| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Anatomy and Morphology |
Department of Plant Sciences, Mail Stop 2, University of California–Davis, One Shields Ave., Davis, California 95616 USA
Received for publication December 21, 2004. Accepted for publication June 14, 2005.
ABSTRACT
Stomatal architecture—the number, form, and arrangement of specialized epidermal cells associated with stomatal guard cells—of 46 species of basal angiosperms representing all ANITA grade families and Chloranthaceae was investigated. Leaf clearings and cuticular preparations were examined with light microscopy, and a sample of 100 stomata from each specimen was coded for stomatal type and five other characters contributing to stomatal architecture. New stomatal types were defined, and many species were examined and illustrated for the first time. Character evolution was examined in light of the ANITA hypothesis using MacClade software. Analysis of character evolution, along with other evidence from this study and evidence from the literature on fossil angiosperms and other seed plant lineages, suggests that the ancestral condition of angiosperms can be described as anomo-stephanocytic, a system in which complexes lacking subdidiaries (anomocytic) intergrade with those having weakly differentiated subsidiaries arranged in a rosette (stephanocytic). From this ancestral condition, tangential divisions of contact cells led to the profusion of different types seen in early fossil angiosperms and Amborellaceae, Austrobaileyales, and derived Chloranthaceae, while the state in Nymphaeales is little modified. Formation of new, derived types by tangential division appears to be a recurrent theme in seed plant evolution.
Key Words: Amborellaceae • Austrobaileyales • basal angiosperms • epidermal anatomy • evolution • fossil angiosperms • Nymphaeales • stomata
Over the last six years, many researchers interested in angiosperm origins and early evolution have directed their efforts toward the study of three clades of flowering plants convincingly shown by several phylogenetic analyses of multiple genes to have arisen from the deepest lineage-splitting events within the extant angiosperm clade (e.g., Qiu et al., 1999
, 2000
; Soltis et al., 1999
; Mathews and Donoghue, 1999
, 2000
; Graham and Olmstead, 2000
). These analyses placed Amborella trichopoda Baill., a shrub with unisexual flowers and vesselless xylem endemic to New Caledonia, as the sister to the rest of the extant angiosperms. Nymphaeales (water lilies), comprising two families and about 70 species of herbaceous, vesselless aquatics, were resolved as the next lineage above the Amborellaceae, followed by the Austrobaileyales, an order of four small families, and roughly 80 species of woody vines (Austrobaileyaceae, Schisandraceae), and shrubs and trees (Illiciaceae, Trimeniaceae). Combined analysis of morphological and molecular data (Doyle and Endress, 2000
) corroborated these findings. These three lineages have come to be known as the "ANITA" grade, an acronym formed from Amborella, Nymphaeales, Illiciales, Trimeniaceae, and Austrobaileya. Many different aspects of the biology of ANITA angiosperms have been investigated, including their floral morphology and reproductive biology (e.g., Endress and Igersheim, 2000a
, b
; Endress, 2001
; Friedman, 2001
; Williams and Friedman, 2002
; De Craene et al., 2003
), vegetative anatomy (Carlquist, 2001
; Carlquist and Schneider, 2001
, 2002
), and ecology (e.g., Bernhardt et al., 2003
; Feild et al., 2004
).
One aspect of ANITA angiosperms that has attracted little attention is leaf epidermal anatomy, particularly stomatal architecture—the number, form, and arrangement of specialized epidermal cells associated with the stomatal guard cells. Stomatographic studies (i.e., those on stomatal architecture and sometimes other features) have shown that stomata can provide valuable taxonomic and systematic evidence in both living and fossil plants (e.g., Florin, 1931
, 1951
; Bailey and Nast, 1948
; Stebbins and Khush, 1961
; Stace, 1965
; Van Cotthem, 1970
; den Hartog and Baas, 1978
; Wilkinson, 1979
; Upchurch, 1984a
, b
, 1995
; Baranova, 1972
, 1992a
, b
; Kong, 2001
) and also have played a significant role in framing hypotheses about early angiosperm evolution (Upchurch, 1984a
, b
). Although stomatographic studies on these families were done prior to their placement at the base of the angiosperm tree, most were done before important advances in describing and understanding structure and development of different stomatal types, and many ANITA taxa still remain to be examined. Carlquist (2001)
and Baranova (2004)
presented brief treatments of stomatal architecture in Austrobaileya scandens C.T. White, and Amborella trichopoda was briefly summarized by Carlquist and Schneider (2001)
. However, I believe that a large-scale, comparative stomatographic survey using modern concepts and terminology, across all ANITA families, is needed to elucidate ancestral stomatal architecture and evolution in light of the this new consensus on angiosperm phylogeny.
In this study, I have examined stomatal type and five other characters contributing to stomatal architecture of mature complexes in 42 species of ANITA angiosperms representing all seven families and 11 of 14 genera, as well as four species of Chloranthaceae, which were inferred by Doyle and Endress (2000)
to branch above the ANITA grade. I examined the evolution of these characters in light of the ANITA hypothesis and used the results, along with evidence from the literature, to address several important issues. These include (1) the character states (including stomatal type) present in the common ancestor of all angiosperms and at other important internal nodes near the base of the angiosperm tree, (2) the evolution and polarity of stomatal characters within basal clades, (3) what the architecture of mature stomatal complexes may suggest about their development, (4) the degree of similarity in stomatal architecture between ANITA taxa and early angiosperm fossils and what this may suggest about early evolution of angiosperms, and (5) the degree of similarity between ANITA angiosperms and other living and fossil seed plant lineages and what this may imply. In addition to these issues, data and hypotheses generated by this study will shed light on other issues of interest for future study, such as how early angiosperm fossils may be related to extant taxa and the character states likely to be present in fossil taxa on the stem lineage to angiosperms. Many species included here are examined and illustrated for the first time.
Terminology
Because terminology pertaining to stomata is very specialized and has not been used consistently, it is necessary to define the terms used here. Stoma (stomata, pl.) refers to the stomatal pore and the pair of guard cells that form it. Stomatal complex refers to the stoma plus any specialized epidermal cells adjacent to it. Such specialized epidermal cells are called subsidiary cells (or simply, subsidiaries) and differ from unmodified epidermal cells in shape, size, staining properties, orientation, or some other feature. These definitions have been followed by the majority of authors (e.g., Metcalfe and Chalk, 1950
; Esau, 1953
; Pant, 1965
; Stace, 1965
; Van Cotthem, 1970
; Wilkinson, 1979
; Baranova, 1987b
, 1992b
). Nonspecialized epidermal cells adjacent to the stoma have been called neighboring cells by many of these same authors. However, I have adopted the term contact cell (used by Upchurch, 1984a
) to refer to any cell, specialized or not, that is adjacent to (i.e., in contact with) the stoma. Thus, the term nonspecialized contact cell takes the place of neighboring cell. Specialized noncontact cell refers to any specialized cell that is not immediately adjacent to the stoma, but still comprises a portion of the stomatal complex. Stomatal pole refers to the two opposite areas where the two guard cells meet, while the term lateral encompasses the sides of the stoma (i.e., everything except the poles; see Fig. 1). A radial wall is a cell wall oriented more or less perpendicular to the stoma (i.e., radiating away from it), while a tangential wall is one that is oriented more or less parallel to it (Fig. 1).
|
Specimen selection
Selection of taxa was guided by recent phylogenetic analyses of basal angiosperm families and other relevant literature as noted later. For certain critical taxa such as Amborella trichopoda and Austrobaileya scandens, samples were taken from both different individuals and different leaves. Leaf material was obtained from the wild, botanic gardens, and herbaria. Prepared slides of three species were observed and photographed in the Jodrell laboratory at the Royal Botanic Gardens, Kew. The Appendix lists taxa examined with source and voucher information.
Specimen preparation
Leaf clearings
For each specimen, a mature leaf was selected, and samples were excised near the mid-section of the blade. These were immersed in 5% potassium hydroxide overnight (12–24 h), rinsed in deionized water, and immersed in a fresh solution of potassium hydroxide overnight (12–24 h). Pieces were rinsed in deionized water and treated briefly (5 min) in glacial acetic acid, followed by Clorox bleach (sodium hypochlorite, bottle strength) until clear. The time required ranged from 5 s to over 1 h. After clearing, pieces were dehydrated in an ethanol series, stained for a minimum of 3 d in 1% safranin O in 100% ethanol, destained in 100% ethanol, and mounted onto microscope slides in Bioquip's Euparal (Rancho Dominguez, California, USA). Where possible, the mesophyll tissue was removed under a dissecting microscope, leaving only the abaxial epidermis (adaxial in Nymphaeales) to be mounted. Certain species of Nymphaeales were stained in 0.5% toluidine blue (aqueous), instead of safranin O, for 10 min to improve contrast.
Cuticular preparations
Cuticular preparations were made for a minority of the specimens. Leaves were sampled as before, and pieces were macerated according to G. R. Upchurch's (Texas State University, San Marcos, unpublished data) modification of Stace's (1965)
method. The two-step oxidative maceration involves a treatment in 70% nitric acid, until leaf segments turn golden-brown or become translucent at the edges (usually complete in 1–30 min for most mesophytic leaves). Segments are then rinsed several times in deionized water, then immersed in 30% chromium trioxide overnight (12–24 h). Leaf cuticles are then rinsed several times in deionized water, dehydrated in 100% methanol, and stained for a minimum of 3 d in 1% safranin O in 100% methanol. Cuticles are destained in methanol, and mounted onto coverslips in Union Carbide AYAF (Somerset, New Jersey, USA), a discontinued, alcohol-soluble, synthetic resin. After curing (1 h on a 45°C hotplate), coverslips are mounted to slides in Buehler Canada balsam (Lake Bluff, Illinois, USA).
Data collection and analysis
Slides were examined and photographed with an Olympus (Tokyo, Japan) BH-2 light microscope and Microfire (Tokyo, Japan) digital camera. One hundred stomatal complexes of each specimen were numbered using PowerPoint (Microsoft; Redmond, Washington, USA), examined on screen, and scored on a spreadsheet. Most of the stomatal characters examined are based on those from Upchurch's (1984a)
system. Characters examined here include: (1) Stomatal type: Dilcher's (1974)
system was followed for paracytic and other types, while types devised by other authors such as den Hartog and Baas' (1978)
laterocytic type, and Baranova's (1987a
, b)
stephanocytic type were also recognized. Additionally, I defined several new stomatal types. Definitions of stomatal types and their distributions in basal angiosperm taxa are presented in Fig. 3. (2) Total number of contact cells: the number of cells in contact with the stoma. Subsidiary cells and nonspecialized contact cells are both included in this number. (3) Presence or absence of lateral subsidiary cell(s). (4) Presence or absence of polar subsidiary cell(s). (5) Presence or absence of specialized noncontact cell(s). (6) Orientation of contact cell walls: radial orientation, with contact cell walls arranged in a spoke-like pattern (Fig. 2); bilateral or biradial orientation, with polar cells perpendicular to the stomatal pore (Fig. 2). The various stomatal types and character states often intergrade without sharp boundaries in these taxa; many are particularly difficult to code for certain characters as a result. In an effort to ensure consistency across the entire sample, each specimen was coded at least three different times.
|
). Characters were mapped onto the basal portion of the single most parsimonious tree found by Doyle and Endress (2000)
in their combined analysis of molecular and morphological data. Within this, I expanded the topologies of Nymphaeales and Chloranthaceae based on the analyses of Les et al. (1999)
and Qiu et al. (1999
, 2000)
, respectively. For assigning character states to chloranthaceous genera, my observations were supplemented by data from Baranova (1987b)
and Kong (2001)
. Because I did not examine Sarcandra, I have coded characters pertaining to this genus as uncertain when they were not explicitly discussed by these authors or could not be inferred with certainty by their coding of stomatal type. Character evolution was examined using the most parsimonious state at each node option in MacClade (i.e., the option that makes the fewest assumptions), and character states were assumed to be unordered. For stomatal type, the character states mapped on the tree represent the most frequently observed type in each genus. A variety of different stomatal types belonging to such broader categories as laterocytic, paracytic, and stephanocytic were observed (Fig. 3); however, the different types are grouped into these larger categories for study of character evolution. Because the other characters reconstructed are all binary, the assigned character states represent the majority character state within each genus examined. Thus, it is important to note that the character states plotted onto the tree represent relative frequencies, and when an internal node is reconstructed as present, this means that the character state is present in a majority of stomatal complexes in that ancestor. The use of frequencies as characters may be viewed as questionable by some, but mature stomatal complexes in basal angiosperm taxa often vary so greatly that few character states can be said to be wholly absent. Use of most common type as a character state has been accepted as a suitable method by other researchers reconstructing stomatal characters (e.g., Doyle and Endress, 2000
; Eklund et al., 2004
). RESULTS
The results reveal two divergent systems of stomatal architecture in basal angiosperms. System 1, present in Amborellaceae, Austrobaileyales, and the chloranthaceous genera Ascarina and Chloranthus, has a majority of stomatal complexes with strongly specialized subsidiary cells (distinguished from other epidermal cells by one or more features such as smaller size, distinctly curved or straight wall contour, thinner cell walls, different staining properties, or cuticular sculpture, etc.). Stomatal types in these taxa are predominantly laterocytic or paracytic, and, in many taxa, most complexes have specialized noncontact cells. Some have two concentric rings of specialized cells. Most taxa with system 1 architecture have a low frequency of polar subsidiaries. System 2, found in Nymphaeales, has a small minority of stomatal complexes with subsidiary cells specialized as in system 1. The majority of stomatal complexes comprises a continuum of forms that either lack subsidiaries (i.e., the anomocytic type) or have weakly specialized subsidiaries (differing little from other epidermal cells) arranged in a circle or ellipse. Some form well-defined rings, as in the stephanocytic or actinocytic types, while some rings are weaker, or interrupted at one or more points by nonspecialized cells. Due to the ringlike arrangement of subsidiaries, most complexes in each specimen have polar as well as lateral subsidiaries. Most complexes lack specialized noncontact cells, and complexes with concentric rings are rare. The genus Hedyosmum (Chloranthaceae) has properties of both systems, as noted in results by taxon. Table 1 shows the frequencies of character states by taxon.
|
Results by taxon
General summaries of basal angiosperm taxa follow, but a complete listing of character states for each taxon is presented in Table 1.
Amborellaceae
Two leaves from each of two individuals were examined. Stomata of Amborella trichopoda are shown in Figs. 4–13. The majority (62–77%) of stomatal complexes in each specimen comprises paracytic types, including variants outlined by Dilcher (1974)
, such as brachyparacytic (the majority; Figs. 4–6, 8, 9, 11, 12), weakly brachyparacytic (my terminology; Figs. 5, 7), amphibrachyparacytic (Fig. 4), hemiparacytic, and in one specimen, a single holoparacytic complex. The remaining complexes include laterocytic types, mostly 1 + 2, rarer 2 + 2 types, and latero-cyclocytics. Stephanocytic (Fig. 11), and rare anomocytic types are also observed. Irregular complexes do not conform closely to any of these stomatal types. Many of these are characterized by polar subsidiaries (Figs. 7, 9), while some have two lateral subsidiaries, one extending the full length of the stoma, and one on the other side, that extends about half the length of the stoma (Figs. 4, 7). The latter resembles a laterocytic 1 + 2 type that has lost (or never formed) specialization in one cell. I have named this type irregular-incomplete laterocytic.
|
|
|
Schisandraceae
Fourteen species were examined, including eight species of Schisandra representing all three subgenera (Saunders, 2000
) and six species of Kadsura representing both subgenera (Saunders, 1998
). Stomata of Schisandraceae are shown in Figs. 42–56. Laterocytic types are most common (13–64%) in the majority of specimens examined (Figs. 42– 49, 51–53, 55, 56) and a wide variety of forms is observed, including laterocytic 1 + 2 (the most common type in most), 1 + 3, 2 + 2, 2 + 3, 3 + 3, latero-cyclocytic, and weakly laterocytic 1 + 2, and 2 + 2. Paracytic types (Figs. 42, 44, 46, 55) are also common (and represent the most common type in S. henryi, and K. borneensis), and include brachyparacytics, amphibrachyparacytics, and weakly brachyparacytics. Stephanocytic types also occur in all specimens examined and consist of types with one cycle of cells (stephanocytic types; Figs. 43, 47), and more commonly, stephanocytic bicyclic types (Figs. 42, 43, 45, 46, 48–52, 54, 56). Irregular types with polar subsidiaries (Figs. 44 and 45) are less common, as are irregular-incomplete laterocytics (Figs. 47, 51). Anomocytic complexes are absent.
|
, and the major clades found in the phylogenetic analysis of Hao et al. (2000)
. Stomata of Illiciaceae are shown in Figs. 57–63. Paracytic types predominate in this family (51–95% across the sample). For each specimen, either brachyparacytic (Figs. 58, 61, 62) or amphibrachyparacytic types (Figs. 57, 59, 60, 63) are most common; a small minority of weakly brachyparacytics is also observed. Stephanocytic types including unicyclic and bicyclic types (Figs. 57, 58, 60, 62, 63) are less common than in Schisandraceae. A small number of complexes appear paracytic but are surrounded by a well-defined rosette of cells without close alignment of the two cycles of cells; these are coded stephanocytic bicyclic here (e.g., Fig. 58). Laterocytics (Figs. 59, 61) including laterocytic 1 + 2, 2 + 2, weakly laterocytic 1 + 2, and latero-cyclocytics, and irregulars with polar subsidiaries comprise the remaining complexes.
|
Character evolution
MacClade reconstructed the evolution of characters 1, 3, 4, and 6 as follows: (1) For stomatal type, the ancestral state of angiosperms—present at the node that gave rise to Amborella + Other Angiosperms (henceforth abbreviated as node 1)—is reconstructed as equivocal between a state in which stephanocytic stomatal complexes are most common, and a state in which the most common type is paracytic (Fig. 73). Likewise, equivocal reconstructions, all involving a possible stephanocytic state, are found for other important nodes including the ancestor of Nymphaeales + Other Angiosperms (node 2), the ancestor of Austrobaileyales + Other Angiosperms (node 3), and the ancestor of Chloranthaceae + Other Angiosperms (node 4). (3) All taxa are reported as having a majority of complexes with lateral subsidiaries, hence all internal nodes in the tree, including the common ancestor of all angiosperms, are reconstructed unequivocally as having this condition as well. (4) Nodes 2, 3, and 4 are reconstructed as having a majority of complexes with polar subsidiaries, while node 1 is reconstructed as equivocal. (Fig. 71). (6) Radial orientation of contact cells in the majority of stomatal complexes is the reconstructed state for nodes 2, 3, and 4, while node 1 is reconstructed as equivocal (Fig. 72). Character 5, presence or absence of specialized noncontact cells, exhibits considerable variability within genera, especially Amborella, and Illicium, suggesting that reconstructions of this character with these data may not be meaningful; however, the presence of specialized noncontact cells is of great importance in the study, as discussed later. Character 2, number of contact cells per complex, exhibits enough overlap among the taxa to cast doubt on the utility of MacClade reconstructions.
|
Previous stomatographic studies of ANITA angiosperms and Chloranthaceae
Prior to the placement of the ANITA grade at the base of the angiosperm phylogenetic tree, many aspects of its members' anatomy, including stomata, attracted the attention of botanists. This was due to their membership in the putatively primitive groups in earlier classification schemes of angiosperms (e.g., Bessey, 1915
; Cronquist, 1968
, 1981
) and also to the interest sparked by unusual features of certain taxa, such as the vesselless xylem of Amborella (Bailey and Swamy, 1948
).
In Amborella trichopoda, Bailey and Swamy (1948)
reported stomata conforming to the definitions of paracytic and anomocytic (terminology not devised until Metcalfe and Chalk, 1950
). Upchurch (1984a
, b)
added laterocytic, cyclocytic, and intermediates, etc., and Carlquist and Schneider (2001
, p. 306) reported paracytics, "variations on a paracytic type," and others in lower frequencies. The present study confirms a predominance of paracytic complexes in Amborella, but the rarity of anomocytic complexes reported here is a point of disagreement with Bailey and Swamy (1948)
. A likely explanation for this discrepancy is that two types of stomatal complexes encountered in Amborella, laterocytic and stephanocytic, were not described or recognized until later by den Hartog and Baas (1978)
and Baranova (1987a
, b)
, respectively. Baranova (1983)
pointed out that prior to recognition of the laterocytic type, earlier workers often classified laterocytic types as anomocytics or "special types." Also, Baranova (1987b)
noted that stephanocytic and related types (e.g., actinocytic) can often be difficult to discern from anomocytic; she believed they were in fact, "simply a modification of the anomocytic type." New to this study is the identification of stephanocytic and related types in Amborella, including some bicyclics.
Stomata of Nymphaeales were described as anomocytic by Williamson et al. (1989)
, Schneider and Williamson (1993)
, and Williamson and Schneider (1993)
, but cyclocytic stomata were reported in Barclaya rotundifolia Hotta by Schneider et al. (1992)
. In the present study, although anomocytic stomata are found to be common in each specimen (Nymphaeaceae), they are less common than the total of all types I have included as stephanocytic variants (Fig. 3) in every species except Nuphar luteum Sibth. & Sm. While my totals of stephanocytic types include the weakly specialized variants (e.g., weakly stephanocytic), these comprise no more than 50% of the total of all stephanocytic types across the sample (23–50%). Hypothetically, even if one were to consign all weakly specialized variants to the anomocytic category, the majority of specimens in Nymphaeaceae (8 of 13) would still have stephanocytic variants as their most common type. Hence, I maintain that the majority of stomatal complexes in this family belong to the stephanocytic category. Some contact cells are also radially elongated, thus forming actinocytic types (Figs. 17–19, 21, 23), while some complexes are intermediate between actinocytic and stephanocytic (Figs. 15, 16, 18–23). Cabombaceae, in contrast to Nymphaeaceae, have few anomocytic types, and a great majority of stephanocytic variants. Despite the increase in the number of stomatal types recognized here, it is important to note that anomocytic types and the different stephanocytic types all grade into one another without sharp boundaries, and criteria for distinguishing them are likely to vary somewhat among researchers.
In Austrobaileya scandens, Bailey and Swamy (1949
, p. 215) described stomata as either "rubiaceous" (i.e., paracytic), "surrounded by ordinary epidermal cells" (i.e., anomocytic), or intermediate between these. Carlquist (2001
, p. 9) offered a similar assessment, but added, "These stomata do not correspond exactly to the concept of paracytic subsidiary cells." Both of the stomata he illustrated actually appear laterocytic. Upchurch (1984a)
added laterocytics and others and also noted that complexes in this taxon invariably lack polar subsidiaries, an observation corroborated by Baranova (1992a)
. Baranova (2004)
criticized Carlquist's failure to include laterocytic types and reiterated her assessment that stomatal types in Austrobaileya consist of laterocytic, anomocytic, and paracytic.
My results for Austrobaileya scandens differ considerably from these studies. First, I do not recognize a single anomocytic complex in any of the Austrobaileya leaves or individuals I examined, and second, I note many stephanocytic bicyclic complexes. Third, I note polar subsidiaries in the majority of complexes in each sample (i.e., as strongly specialized cells or weakly specialized cells of the outer ring of stephanocytic bicyclics; see Figs. 24, 25, 27–29, 31, 32, 34, 37, 38). I offer two explanations for these discrepancies. First, as with Bailey and Swamy's (1948)
account of Amborella trichopoda, their (1949)
account of Austrobaileya preceded the recognition of laterocytic and stephanocytic types, which were often consigned to the anomocytic type. Second, I contend that specimen preparation techniques previously employed for stomatographic studies of Austrobaileya are inadequate. Upchurch (1984a)
and Baranova (1992a
, 2004)
based their results on the examination of cuticular preparations. However, not only does Austrobaileya have a highly grooved cuticular topography that obscures cuticular flanges, but cuticular flanges in many of the cells, particularly those associated with strongly specialized cells (e.g., the inner ring of bicyclic types), are weakly developed and difficult or impossible to see. Upchurch (1984a
, p. 531) mentioned "thinly cutinized tangential walls" in one figure legend, and, in another wrote "anomocytic (?) stoma." Carlquist (2001)
also placed a question mark after the word anomocytic in his table. This uncertainty argues against the use of these previous approaches for stomatographic studies in Austrobaileya. Furthermore, Upchurch (1984a)
reported a range of 4–6 contact cells per stomatal complex; however, my data show a range of 4–10. Complexes falling outside of the range Upchurch (1984a)
reported (i.e., 7–10) are not rare, but range from 28 to 46% across the four samples (Figs. 24, 25, 27, 29, 31, 32). My results strengthen the stomatographic evidence for a relationship between Austrobaileyaceae and Schisandraceae, as hypothesized by Baranova (1992a)
, who noted the presence of anomocytics in the former as the sole stomatographic difference between the two families. Also, the presence of polar subsidiaries in Austrobaileya further increases its similarity to Early Cretaceous (zone I) angiosperm cuticles, as the lack of these was mentioned by Upchurch (1984a)
as a dissimilarity.
Trimeniaceae were described by Money et al. (1950)
, Baranova (1992b)
, and Philipson (1993)
as paracytic, and although the majority of complexes in both species I examined are paracytic types, the other types present are important to recognize as they illustrate a connection to Austrobaileya where paracytics are uncommon. Laterocytics, stephanocytics, and irregular complexes are found in both families and serve to increase stomatographic evidence for a relationship between them. Some paracytic complexes in Trimeniaceae do not closely resemble those in other ANITA taxa; higher proportions of holoparacytic and paracytic complexes in which subsidiary cells meet at one pole but not the other are observed. These are rare or absent in other basal taxa.
Bailey and Nast (1948)
described Schisandraceae as having stomatal complexes conforming to paracytic and anomocytic types, as well as intermediate forms. Baranova (1983)
included Schisandraceae in the list of families with the laterocytic type (den Hartog and Baas, 1978
), noting many different laterocytic variants (e.g., 1 + 2, 2 + 2, 2 + 3), in addition to paracytic complexes. Upchurch (1984a)
offered a similar accounting and added that Schisandraceae (like Austrobaileyaceae) resemble Early Cretaceous (zone I) angiosperms except that they have invariably unmodified polar contact cells. Baranova (1992a)
concurred with this and added that anomocytic complexes are absent in Schisandraceae. My data confirm a high frequency of both laterocytic and paracytic types in the family and also the absence of the anomocytic type. However, along with laterocytics and paracytics, I note stephanocytic types, most of them bicyclic. Also, I have found that many complexes have polar subsidiaries: 8–43% of the complexes across the sample have full specialization of at least one of the stomatal poles (Figs. 42–52, 54, 56), some of which have strongly specialized subsidiaries (most notably, Figs. 43, 44, 47, 54). Jalan (1962)
likewise recognized polar subsidiaries in the genus. As with Austrobaileya, the present study's discovery of polar subsidiaries removes a point of supposed dissimilarity to Early Cretaceous (zone I) angiosperm fossils.
My recognition of stephanocytic types (including bicyclics; Figs. 57, 58, 60, 62, 63) in Illiciaceae contrasts to previous studies that found only paracytics, and less commonly, laterocytics. Bailey and Nast (1948
, p. 82) described the stomata of this family as, "accompanied by 2 to 4 subsidiary cells oriented parallel to them," a definition which would include paracytic and laterocytic types. Baranova (1972)
classified Illiciaceae as belonging to a group characterized by mostly paracytic stomatal complexes. However, Metcalfe (1987)
did mention that arrangements of subsidiary cells are often variable on a single leaf, an observation corroborated by my results.
In Chloranthaceae, Baranova (1987b)
and Kong (2001)
both stated that all investigated species of Hedyosmum are stephanocytic, an assertion generally supported by my results. The latter reported 100% of all complexes as stephanocytic in the two species he investigated. Interestingly, an earlier work by Baranova (1983)
claimed that laterocytic complexes were present in this genus, and I did observe a single laterocytic-like complex, which I coded as irregular-incomplete laterocytic (Fig. 64) in which strongly specialized lateral subsidiaries are closely aligned with specialized noncontact cells, suggesting formation by tangential division. Ascarina was described by both authors as cyclocytic; however, in the single specimen of A. lucida Hook f. I examined, I noted stephanocytic types, including many bicyclics (Figs. 66–68), as well as laterocytics, and a few paracytics and irregulars, but none I would describe as cyclocytic. Perhaps most of the stephanocytic types in my study would have been identified as cyclocytic by other authors who did not recognize the outer ring. Chloranthus spicatus Makino was described by Kong (2001)
as having 100% paracytic stomata. My data show a predominance of paracytic types, but also stephanocytic bicyclics (Fig. 69) and laterocytics, apparently formed by tangential divisions (Fig. 70).
The ancestral condition of angiosperms
My results reveal two divergent systems of stomatal architecture in basal angiosperms—systems 1 and 2. I hypothesize that each of the four basal clades examined here underwent an independent elaboration of the ancestral angiosperm stomatal architecture to form one or the other system. Specifically, the ancestral state can be called anomo-stephanocytic, a system with a continuum of forms ranging from those lacking subsidiaries (anomocytic), to loosely organized forms (e.g., weakly stephanocytic), to more strongly organized forms (e.g., stephanocytic, actinocytic) and intermediates (e.g., actino-stephanocytic). From this ancestral condition, Nymphaeales (system 2) underwent modification through elongation of radial walls of some complexes to form actinocytic types, and in a few, strong specialization of subsidiaries through tangential division. Otherwise, stomatal architecture in Nymphaeales remains close to the ancestral condition. System 1, seen in Amborellaceae, Austrobaileyales, and the chloranthaceous genera Ascarina and Chloranthus, arose through formation of tangential divisions in some contact cells to produce stephanocytic bicyclic, laterocytic, paracytic, and other forms. During the course of development, some complexes lost evidence of the outer ring, thus appearing as laterocytic, paracytic, or another stomatal type with strongly specialized subsidiaries. Also, a new type of stomatal development that gives rise to paracytic types by a different mechanism may have arisen (see next paragraph). A proposed transformation series is shown in Fig. 75. The evolution of stomatal architecture in the three clades with system 1 represents a common theme that is played out independently with variations, not only in basal angiosperms, but also in other seed plant lineages. Several lines of evidence support these hypotheses: further analysis of stomatal characters, the reconstruction of ancestral character states by MacClade, the early angiosperm fossil record, and evidence from the literature on other extant and fossil seed plant lineages.
|
noted in extant Ginkgo. Also, despite the fact that mature leaves were examined, Chloranthus spicatus and a few species of Illicium displayed a few dividing stomatal meristemoids indicative of mesogenous, specifically, mesoparacytic (Pant, 1965
) stomatal development (K. Carpenter, personal observation). This process may form paracytic stomatal complexes without undergoing the tangential divisions of anomocytic or stephanocytic precursors I hypothesize. Although this study is not intended to infer type of stomatal development at such early stages, it is important to note that evidence of mesoparacytic development, especially in these more derived taxa, does not undermine the present theory of modification through tangential divisions, nor that of anomo-stephanocytic ancestral architecture. It is possible that a parallel shift toward mesogenous (mesoparacytic) stomatal ontogeny, perhaps away from perigenous (Pant, 1965
) or some other ontogeny (see also Payne, 1979
), is occurring in both lineages, and possibly in Amborella as well. It is also possible that paracytic and other types, even within a single leaf in these taxa, do not all develop in the same way. This is observed in other taxa; see Baranova (1992b)
.
In evaluating these hypotheses, it is necessary to consider an alternative mechanism that has been proposed for the origin of laterocytic complexes. Baranova (1983)
suggested that the subsidiary cells of laterocytic complexes in any given taxon could be derived in one of two ways: they could arise from subdivision of paracytic subsidiary cells, or they could arise from "ordinary epidermal cells" (i.e., of an anomocytic complex). Thin radial walls in subsidiary cells are interpreted as evidence for the former, while thick radial walls are interpreted as evidence for the latter. While Baranova (1983)
never mentioned the possibility that laterocytic complexes could be derived from stephanocytic types, origin from anomocytic complexes is clearly much closer to my hypothesized mechanism. (Baranova (1987b)
noted that stephanocytic and anomocytic complexes are related and sometimes difficult to distinguish.) Evidence from this study strongly refutes the derivation of laterocytic subsidiary cells from subdivision of paracytic subsidiary cells in these taxa. First, radial walls are generally of equal or greater thickness than tangential walls (e.g., Figs. 30, 33, 36, 43, 47, 51, 59). Second, the subsidiary cells of many laterocytic complexes are distinctly unequal in size and/or shape (e.g., Figs. 8, 10, 33, 36, 42–44, 46–48, 53, 55, 59, 61, 70), a condition that would not be expected from a subdivided paracytic, a type that exemplifies "ontogenetic unity of the cells of the series" (Baranova, 1983
, p. 94). Third, the presence of latero-cyclocytic and other types with strongly specialized polar subsidiaries further argues against the subdivided paracytic hypothesis. For the subdivided paracytic hypothesis to account for these polar subsidiaries, those stomatal complexes having them would have to have been derived from a holoparacytic stomatal type; however, holoparacytics occur only in Amborellaceae and Trimeniaceae, and are rare even in these families. The fourth and perhaps most intriguing line of evidence lies in the irregular-incomplete laterocytic type (Fig. 3). These are seen in Amborellaceae (3–8% of the four samples; Figs. 4, 7), Austrobaileyaceae (0–1% of the four samples; Fig. 35), Schisandraceae (1–9%; Fig. 51), and Chloranthaceae (0–1%; Fig. 64). These suggest formation by a precursor that underwent tangential divisions in some but not all of the cells necessary for the formation of a laterocytic complex. Such complexes intergrade seamlessly with those scored as weakly paracytic (some of which have short lateral subsidiaries), and thus their reported numbers here may be something of an underestimate.
Reconstruction of ancestral character states
Basal angiosperms
Reconstruction of characters at internal nodes near the base of the angiosperm tree yields further evidence for these hypotheses. Despite the equivocal reconstructions for stomatal type at nodes 1, 2, 3, and 4 (Fig. 73), the reconstruction of two other characters suggests that the stephanocytic hypothesis should be favored, at least for nodes 2, 3, and 4. These characters, presence or absence of polar subsidiaries (Fig. 71), and presence or absence of radial orientation (Fig. 72), are reconstructed as present in the majority of stomatal complexes at these three nodes. Because these character states are generally lacking in both paracytic and laterocytic complexes, but present in the great majority of stephanocytic types, the reconstruction of these two characters supports the stephanocytic hypothesis and also illustrates the value of examining other stomatal characters in addition to type.
Unfortunately, the reconstruction of node 1 is not so straightforward. The ancestor of all angiosperms is reconstructed as equivocal for all three of these characters (Figs. 71–73). Despite this uncertainty, the results nevertheless suggest that stomatal complexes with stephanocytic-like organization played an important role in the evolution of stomatal complexes of extant basal angiosperms and likely represented the ancestral stomatal type from which, at the very least, the various types of stomatal complexes found in Nymphaeales, Austrobaileyales, and Chloranthaceae were derived. I argue that the scenario for stomatal evolution in Chloranthaceae is the same as for Amborellaceae and Austrobaileyales. Specifically, paracytic, laterocytic, and other stomatal complexes with strongly specialized subsidiaries (i.e., as in Ascarina, Sarcandra, and Chloranthus) represent advanced character states derived from the more primitive stephanocytic state found in Hedyosmum. In discussing character evolution in Chloranthaceae, Eklund et al. (2004)
, however, argue for the opposite scenario. Specifically, they claim the stephanocytic stomatal type in Chloranthaceae was derived from paracytic, which they argue represents the ancestral state for the family, although they offer no explanation for how this transformation was supposed to occur. This result was obtained with the assignment of paracytic as the most common stomatal type in the outgroup Austrobaileya, which my data indicate is incorrect (Table 1). Baranova (1992a)
likewise noted that paracytic complexes are uncommon in Austrobaileya.
Although node 1 is reconstructed as equivocal, I argue that the stephanocytic hypothesis is better supported by several lines of evidence and should be favored over the paracytic hypothesis. First, the equivocal reconstruction at this node is due solely to a predominance of paracytic types in Amborella. A look at the next three basal angiosperm lineages reveals that Nymphaeales lack paracytic types entirely, and taxa with high proportions of paracytic types are nested within Austrobaileyales, and Chloranthaceae, thus representing more derived groups. Second, although stephanocytic types occur less frequently in Amborella than in the other basal angiosperm clades, Amborella reveals clear examples of stephanocytic bicyclic types (Figs. 4–6, 13), illustrating that the tangential division underlies at least some of the stomatal diversity encountered in this species. Also, the proportion of complexes with polar subsidiaries (25–39%) seems quite high for a system that is primitively paracytic, while the presence of the irregular incomplete laterocytic type, and the high proportion of complexes with specialized noncontact cells (29–67%) are both to be expected in a system based on the tangential subdivision of cells of stephanocytic-like complexes. Furthermore, since Amborella trichopoda is likely a product of one of the oldest lineage-splitting events in all angiosperms and has an extremely restricted present-day distribution, it seems unlikely that it is the sole member of its lineage ever to have evolved (i.e., that there has not been extinction in the clade). Extinction within the Amborellaceae could obscure the pattern of stomatal evolution in this clade, one that I propose was similar to that in Austrobaileyales and Chloranthaceae. These clades both exhibit a similar pattern, i.e., one in which derived taxa exhibit mostly paracytic (as in Illiciaceae), or a mixture of paracytic and laterocytic, stomatal types (Sarcandra and Chloranthus). Both stomatal architecture and MacClade reconstructions indicate these were derived from the stephanocytic-like progenitors in the basal members of their clade: Austrobaileya in Austrobaileyales, and Hedyosmum in Chloranthaceae. Concomitant with this increase in paracytic and laterocytic types in Austrobaileyales is a decrease in the frequency of polar subsidiaries (a character inextricably linked to the stephanocytic stomatal type), with Illiciaceae having far fewer (8–30%) than Austrobaileyaceae (63–67%), for example. The same is true for Chloranthaceae: Hedyosmum has 99.5% of complexes with polar subsidiaries, Ascarina with 66%, and Chloranthus with 29%. Supposing all members of the Austrobaileyales except a given species of Illicium, Schisandra, or Kadsura had become extinct, we would be presented in many ways with a system of stomatal architecture similar to Amborella from which to make inferences. A high proportion of paracytic and/or laterocytic types in such a system may obscure the presence of other important but rarer types, such as stephanocytic bicyclics, which hint at an alternate explanation.
Reconstruction of ancestral character states
Expanded analysis
The inclusion of other lineages of extant and fossil seed plants in the analysis of character evolution provides further support for these hypotheses, and resolves node 1 with respect to stomatal type.
Because relationships of extant seed plants remain controversial, I have used two different hypotheses of seed plant phylogeny to examine the evolution of stomatal type. The "anthophyte hypothesis," that angiosperms and Gnetales together form a clade along with certain fossil groups, all of which share flower-like reproductive structures, has been supported by several analyses (e.g., Crane, 1985
; Doyle and Donoghue, 1986
, 1992
; Nixon et al., 1994
; Doyle, 1996
). Recent phylogenetic analyses based on sequence data from many different genes (Hansen et al., 1999
; Qiu et al., 1999
, 2000
; Winter, 1999
; Bowe et al., 2000
; Chaw et al., 2000
) refute this however, and indicate that Gnetales are related to, or nested within conifers ("Gne-Pine" hypothesis), which are themselves nested within a gymnosperm clade. In addition to analyzing character evolution with these topologies with only extant plants, I have examined two different placements of a key fossil taxon for each (giving six analyses total), and I have added eudicots as the next clade above Chloranthaceae (as in Doyle and Endress, 2000
).
The eudicot or "tricolpate" clade, contains the majority of extant non-monocotyledonous angiosperms. A profusion of taxa characterized by anomocytic stomata is seen near the base of this large clade, including Papaveraceae, Fumariaceae (Kidwai, 1972
), Lardizabalaceae, Berberidaceae, Ranunculaceae (Metcalfe and Chalk, 1950
), Nelumbonaceae (Gupta et al., 1968
). It therefore seems fitting to code the eudicot clade as primitively lacking paracytic stomatal complexes. In other living and fossil non-angiosperm seed plant lineages, the only occurrences of the paracytic type are in the gnetalian genera Gnetum and Welwitschia and in the Mesozoic seed fern group Bennettitales (Florin, 1951
), hence its importance in this analysis. Doyle's (1996)
analysis found the position of Bennettitales to be variable; in some most parsimonious trees it is sister to Caytonia + angiosperms, and in others it is related to Gnetales or to another clade. I have examined the first two possibilities here. Caytonia, another Mesozoic seed fern, was found to be sister to angiosperms (Doyle, 1996
) and is included here, as is Elkinsia, one of the earliest seed plants from the Paleozoic, which is used to root the entire seed plant tree (as in Doyle, 1996
). To simplify the analysis and avoid conflating different nonparacytic stomatal types in these lineages, I have redefined the character from "stomatal type" to "presence of paracytic stomata," a binary character with two states: present or absent.
All six analyses (one of which is shown in Fig. 74) unequivocally reconstruct the ancestor of extant angiosperms, as well as nodes 2, 3, and 4 as nonparacytic, leaving us to adopt the stephanocytic hypothesis for these nodes. This is true for all three resolving options in MacClade. Under the anthophyte hypothesis, it appears that paracytic stomata are never homologous in angiosperms and Gnetales, but have evolved independently in both groups. This is true with both placements of Bennettitales. Considering all three topologies, paracytic stomata could have evolved independently no fewer than four times in the included taxa, a number that would most likely increase considerably if one were to consider all angiosperms. The same is true under the Gne-pine hypothesis with only extant taxa, and with both placements of Bennettitales; paracytic stomata evolved independently a minimum of four times within the taxa examined. Thus, the evidence indicates that the ancestor of extant angiosperms had an anomo-stephanocytic stomatal architecture and that the first four clades of extant angiosperms each began with this ancestral state, and independently derived the more advanced types of system 1 or system 2.
Although the paracytic hypothesis for the ancestral state of angiosperms was probably the most widely accepted hypothesis prior to the placement of the ANITA grade at the base of the angiosperm tree (as in Takhtajan, 1969
; Baranova, 1972
, 1992b
; Cronquist, 1988
), some authors have advocated other hypotheses. In an earlier work, Cronquist (1968)
argued that the ancestral state was anomocytic, with later specialization giving rise to other types. Payne (1979
, p. 129) also argued against the paracytic hypothesis, and noted importantly that, "It may be appropriate to make a distinction, however, between occurrence in a primitive group, and occurrence in the progenitor," an observation that I find particularly relevant here in considering the predominantly paracytic stomata in Amborella. Upchurch (1984a
, b)
suggested that a uniformly paracytic condition (e.g., as seen in many Magnoliales) may represent an advanced state over the variable forms seen in Early Cretaceous fossil angiosperms and extant groups such as Amborellaceae, Austrobaileyaceae, Schisandraceae, and certain Chloranthaceae. This hypothesis is concordant with evidence from the present study. However, I argue that the varied forms seen in these groups do not represent the ancestral state, but rather represent a derived state themselves, arising from the ancestral anomo-stephanocytic system by tangential divisions.
The angiosperm fossil record
Given this anomo-stephanocytic system as ancestral in angiosperms, we may expect at least some early angiosperm fossils to resemble this condition, at least more so than the taxa with system 1 examined here, whose condition, I argue, is derived. Further, although we cannot rule out that a particular fossil taxon may exhibit a derived condition, we would not expect to see many early ones with a uniformly paracytic condition. In a study of early angiosperm leaf cuticles from the Early Cretaceous Potomac Group, Upchurch (1984a)
noted similarities in stomatal architecture between zone I (Aptian) angiosperms and extant Amborellaceae, certain Chloranthaceae, Austrobaileyaceae, and certain Schisandraceae. These similarities include a highly variable architecture encompassing numerous types and intermediates, observations supported by the present study. However, in possessing anomocytic complexes, many zone I fossils (e.g., Eucalyptophyllum oblongifolium, cf. Ficophyllum, and Dispersed Cuticle #1) show a more primitive condition than Amborella, where anomocytics are rare, and Austrobaileyales and Chloranthaceae, in which they are absent. Upchurch (1984a)
also reported anomocytic complexes in the later zone II (Albian) forms including Populophyllum reniforme Font. and Menispermites potomacensis Berry. Many other types of complexes in fossil taxa from both zones bear evidence of derivation from anomocytic or weakly stephanocytic complexes by tangential division. Forms similar to stephanocytic bicyclic and forms in which an inner cycle of cells appears derived from anomocytic-like complexes are seen in Upchurch's (1984a)
illustrations of the New Serrate Red Pointe and Sapindopsis/Platanoid patterns, as well as a representative "zone I pattern." However, in contrast to those in Amborella, Austrobaileyales, and Chloranthaceae, complexes with complete inner cycles of cells seem common in fossil taxa as inferred from Upchurch's (1984a, b)
illustrations. I suggest that complexes with this state precede those with incomplete inner rings (Fig. 75) and represent another aspect in which these fossils are more primitive than Amborellaceae, Austrobaileyales, and Chloranthaceae. He also notes two other features in the fossil taxa that are characteristic of systems based on tangential division: lateral subsidiaries that often do not extend the full length of the stoma (as in my irregular, incomplete laterocytic type) and the presence of specialized noncontact cells in complexes of most taxa. The one fossil taxon noted to lack specialized noncontact cells—cf. Ficophyllum—is reported to have a high frequency of anomocytic complexes, and as illustrated, bears some resemblance to Nymphaeaceae, whose system I inferred as being close to the ancestral condition of angiosperms. Also, given that the earliest angiosperm pollen dates from the Hauterivian or possibly earlier (Hughes and McDougall, 1994
), the zone I (Aptian) leaf fossils observed by Upchurch (1984a)
would have to be approximately 5–24 million years younger than the currently accepted date for the earliest appearance of angiosperms in the fossil record. Thus, the presence of some aspects of derived stomatal architecture in these fossils is not surprising. In terms of stomatal architecture, their intermediacy between Nymphaeaceae and the other three basal clades suggests some of these fossils may represent stem lineage taxa on the line to extant Amborella, Austrobaileyales, or the more derived Chloranthaceae (i.e., all except Hedyosmum). Alternately, they could be intercalated somewhere in between these lineages, in which
