation of multiple vessel origins (vs. fewer vessel losses) have been made on functional, structural, and adaptive grounds (e.g., Carlquist, 1983b, 1987, 1996; Baas & Wheeler, 1996), but the reverse of many of these arguments also has been presented (e.g., Donoghue & Doyle, 1989: 32). Additional data, such as from xylem differentiation, should be obtained to address the question of whether not having vessels is a basic or derived feature in flowering plants, but ultimately such questions need to be framed in terms of hypotheses and tested against the detailed distribution of other features.
The most rigorous test of the hypothesized major trends in angiosperm wood evolution will be realized through comprehensive cladistic analyses that include wood anatomical characters among the suite of characters used in the analysis (Herendeen, work in progress), but preliminary insights can emerge from an examination of the distribution of several wood anatomical characters in families of Magnoliidae and "lower" Hamamelididae against current cladistic hypotheses of their interrelationships.
A. WOOD STRUCTURE IN MAGNOLIIDAE AND "LOWER" HAMAMELIDIDAE
Phylogenetic relationships among families of Magnoliidae and "lower" Hamamelididae have been investigated by Lammers et al. (1986), Donoghue and Doyle (1989), Loconte and Stevenson (1991), Chase et al. (1993), and Doyle and Donoghue (1993). We will use three different topologies, one from Chase et al. (1993) and two from Doyle and Donoghue (1993), including two alternative rootings for the angiosperms (Magnoliales and "paleoherbs") to examine the distribution patterns of the wood characters and their evolutionary implications.
This comparison is presented primarily in tabular form (Table III). Many of the data came from descriptions given by Metcalfe and Chalk (1950) and Metcalfe (1987), and from a series of papers on wood of the Magnoliidae (Canright, 1955; Wilson, 1960; Shutts, 1960; Carlquist, 1981, 1982a, 1982b, 1983a, 1983b, 1983c, 1984, 1988a, 1989a, 1989b, 1990a, 1990b, 1992a, 1992b, 1993; Takahashi, 1985; Chen et al., 1993). The data were further supplemented by studies of slides from the Bailey-Wetmore Laboratory (Harvard University) and the David A. Kribs collection (North Carolina State University). The wood anatomical features presented here, and many others, are illustrated by Carlquist (1988b) and the IAWA Committee (1989).
Because the vessel element has been emphasized in evolutionary studies, considerable detail is available on features of vessel elements, such as length, perforation plate structure, and vessel-vessel pitting. It needs to be emphasized that many quantitative wood features vary with stem diameter. Wood produced by a young vascular cambium differs quantitatively from wood produced by an older vascular cambium (Panshin & DeZeeuw, 1980). Vessel element diameter and length usually increase with cambial age, whereas vessel frequency decreases. In many woods, ray width and height change with increases in stem diameter. Consequently, these features must be used with considerable caution in phylogenetic analyses. Evidence to date indicates that characters less affected by stem diameter are vessel distribution and grouping, perforation plate type (although in woods with both simple and scalariform plates, the incidence of scalariform plates decreases with increases in stem diameter), type of intervessel and vessel-ray parenchyma pitting, imperforate element pitting, and distribution of axial parenchyma (Stem & Greene, 1958; Takahashi, 1985). Vessel element features (length, diameter, frequency, groupings, arrangement, and perforation type) also are correlated with environment and habit (cf. Baas, 1986; Carlquist, 1988b; Zhang & Baas, 1992). Thus, other wood anatomical features, such as axial parenchyma distribution and vessel-ray parenchyma pits, which may be less affected by environment or habit, should be used in addition to vessel characteristics to evaluate relationships.
Of the taxa included here, the orders Magnoliales and Laurales show the greatest diversity of wood anatomical features and considerable heterogeneity within families (Table III). For example, there are families in both orders with exclusively scalariform perforations, mixed scalariform and simple perforations, and exclusively simple perforations. Likewise, there are families with exclusively scalariform intervessel pits, scalariform to opposite intervessel pits, and exclusively alternate intervessel pits. Of the 32 families treated here, ten are polymorphic for intervascular pit distribution, and eight are polymorphic for imperforate element pit structure. The Monimiaceae s.1. (e.g., Cronquist, 1981) is the most variable of the lauralean families, both morphologically and wood anatomically. In our summary we utilize a narrower definition of Monimiaceae. We present wood features for Monimiaceae s.str. (incl. Hortonioideae), Atherospermataceae, and Siparunaceae.
Although vessel element perforations and pits are not useful for distinguishing between the Magnoliales and Laurales, patterns of parenchyma distribution, vessel-ray parenchyma pitting, and frequency of uniseriate rays differ between Magnoliales (scanty diffuse to apotracheal banded, reduced borders but of uniform shape, few uniseriates) and Laurales (scanty to abundant paratracheal, reduced borders to irregular shapes, frequent uniseriates). The Canellaceae are anomalous within the Magnoliales in that they have unilateral paratracheal parenchyma, they lack apotracheal parenchyma, their vessel-ray parenchyma pits are half-bordered and similar to intervessel pits, and they commonly have crystals in procumbent and upright ray cells, a feature not seen in any other family of the Magnoliales. Recent cladistic analyses found that the placement of Canellaceae in the Magnoliales is only weakly supported (Chase et al., 1993; Qiu et al., 1993).
Although it has been suggested that the Chloranthaceae are related to the Piperales (e.g., Cronquist, 1981), this relationship has not been supported in recent cladistic analyses. The Chloranthaceae are grouped with the Trimeniaceae in the Laurales in the cladograms presented here (based on Donoghue & Doyle, 1989; Chase et al., 1993). The Chloranthaceae and Piperaceae differ in significant features of both the axial and radial systems, and there are very few shared character states. These families differ in perforation plate type, intervessel pitting, vessel grouping, and axial parenchyma distribution (Table III). Additionally, the Piperaceae and Lactoridaceae are characterized by a "vessel restriction pattern" (i.e., vessels are not in contact with rays; Carlquist, 1993), which does not occur in the Chloranthaceae.
B. IMPLICATIONS FOR CHARACTER EVOLUTION
Many wood anatomical features are quantitative, and both quantitative and qualitative discrete characters are often highly variable within the families treated here. Extensive variation within families for characters such as perforation plate structure, intervessel pit arrangement, vessel element length, and imperforate element pit structure imply substantial homoplasy for these characters. As a means of objectively evaluating the evolutionary history of wood features among families of Magnoliidae and lower Hamamelididae we defined selected characters in a context appropriate for cladistic analyses (Table IV) and scored the families for these characters (Table V) based on the descriptive data in Table III. We then mapped these characters onto previously published cladograms to examine the implications for character evolution among these families [ILLUSTRATION FOR FIGURES 2-5 OMITTED]. Several cladistic studies have included many of the families addressed here (Donoghue & Doyle, 1989; Loconte & Stevenson, 1991; Chase et al., 1993; Doyle & Donoghue, 1993). Although each of these studies can be criticized on technical grounds for one reason or another, we have no basis for preferring one over the others. We elected to utilize three different topologies, one from Chase et al. (1993) and two from Doyle and Donoghue (1993), one with the cladogram rooted near Magnoliales and the other rooted near the "paleoherbs." These three topologies present contrasting implications for character evolution. Because Doyle and Donoghue (1993) and Chase et al. (1993) did not include the same suite of families as each other or as treated here, we have modified their cladograms to accommodate certain groups, most notably the segregate families of Monimiaceae.
[TABULAR DATA FOR TABLE III OMITTED]
1. Vessel Presence/Absence
It is evident from all three topologies that vessel absence is most parsimoniously interpreted as derived within angiosperms and that vessels have been lost three times among the families treated here [ILLUSTRATION FOR FIGURE 2 OMITTED]. Even if one were to specify that angiosperms were primitively vesselless, or to root angiosperms using one of the vesselless groups (Amborellaceae, Winteraceae, Tetracentraceae-Trochodendraceae), the most parsimonious interpretation of the remaining vesselless taxa would remain unchanged: vessels have been lost in these lineages. Although arguments against vessel loss have been presented elsewhere (see the review in Baas & Wheeler, 1996), cladistic methodology and the principle of parsimony leads inevitably to the conclusion that vessels have been lost in some angiosperm taxa. Additional structural and developmental data may be sought to pursue this issue further. More definitive data may come from the fossil record of flowering plants. If angiosperms were originally vesselless, then one should expect to find vesselless angiosperm wood among the earliest fossil angiosperm woods, which so far is not the case, although the available data are few (Wheeler & Baas, 1991). Paleobotanical data may also be found to refute the hypothesis of the primitively vesselless angiosperm. A vessel-bearing fossil wood that can be tied unequivocally to one of the extant vesselless lineages would be strong support for vessel loss. Thus, the currently available data do not support the hypothesis that Amborellaceae, Winteraceae, and Tetracentraceae-Trochodendraceae are primitively vesselless, and further inquiry into this issue should be directed at discovering new data that can refute one of the competing hypotheses.
[TABULAR DATA FOR TABLE V OMITTED]
2. Perforation Plate Structure
Mapping the distribution of perforation plate structure provides graphic illustration of the high degree of homoplasy in this character [ILLUSTRATION FOR FIGURE 3 OMITTED]. At least nine steps are required in the Chase topology, and at least ten steps are required in the magnolialean and paleoherb topologies to explain the distribution pattern of this character. These figures do not include character evolution within the five families that are polymorphic for perforation plate structure. Baas (1993) and Baas and Wheeler (1996) also found a high degree of homoplasy for perforation plate type when the distribution of this character was mapped onto existing cladograms. All three topologies demonstrate that simple perforation plates have arisen numerous times among these families, which is consistent with traditional interpretations of character evolution. However, the paleoherb tree [ILLUSTRATION FOR FIGURE 3C OMITTED] is inconsistent with traditional interpretations in its implication that exclusively simple perforation plates are plesiomorphic and that scalariform plates are derived within angiosperms. Furthermore, in all three topologies the scalariform perforation plates found in some members of Saururaceae are most parsimoniously interpreted as derived relative to simple perforation plates, which are shared by the remaining members of Saururaceae and other paleoherb families (Piperaceae, Aristolochiaceae, Lactoridaceae).
3. Intervessel Pitting
Although the implications for the evolution of intervessel pit distribution patterns differ among the three topologies, exclusively scalariform intervessel pitting is most parsimoniously interpreted as being derived in all three phylogenetic hypotheses, which is not consistent with the Baileyan trends [ILLUSTRATION FOR FIGURE 4 OMITTED]. A further departure from the Baileyan model is implicated in the paleoherb tree, in which alternate intervessel pitting is hypothesized to be plesiomorphic among the families included here. Less dramatic, but still counter to the Baileyan model, is the implication from the Chase tree that transitional pitting (scalariform and opposite pitting in the same vessel element) is the plesiomorphic condition. Although the optimization of this character on the deeper branches of the magnolialean tree is ambiguous (transitional and alternate pitting are equally parsimonious), scalariform pitting is most parsimoniously interpreted as being derived. There is considerable homoplasy in this character among these families. In addition, ten families are polymorphic for this character, eight of which exhibit exclusively scalariform pitting among some members. Paleontological data may help to clarify wood anatomical features among early members of these families and thus will reduce ambiguity due to polymorphic coding.
4. Imperforate Element Pit Structure
Distribution patterns for imperforate element pit structure imply that pits with indistinct borders have evolved numerous times among the families included here [ILLUSTRATION FOR FIGURE 5 OMITTED]. Although this reduction of pit borders is in accord with the Baileyan trends, all three cladogram topologies also demonstrate the reverse transition. Within the "paleoherb" families, the Saururaceae and Aristolochiaceae are polymorphic for this character, and the phylogenetic context provided by these cladograms implies that the distinctly bordered pits in some Saururaceae and Aristolochiaceae represent reversals.
The most significant source of ambiguity, especially in family-level analyses, for the kind of analyses presented here is character polymorphism within many families. The ideal solution for dealing with polymorphic taxa is to divide them so as to form two or more homogeneous taxa for the analysis (Nixon & Davis, 1991), but this becomes impractical when numerous taxa are polymorphic for several characters. Another solution is to score polymorphic characters based on an understanding of the plesiomorphic conditions in those characters for each taxon (Nixon & Davis, 1991). Such information is best derived from a more detailed analysis of relationships within a group. In fact, because of the character variability observed within many of the families included here, some of these characters may be more appropriately used in cladistic analyses of relationships within families or genera.
VII. The Unrealized Potential Contribution of Fossil Woods in Angiosperm Phylogeny
In addition to data from extant taxa, paleontological data can also make a significant contribution to understanding the evolution of wood structure and the development of phylogenetic hypotheses. Fossil woods are widespread and often abundant in the paleobotanical record, but relatively few have been studied in detail. Of the approximately 100 records of Cretaceous dicotyledonous woods, most are of Late-Cretaceous age (Wheeler & Baas, 1991), and nearly half (47) of them are from a single locality in California (Page, 1967, 1968, 1970, 1979, 1980, 1981). By the end of the Cretaceous, woods with combinations of characters present in extant Magnoliales (Magnoliaceae, Eupomatiaceae), Laurales (Chloranthaceae, Monimiaceae s.l., Lauraceae), "winteroids" (Winteraceae, Illiciaceae), and eudicots (Eupteleaceae, Platanaceae, Hamamelidaceae, Tetracentraceae) are known (Page, 1981; Wheeler & Baas, 1991), but the precise relationships of many of these fossils and their evolutionary implications remain unknown.
Most fossil woods that have been described are permineralized and usually occur in deposits that lack other fossil plant organs. As a consequence, it is usually impossible to link the fossil wood specimens to other plant parts, such as leaves or reproductive structures. This is a significant problem because angiosperm woods (both fossil and extant) have relatively few unique, systematically diagnostic characters and few characters that are amenable to cladistic analysis, as discussed above. As a result, it is often impossible to assign an unknown fossil wood to anything more precise than a group of families or genera. This problem is most severe with Cretaceous woods (Page, 1967, 1968, 1970, 1979, 1980, 1981; Herendeen, 1991a, 1991b; Wheeler & Baas, 1991; Wheeler et al., 1995).
For example, at many Cretaceous sites two types of fossil angiosperm, Icacinoxylon and Paraphyllanthoxylon, are most abundant (Wheeler & Baas, 1991). Icacinoxylon is characterized by long scalariform perforation plates, scalariform to transitional lateral pitting, wide heterocellular rays, and scanty diffuse axial parenchyma and may be representative of several families, such as Chloranthaceae, Platanaceae, and Hamamelidaceae, all of which are known from the Mid-Cretaceous of eastern North America based on reproductive structures (Crane & Herendeen, 1996; Magallon-Puebla et al., 1996, 1997). Paraphyllanthoxylon is characterized by simple perforation plates, alternate intervascular pitting, relatively narrow heterocellular rays (2-5 seriate), and septate fibers and also may be representative of several different families (Wheeler & Baas, 1991). Thus, if fossil woods are to make any significant contribution, either to understanding phylogenetic relationships within flowering plants or to clarifying the details of structural evolution in angiosperm woods, we must be able to establish links between the fossil wood and fossil reproductive structures or foliage.
Very few fossil plant localities yield reproductive structures in attachment to anatomically preserved axes (e.g., Crane et al., 1990; Manchester, 1994). The greatest potential for establishing links between reproductive structures, twigs and reproductive axes with juvenile wood features, and larger wood specimens with mature wood structure is likely to be realized in mesofossil assemblages of dispersed plant parts (flowers, fruits, seeds, twigs, wood) that are anatomically preserved, often as charcoal. Such mesofossil assemblages are common in the Cretaceous, especially in eastern North America (Crane 8,: Herendeen, 1996), and both Icacinoxylon and Paraphyllanthoxylon are well represented in numerous Mid- and Late-Cretaceous mesofossil floras from New Jersey to Georgia (Herendeen, 1991a, 1991b, unpubl. data). These fossil woods, which are preserved as charcoal and demonstrate exceptional anatomical details, occur with three-dimensionally preserved fossil flowers, fruits, seeds, twigs, and inflorescence fragments.
Icacinoxylon represents the classical Baileyan primitive wood: long scalariform perforation plates, scalariform lateral pitting, scanty diffuse parenchyma, and very wide heterocellular rays. Detailed study of specimens from the Atlantic Coastal Plain localities indicates that there are several types of Icacinoxylon wood, which differ in features such as presence or absence of uniseriate rays and septate fibers and detailed structure of the multiseriate rays (Herendeen, unpubl, data). These several types of Icacinoxylon most certainly represent families other than Icacinaceae. Fossil reproductive structures of families such as Chloranthaceae, Calycanthaceae, Hamamelidaceae, Platanaceae, and several basal rosid families have been documented from the Late Cretaceous (Crane & Herendeen, 1996), and one or more of these co-occur with Icacinoxylon at numerous mesofossil assemblages (Herendeen 1991b, unpubl. data). It is quite likely that each of these reproductive structures will be found to be associated with one type or another of Icacinoxylon wood, and discovering connections or associations between a particular type of wood and reproductive structure or leaf type is needed.
Some species of Icacinoxylon and Plataninium from the Late Cretaceous and Early Tertiary, it has been suggested, are related to the Platanaceae, based on association with platanoid reproductive structures and structural comparisons with wood of extant Platanus (e.g., Manchester, 1986; Herendeen, 1991b). These fossil woods differ from wood of extant Platanus in several respects. Perhaps most significant is that perforation plates in the fossil taxa are entirely scalariform, whereas they are mixed scalariform and simple in the extant species. Thus, these fossil woods help to clarify the original structure of perforation plates in the Platanaceae and thereby contribute useful information for phylogenetic studies.
In one case it has been possible to tie a particular fossil wood to certain reproductive structures that are found in the same fossil assemblage. Herendeen (1991a) established that the fossil wood Paraphyllanthoxylon marylandense from the Early Cenomanian Mauldin Mountain locality was produced by the extinct genus Mauldinia (Lauraceae), which was described by Drinnan et al. (1990) from fossil inflorescences, flowers, and fruits. The link between Mauldinia and Paraphyllanthoxylon was established based on anatomical similarities between charcoalified pieces of the inflorescence axis of Mauldinia and charcoalified twig wood and mature wood of Paraphyllanthoxylon. This discovery is significant because it helps to document wood structure in an early member of the Lauraceae. Wood anatomy of extant Lauraceae is variable, especially in axial parenchyma abundance and distribution (Table III). Most extant members of the family have abundant paratracheal parenchyma, whereas Paraphyllanthoxylon has very scanty paratracheal parenchyma. Thus, because Paraphyllanthoxylon is the oldest confirmed fossil wood of the Lauraceae, this fossil evidence could be used to hypothesize the plesiomorphic states for the wood anatomical characters that are variable among the extant taxa. However, because the Cretaceous record for fossil angiosperm woods is still poorly known, especially for low paleolatitude areas, it would be premature to use the currently available stratigraphic evidence to resolve questions of character polarity and variability.
Paraphyllanthoxylon has the potential to provide a better understanding of the evolutionary history of the Lauraceae and wood structure within the family if it is incorporated into phylogenetic analyses. Mauldinia mirabilis and Paraphyllanthoxylon marylandense are known in sufficient detail to include them as a single taxon in cladistic analyses of relationships among Lauraceae and related families. Together, these fossils provide character information for inflorescence structure, flower morphology, fruit structure, and mature and juvenile wood anatomy. In addition to Mauldinia-Paraphyllanthoxylon marylandense, other Lauraceae are known from the Late Cretaceous (e.g., Herendeen et al., 1994), and ideally multiple fossil taxa should be incorporated in comprehensive cladistic analyses that also include diverse extant taxa of Lauraceae, Monimiaceae s.l., and Hernandiaceae.
It is through this type of approach, establishing links between fossil woods and associated reproductive structures, and incorporating these extinct taxa into cladistic analyses, that fossil woods will make their greatest contribution to understanding the phylogenetic relationships and evolutionary history of fossil and extant plants.
List of families of dicotyledons that are polymorphic for perforation plate structure (scalariform vs. simple perforation plates)
a Polymorphic when fossil taxa with exclusively scalariform perforation plates are included (Manchester, 1986; Wheeler et al., 1995).
List of families of dicotyledons that are polymorphic for imperforate element pit structure (distinctly bordered vs. indistinctly bordered or simple pits)
Selected wood anatomical characters defined for cladistic analysis
1. Vessels: 0 = present; 1 = absent
2. Perforation plate structure: 0 = exclusively scalariform; 1 = scalariform and simple plates present in the same specimen; 2 = exclusively simple perforations
3. Intervascular pit distribution: 0 = scalariform; 1 = transitional (mixture of scalariform and opposite pits in same vessel element); 2 = opposite; 3 = alternate
4. Vessel to ray pit structure: 0 = obvious borders; 1 = borders reduced to absent (pits simple)
5. Paratracheal parenchyma distribution: 0 = absent; 1 = scanty; 2 = vasicentric; 3 = aliform to confluent
6. Apotracheal parenchyma distribution: 0 = absent; 1 = diffuse or diffuse in aggregates (groups of few cells); 2 = banded
7. Marginal parenchyma: 0 = absent; 1 = present
8. Multiseriate rays over five cells in width: 0 = absent; 1 = present. With few exceptions there seems to be a discrete gap in ray structure between those families with narrow multiseriate rays that are 2-3 cells wide, and those families with wide multiseriate rays that are greater than five cells wide.
9. Multiseriate ray structure: 0 = heterocellular; 1 = homocellular composed of upright cells; 2 = homocellular composed of procumbent cells
10. Uniseriate rays: 0 = present; 1 = absent
11. Imperforate elements: 0 = nonseptate; 1 = septate
12. Imperforate element pit structure: 0 = pit borders distinct ([greater than]3 [[micro]meter]); 1 = pits simple or borders indistinct ([less than]2 [[micro]meter]). Differences in terminology among workers present difficulties in scoring families for this character. Also, differences in sample preparation techniques, microscopy resources (light vs. SEM), and quality of published photographs are also potential sources of difficulty.
We thank Steven Manchester and Peter R. Crane for their comments on an earlier version of this paper.
IX. Literature Cited
Anderberg, A. A. 1992. The circumscription of the Ericales, and their cladistic relationships to other families of "higher" dicotyledons. Syst. Bot. 17: 600-675.
Baas, P. 1986. Ecological patterns of xylem anatomy. Pp. 327-352 in J. Givnish (ed.), On the economy of plant form and function. Cambridge University Press, New York.
-----. 1993. Reversibility and parallelism in wood evolution. XV International Botanical Congress, Yokohama, Japan. Abstracts p. 86.
----- & E. A. Wheeler. 1996. Parallelism and reversibility on xylem evolution - A review. IAWA J. 17:351-364.
-----, P.M. Esser, M. E. T. van der Westen & M. Zandee. 1988. Wood anatomy of the Oleaceae. IAWA Bull., n.s. 9: 103-182.
Bailey, I. W. 1944. The development of vessels in angiosperms and its significance in morphological research. Amer. J. Bot. 31: 421-428.
-----. 1957. The potentialities and limitations of wood anatomy in the study of the phylogeny and classification of angiosperms. J. Arnold Arbor. 38: 243-254.
----- & W. W. Tupper. 1918. Size variation in tracheary cells. I. A comparison between the secondary xylems of vascular cryptogams, gymnosperms and angiosperms. Proc. Amer. Acad. Arts 54: 149-204.
Barghoorn, E. S. 1940. The ontogenetic development and phylogenetic specialization of rays in the xylem of dicotyledons. I. The primitive ray structure. Amer. J. Bot. 27: 918-928.
-----. 1941a. The ontogenetic development and phylogenetic specialization of rays in the xylem of dicotyledons. II. Modification of the multiseriate and uniseriate rays. Amer. J. Bot. 28: 273-282.
-----. 1941b. The ontogenetic development and phylogenetic specialization of rays in the xylem of dicotyledons. III. The elimination of rays. Amer. J. Bot. 68: 317-325.
Canright, J. E. 1955. The comparative morphology and relationships of the Magnoliaceae. IV. Wood and nodal anatomy. J. Arnold Arbor. 36:119-140.
Carlquist, S. 1981. Wood anatomy of Zygogynum (Winteraceae); field observations. Bull. Mus. Natl. Hist. Nat., B, Adansonia, Paris 4e ser., 3: 281-292.
-----. 1982a. Exospermum stipitatum (Winteraceae): Observations on wood, leaves, flowers, pollen, and fruit. Aliso 10: 277-289.
-----. 1982b. Wood anatomy of Illicium (Illiciaceae): Phylogenetic, ecological, and functional interpretations. Amer. J. Bot. 69: 1587-1598.
-----. 1983a. Wood anatomy of Belliloum (Winteraceae) and a note on flowering. J. Arnold Arbor. 64: 161-169.
-----. 1983b. Wood anatomy of Bubbia (Winteraceae), with comments on the origin of vessels in dicotyledons. Amer. J. Bot. 70: 578-590.
-----. 1983c. Wood anatomy of Calycanthaceae: Ecological and systematic implications. Aliso 10: 427-441.
-----. 1984. Wood anatomy of Trimeniaceae. PI. Syst. Evol. 144: 103-118.
-----. 1987. Presence of vessels in wood of Sarcandra (Chloranthaceae)', comments on vessel origins in angiosperms. Amer. J. Bot. 74: 1765-1771.
-----. 1988a. Wood anatomy of Drimys s.s. (Winteraceae). Aliso 12: 81-95.
-----. 1988b. Comparative wood anatomy. Springer-Verlag, Berlin.
-----. 1989a. Wood anatomy of Tasmannia; summary of wood anatomy of Winteraceae. Aliso 12: 257-275.
-----. 1989b. Wood and bark anatomy of Degeneria. Aliso 12: 485-495.
-----. 1990a. Wood anatomy and relationships of Lactoridaceae. Amer. J. Bot. 77: 1489-1505.
-----. 1990b. Wood anatomy of Ascarina (Chloranthaceae) and the tracheid-vessel element transition. Aliso 12: 667-684.
-----. 1991. Wood and bark anatomy of Ticodendron: Comments on relationships. Ann. Missouri Bot. Gard. 78: 96-104.
-----. 1992a. Wood anatomy and stem of Chloranthus; summary of wood anatomy of Chloranthaceae, with comments on relationships, vessellessness, and the origin of monocotyledons. IAWA Bull., n.s. 13: 3-36.
-----. 1992b. Vegetative anatomy and relationships of Eupomatiaceae. Bull. Torrey Bot. Club 119: 167-190.
-----. 1993. Wood and bark anatomy of Aristolochiaceae: Systematic and habital correlations. IAWA Bull., n.s. 14: 341-357.
-----. 1996. Wood anatomy of primitive angiosperms: New perspectives and syntheses. Pp. 68-90 in D. W. Taylor and L. J. Hickey (eds.), Flowering plant origin, evolution, and phylogeny. Chapman & Hall, New York.
Chase, M. W., D. E. Soltis, R. G. Olmstead, D. Morgan, D. H. Les, B. D. Mishler, M. R. Duvall, R. A. Price, H. G. Hills, Y.-L. Qiu, K. A. Kron, J. H. Rettig, E. Conti, J. D. Palmer, J. R. Manhart, K. J. Sytsma, H. J. Michaels, W. J. Kress, K. G. Karol, W. D. Clark, M. Hedren, B. S. Gaut, R. K. Jansen, K.-J. Kim, C. F. Wimpee, J. F. Smith, G. R. Furnier, S. H. Strauss, Q.-Y. Xiang, G. M. Plunkett, P. S. Soltis, S. M. Swensen, S. E. Williams, P. A. Gadek, C. J. Quinn, L. E. Eguiarte, E. Golenberg, G. H. Learn, S. W. Graham, S.C. H. Barrett, S. Dayanandan & V. A. Albert. 1993. Phylogenetics of seed plants: An analysis of nucleotide sequences from the plastid gene rbcL. Ann. Missouri Bot. Gard. 80: 528-580.
Chen, B. L., P. Baas, E. A. Wheeler & S. M. Wu. 1993. Wood anatomy of trees and shrubs from China. VI. Magnoliaceae. IAWA J. 14: 391-412.
Crane, P. R. & P.S. Herendeen. 1996. Cretaceous floras containing angiosperm flowers and fruits from eastern North America. Rev. Palaeobot. & Palynol. 90: 319-337.
-----, S. R. Manchester & D. L. Dilcher. 1990. A preliminary survey of fossil leaves and well-preserved reproductive structures from the Sentinel Butte Formation (Paleocene) near Almont, North Dakota. Fieldiana, Geol. 20: 1-63.
Cronquist, A. 1981. An integrated system of classification of flowering plants. Columbia University Press, New York.
Dahlgren, R. T. & K. Bremer. 1985. Major clades of angiosperms. Cladistics 1: 349-368.
Donoghue, M. & J. A. Doyle. 1989. Phylogenetic relationships of angiosperms and the relationships of the Hamamelidae. Pp. 17-45 in P. R. Crane and S. Blackmore (eds.), Evolution, systematics, and fossil history of the Hamamelidae. Clarendon Press, Oxford.
Doyle, J. A. & M. J. Donoghue. 1993. Phylogenies and angiosperm diversification. Paleobiology 19: 141-167.
Drinnan, A. N., P. R. Crane, E. M. Friis & K. R. Pedersen. 1990. Lauraceous flowers from the Potomac Group (mid-Cretaceous) of eastern North America. Bot. Gaz. 151: 370-384.
Frost, F. H. 1930a. Specialization in secondary xylem of dicotyledons. I. Origin of vessel. Bot. Gaz. 89: 67-94.
-----. 1930b. Specialization in secondary xylem of dicotyledons. II. Evolution of end wall of vessel. Bot. Gaz. 90: 198-212.
-----. 1931. Specialization in secondary xylem of dicotyledons. III. Specialization of lateral wall of vessel segment. Bot. Gaz. 91: 88-96.
Gomez-Laurito, J. & L. D. Gomez P. 1989. Ticodendron: A new tree from Central America. Ann. Missouri Bot. Gard. 76: 1148-1151.
Herendeen, P.S. 1991a. Lauraceous wood from the mid-Cretaceous Potomac Group of eastern North America: Paraphyllanthoxylon marylandense sp. nov. Rev. Palaeobot. & Palynol. 69: 277-290.
-----. 1991b. Charcoalified angiosperm wood from the Cretaceous of eastern North America and Europe. Rev. Palaeobot. & Palynol. 70: 225-239.
-----, W. L. Crepet & K. C. Nixon. 1994. Fossil flowers and pollen of Lauraceae from the Upper Cretaceous of New Jersey. Pl. Syst. Evol. 189: 29-40.
Hufford, L. 1992. Rosidae and their relationships to other magnoliid dicotyledons: A phylogenetic analysis using morphological and chemical data. Ann. Missouri Bot. Gard. 79: 218-248.
----- & W. C. Dickison. 1992. A phylogenetic analysis of Cunoniaceae. Syst. Bot. 17: 181-200. IAWA Committee. 1989. IAWA list of microscopic features for hardwood identification. IAWA Bull., n.s. 10: 219-332.
Keller, J. A., P.S. Herendeen & P. R. Crane. 1996. Fossil flowers and fruits of the Actinidiaceae from the Campanian (Late Cretaceous) of Georgia. Amer. J. Bot. 83: 528-541.
Klaassen, R. 1999. Wood anatomy of the Sapindaceae. IAWA J. Supplement 2.
Kribs, D. A. 1935. Salient lines of structural specialization in the wood rays of dicotyledons. Bot. Gaz. 94: 547-557.
-----. 1937. Salient lines of structural specialization in the wood parenchyma of dicotyledons. Bull. Torrey Bot. Club 64: 177-186.
Lammers, T. G., T. F. Stuessy & M. Silva O. 1986. Systematic relationships of the Lactoridaceae, an endemic family of the Juan Fernandez Islands, Chile. Pl. Syst. Evol. 152: 243-266.
Loconte, H. & D. W. Stevenson. 1991. Cladistics of the Magnoliidae. Cladistics 7: 267-296.
Magallon-Puebla, S., P.S. Herendeen & P. K. Endress. 1996. Floral remains of the tribe Hamamelideae (Hamamelidaceae) from Campanian strata of southeastern U.S.A. PI. Syst. Evol. 202: 177-198.
-----, ----- & P. R. Crane. 1997. Quadriplatanus georgianus gen. et sp. nov.: Staminate and pistillate platanaceous flowers from the Late Cretaceous (Coniacian-Santonian) of Georgia, U.S.A. Int. J. Pl. Sci. 158: 373-394.
Manchester, S. R. 1986. Vegetative and reproductive morphology of an extinct plane tree (Platanaceae) from the Eocene of western North America. Bot. Gaz. 147: 200-226.
-----. 1994. Fruits and seeds of the Middle Eocene Nut Beds Flora, Clarno Formation, Oregon. Palaeontogr. Amer. 58: 1-205.
Meeuse, A.D. J. 1982. Cladistics, wood anatomy and angiosperm phylogeny. Acta Bot. Neerl. 31: 345-354.
Metcalfe, C. R. 1987. Anatomy of dicotyledons. Ed. 2. Vol. III. Magnoliales, Illiciales, and Laurales. Clarendon Press, Oxford.
----- & L. Chalk. 1950. Anatomy of the dicotyledons. Vols. I and II. Clarendon Press, Oxford.
Nixon, K. C. & J. I. Davis. 1991. Polymorphic taxa, missing values and cladistic analysis. Cladistics 7: 233-241.
Noshiro, S. & P. Baas. 1998. Systematic wood anatomy of Cornaceae and allies. IAWA J. 19: 43-97.
Page, V. M. 1967. Angiosperm wood from the Upper Cretaceous of central California. Part I. Amer. J. Bot. 54: 510-514.
-----. 1968. Angiosperm wood from the Upper Cretaceous of central California. Part Il. Amer. J. Bot. 55: 168-172.
-----. 1970. Angiosperm wood from the Upper Cretaceous of central California. III. Amer. J. Bot. 57: 1139-1144.
-----. 1979. Dicotyledonous wood from the Upper Cretaceous of central California. J. Arnold Arbor. 60: 323-349.
-----. 1980. Dicotyledonous wood from the Upper Cretaceous of central California. II. J. Arnold Arbor. 61: 723-748.
-----. 1981. Dicotyledonous wood from the Upper Cretaceous of central California. III. Conclusions. J. Arnold Arbor. 62: 437-455.
Panshin, A. J. & C. DeZeeuw. 1980. Textbook of wood technology. Ed. 4. McGraw-Hill, New York.
Qiu, Y.-L., M. W. Chase, D. H. Les & C. R. Park. 1993. Molecular phylogenetics of the Magnoliidae: Cladistic analyses of nucleotide sequences of the plastid gene rbcL. Ann. Missouri Bot. Gard. 80: 587-606.
Riggins, R. & J. S. Farris. 1983. Cladistics and the roots of angiosperms. Syst. Bot. 8: 96-101.
Shutts, C. F. 1960. Wood anatomy of Hernandiaceae and Gyrocarpaceae. Trop. Woods 113: 85-123.
Stern, W. L. & S. Greene. 1958. Some aspects of variation in wood. Trop. Woods 108: 65-71.
Takahashi, A. 1985. Wood anatomical studies of the Polycarpicae. I. Magnoliales. Sci. Rep. Coll. Gen. Educ. Osaka Univ. 34: 29-83.
Terrazas, T. S. 1994. Wood anatomy of the Anacardiaceae: Ecological and phylogenetic interpretation. Unpub. dissertation, University of North Carolina-Chapel Hill.
Wheeler, E. A. & P. Baas. 1991. A survey of the fossil record for dicotyledonous wood and its significance for evolutionary and ecological wood anatomy. IAWA Bull., n.s. 12: 271-332.
----- & -----.1993. The potentials and limitations of dicotyledonous wood anatomy for climatic reconstructions. Paleobiology 19: 487-498.
-----, J. McClammer & C. A. LaPasha. 1995. Similarities and differences in dicotyledonous woods of the Cretaceous and Paleocene San Juan Basin, New Mexico. IAWA J. 16: 223-254.
Wilson, T. K. 1960. The comparative morphology of the Canellaceae. I. Synopsis of genera and wood anatomy. Trop. Woods 112: 1-27.
Young, D. A. 1981. Are the angiosperms primitively vesselless? Syst. Bot. 6: 313-330.
Zhang, S. Y. 1992. Systematic wood anatomy of the Rosaceae. Blumea 37: 81-158.
----- & P. Baas. 1992. Wood anatomy of trees and shrubs from China. III. Rosaceae. IAWA Bull., n.s. 13: 21-91.
|Printer friendly Cite/link Email Feedback|
|Author:||Herendeen, Patrick S.; Wheeler, Elisabeth A.; Baas, Pieter|
|Publication:||The Botanical Review|
|Date:||Jul 1, 1999|
|Previous Article:||Implications of Fossil Conifers for the Phylogenetic Relationships of Living Families.|
|Next Article:||The Inflorescence: Introduction.|