| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Structure and Development |
School of Biological Sciences, Washington State University, Pullman, Washington 99164-4236 USA
Received for publication February 21, 2002. Accepted for publication June 21, 2002.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: Amorpheae • corolla • Fabaceae • flowers • homology • Papilionoideae • stamens • synorganization
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Endress, 1994
; Raven and Weyers, 2001
), which we will discuss as synorganization (Hufford, 1996
). Examples include the evolution of the corolla tube from separate petals in the asterid clade and the evolution of syncarpy from separate carpels among early eudicots. Although understanding the evolution of synorganized structures is critical to understanding the diversification of floral form, synorganization can obscure structural homology, making inferences about evolutionary diversification problematic. When the synorganization involves dissimilar structures, such as adnation among floral whorls, homology can be particularly difficult to assess.
Some aspects of synorganization are well known in papilionoid legumes, which include the tribe Amorpheae. For example, the fusion of stamen filaments that results in various conditions ("monadelphy," "diadelphy," and others) has commonly been used in classifications and identification keys for the group. Synorganization between abaxial lateral petals forms the distinctive papilionoid "keel"; synorganization among sepals forms the calyx tube. Less commonly recognized in papilionoid flowers is a region of synorganization below the perianth and androecium, i.e., an hypanthium or floral cup, that surrounds the base of the superior ovary. Some members of Amorpheae have, in addition to these regions, an unusual example of synorganization involving petals and stamens. In Dalea and Marina, the four lateral petals (wing and keel petals) appear attached to the fused filament tube (Bentham, 1865
; Hutchinson, 1964
; Barneby, 1977
). Although homology of the tissue below the insertion points of the petals has not been explicitly assessed, Barneby (1977
, p. 1), in his monograph of four Amorpheae genera, suggests that "the daleoid petal ... appears to be elevated on an outgrowth of receptacular origin."
Our study of the petal-androecium synorganization in Amorpheae is important for several reasons. First, it is a unique arrangement of floral organs; although petal-stamen synorganization is common (e.g., the epipetalous stamens of sympetalous Asteridae), cases in which free petals apparently attach to fused stamens are extremely rare (although a similar-appearing condition occurs in Linaceae [Sharsmith, 1961
]). Second, the synorganization is important to taxonomy within the tribe: species assigned to Dalea and Marina have been described as having petals inserted above the rim of the hypanthium, whereas the remaining taxa are said to have all petals inserted on the hypanthium (Hutchinson, 1964
; Barneby, 1977
). Petal-stamen synorganization, therefore, could be a synapomorphic innovation. Finally, accurate statements of homology are prerequisites for meaningful characterization of floral diversity in Amorpheae.
Definitions of homology have long been debated (e.g., Donoghue, 1992
; Hall, 1994
and references therein). Here we use homology to mean similarity by descent, rephrasing Lankester's (1870)
homogeny, defined as similarity due to inheritance from a common ancestor (Donoghue, 1992
). We choose this definition because of our interest in the processes of evolution, of which inheritance is central. Additionally, defining homology as "similarity by descent" allows us to be explicit about criteria that we use to assess homology. For structures to be comparable among taxa, they need to pass the two tests implied by this definition. First, the structures need to be similar, or structurally homologous. To assess structural homology, we can apply the three principal criteria of Remane (1952)
: correspondence of position, correspondence of special attributes, and the presence of intermediates (Riedl, 1978
). Special attributes could include, e.g., histological characteristics, surface features, or gene expression data. Second, these similarities must have arisen by the process of descent (with or without modification), i.e., they must be phylogenetically homologous. To assess phylogenetic homology we can apply the single criterion of having been inherited from a common ancestor (the congruence test of Patterson, 1988
). To understand the evolution of floral form in Amorpheae, we first need to assess structural homology in order to make correct comparisons of attributes; this is the focus of the current paper. Next we need a phylogenetic hypothesis, in order to assess phylogenetic homology of those attributes found to be structurally homologous; this will be the focus of future work.
Amorpheae comprises eight genera, about 240 species, and is thought to be monophyletic, based on the presence of several morphological characteristics (Barneby, 1977
) and preliminary phylogenetic analyses (McMahon and Hufford, 2000
; McMahon, 2002
). Flowers of Amorpheae demonstrate notable variation, particularly in the context of the fairly conserved papilionoid floral form. Within Amorpheae, only the small genus Psorothamnus has flowers that are typically papilionoid, having five petals, strong bilateral symmetry, and reproductive organs enclosed by the postgenitally fused keel petals. Other genera in Amorpheae have flowers that are open and nearly radially symmetric (most Errazurizia, Eysenhardtia, and Apoplanesia), and still others have flowers with one petal (Amorpha) or no petals (Parryella and Errazurizia in part). Most species in the remaining two genera, Dalea and Marina, have flowers that appear papilionoid but for the synorganization described above.
The goals of this paper are to present the morphology, development, and distribution of the petal-stamen synorganization in Amorpheae and to assess structural homology of the floral parts involved. In particular, we will address three interrelated questions: (1) In the flowers that have petal-stamen synorganization, what is the region below the insertion points of the petals and above the rim of the hypanthium? For ease of discussion, we will refer to flowers that have this condition as "daleoid" and this region as the "stemonozone," a term also applied to petal-stamen synorganization in mimosoid legumes (Lewis and Elias, 1981
). Narrowing our focus to Dalea subgenus Kuhnistera ("prairie clovers," about 22 species), in which five fertile stamens alternate with four sterile, laminate structures, we will ask, (2) Do the prairie clovers have staminodes, i.e., are these sterile structures homologous to stamens or to petals in other daleas? Finally, we wish to pursue a suggestion made by Barneby (1977)
that the daleoid flowers may be the result of a secondary derivation of the papilionoid form. The range of variation among all Amorpheae flowers includes corolla forms that are not papilionoid; this led to a concern about corolla form polarity in the group. Therefore, we will ask, (3) Are daleoid flowers structurally homologous to other papilionoid flowers, or is there structural evidence to suggest non-homology, i.e., a secondary derivation of the general papilionoid form?
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
, 1994
) is here used to denote an evolutionary process in which the condition of fused organs is derived from the ancestral condition of free organs (Hufford, 1996
). Developmentally, the condition of fusion can occur two ways: postgenitally or congenitally. During postgenital fusion, organs that were free at the start of development become fused; this process generally involves epidermal contact between structures (Cusick, 1966
; Sattler, 1978
). During congenital fusion, growth occurs under individual organs or their primordia, raising the organs together (Sattler, 1978
). This process is more descriptively called zonal growth and involves no observable contact among organs (Cusick, 1966
).
Sampling
We selected three taxa, exemplifying different floral forms, for morphological and developmental study. For each taxon we dissected at least 75 flowers at various stages of ontogeny. For the purpose of comparison, we will first describe the adult morphology of a flower with no corolla-androecium synorganization, exemplified by Psorothamnus kingii. Second, a flower is described with corolla-androecium synorganization, exemplified by Marina maritima. Finally, we include a prairie clover, Dalea candida, to allow discussion of homology for this floral form in the context of the other two floral forms (with and without a stemonozone). Following the descriptions of the adult morphologies, we characterize floral development of these species. Adult flowers of other Amorpheae are briefly compared to these three species in order to describe the distribution of the stemonozone, including its presence in species of Psorothamnus for which no such synorganization had been reported previously. For a list of species examined and voucher information, see the Botanical Society of America website (http://ajbsupp.botany.org/v89/).
Morphology
We studied early ontogeny and major structures in later ontogeny primarily using scanning electron microscopy (SEM), and we observed details of mid-stage and late stage ontogeny primarily through serial sections. Inflorescences bearing anthetic and pre-anthetic flowers were collected from natural populations and preserved in FAA (37% formalin, glacial acetic acid, absolute ethanol, and water, 5 : 5 : 45 : 45). Flowers and flower buds were dissected in 50% ethanol, dehydrated in a graded ethanol series, critical point dried, and examined using a scanning electron microscope at 10–20 kV. Whole flowers and flower buds were dehydrated in a graded tertiary butanol series, embedded in Paraplast, and serially cross and longitudinally sectioned at 8–20 µm for anthetic flowers or 6–12 µm for buds, mounted on glass slides, and stained with safranin and fast green. To determine the order of initiation of zonal growth, we examined cross and longitudinal sections and documented the presence of observable tubes, i.e., the products of zonal growth. If a particular structure is observed without other structures present (and never the reverse) that structure is the first to begin developing. For example, if some buds had only a calyx tube and others had a hypanthium as well as a calyx tube, we considered this to be evidence that the calyx tube initiates earlier than the hypanthium.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
). Petals of P. kingii are inserted directly on the rim of the hypanthium (Figs. 3 and 4).
|
|
Dalea candida
The general form of the Dalea candida flower, in contrast to the forms of the previous two species, does not resemble a "typical" papilionaceous flower (Fig. 9). The androecium and corolla are similar to neither P. kingii nor M. maritima. The androecium of D. candida has only five fertile stamens, fused for about one-half of their length in a symmetrical open sheath (Figs. 10 and 15). The free portions of the stamens alternate with four petaloid structures at the apex of the sheath (Figs. 9 and 10). The banner petal is medial and adaxial and shaped similarly to the banners of the previous two species, with a narrow claw and a laminate, cucullate, distal portion. The four petaloid structures also have narrow proximal portions and laminate distal portions. However, these laminate blades are dissimilar to the blades of the wing and keel petals of the other two species described: the blades of D. candida are flat, with no proximal auricle nor sculpturing (as in wing petals), no marginal fusion (as in keel petals), and all four are similar to one another, not distinguishable as wings and keels. The placement of the four petaloid structures is exactly between the five stamens; the claws attach to the top of what appears to be androecial column, in the sinuses between the free stamens (Figs. 9, 10, and 12). We will refer to the region between the insertion of the petaloid structures and stamens and the hypanthium as stemonozone (Fig. 15), although the homology of this region will be discussed later. Other than the corolla and androecium, the flower of Dalea candida is similar to the other two in having a synsepalous calyx, a single carpel, and a hypanthium, although the hypanthium is relatively short (Fig. 11).
Other Amorpheae
All species of Marina investigated (http://ajbsupp.botany.org/v89/) are similar to M. maritima in having relatively short hypanthia and stemonozones (e.g., M. calycosa in Fig. 16). As described by Barneby (1977)
, all but one Marina (M. vetula) have a region below the petal insertion points that is much shorter than the stamen sheath (i.e., the petals are attached low in the flower); our results confirm this condition for the taxa studied.
|
). We studied all species of the genus Psorothamnus and found that two species have forms similar to P. kingii, P. arborescens, and P. polydenius (Fig. 19), in which petals attach directly to the rim of the hypanthium. However, we also found that a very short petal-stamen tube occurs in some members of the genus. Psorothamnus spinosus, P. thompsonae, and P. emoryi (Fig. 20) have petals that attach in such a way that the outside of the petal touches the rim of the hypanthium (where the calyx attaches), but the inside of the petal claw attachment is above the rim. Psorothamnus thompsonae and P. scoparius (Fig. 21) have petal attachments entirely above the rim. Additionally, in P. fremontii, though the lateral and abaxial petals attach on the rim of the floral cup, the banner attaches just above the rim, as in Dalea and Marina.
Floral development
Psorothamnus kingii
Flowers are initiated singly in the axils of bracts (Fig. 22), situated along a short shoot, the apex of which becomes spine-tipped. A pair of bracteoles forms at the apex of each pedicel (Figs. 22 and 23). Sepals initiate unidirectionally (sensu Tucker, 1984
), starting abaxially (Fig. 23). The sepals grow relatively quickly and cover the meristem. Petals and the carpel initiate on the convex meristem (Fig. 24). Last to initiate are the stamens (Fig. 25), beginning with the abaxial antesepalous stamen (Fig. 24). Based on the relative sizes of the stamen primordia shortly after initiation, it appears that the antesepalous stamens initiate unidirectionally (Fig. 25); this may be the case for the antepetalous stamens as well. The adaxial antepetalous stamen lies to one side of the adaxial petal (Fig. 25). As the carpel cleft deepens, the petals elongate and broaden laterally (Fig. 26). At this stage, the antesepalous stamens overtop the antepetalous stamens and the lateral petals (Fig. 26). Later in development, as all organs continue to develop, the petals elongate more extensively and enclose the androecium and gynoecium (Fig. 27).
|
|
|
|
In the flowers of M. maritima, zonal growth occurs in four regions (Fig. 58). The first region to initiate zonal growth is that below the sepals. As this zone elongates, the sepals are raised, and a congenitally fused calyx tube is formed, eventually surrounding the base of the flower (Figs. 34–36). The second region to initiate zonal growth is that below the stamens (Figs. 40–42). This occurs after anther differentiation, while the thecae are forming. Zonal growth below the filaments occurs nonuniformly, with less growth near the banner petal. This results in a stamen tube that is continuous just above the hypanthium, but distally discontinuous with the interrupted portion on the adaxial side of the flower. As zonal growth occurs, the ten stamens are elevated on a staminal column (Figs. 37 and 38), almost entirely open on one side (Fig. 39). The third region to initiate zonal growth is below the petals and the staminal sheath, forming the stemonozone (Figs. 43–45). Again, this growth is nonuniform; the keel petals are elevated more than the wing petals, and the banner is elevated the least (Figs. 38 and 39). In this way, the petals become attached to what appears to be fused filaments. The fourth region of zonal growth is below all organs except the carpel. As this region elongates, the stamens, petals, and sepals are elevated on a common tube of tissue that forms a cup, the hypanthium, around the ovary (Fig. 7). We infer from relative bud size that these processes of zonal growth occur in quick succession. For example, the buds that have a calyx tube, a stamen tube, and a stemonozone (Figs. 43–45) are generally only slightly larger than the buds with a calyx tube, a stamen tube, and no stemonozone (Figs. 40–42). As the flower matures, further elongation occurs in all four of these regions. At anthesis, the M. maritima hypanthium is very short, and the stemonozone is shorter than the stamen sheath, i.e., the petals appear to attach low on the staminal column (Fig. 6).
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Forms of stamen connation
Traditional classifications of papilionoid androecia focus on the patterns of stamen connation among the (usually) ten stamens (Polhill, 1981b
). In diadelphous flowers, nine stamens are fused in a sheath and the adaxial stamen remains free. In pseudomonadelphous flowers (Gillett, 1961
), the adaxial stamen is fused only to its neighbors above the base, leaving a pair of gaps between the adaxial stamen and its neighbors at the base of the androecium. In monadelphous flowers, all stamens are fused. However, monadelphy can take two forms: the stamens can be fused into a complete tube, or, as in the Amorpheae flowers studied here, they can be fused into a sheath, open (usually) on the adaxial side. Distinctions among the traditional classifications (monadelphy, diadelphy, and pseudomonadelphy) lie in the number of stamens fused together. An alternative view would be to focus on symmetry. In one form of monadelphy, the complete tube, androecial symmetry is radial, unlike any other androecial form in papilionoid legumes. In the other form, the sheath, androecial symmetry is bilateral, or approximately so. Therefore, the appropriate comparisons for Amorpheae androecia may be not with other androecia designated monadelphous, but with other bilaterally symmetric androecia.
We have shown for M. maritima and Psorothamnus kingii, and infer for the remaining sheathed taxa, that the sheath develops by zonal growth in an incomplete band of tissue below the stamens, with a gap between the adaxial stamen and one of its neighbors. This zonal growth occurs after initial organ elongation and differentiation, so the anthers are formed and the filaments are distinct atop the elongating sheath. This developmental pattern is similar to that seen in other sheathed taxa (e.g., Machaerium villosum [Klitgaard, 1999
]). It is also similar to that which produces the sheath portion of the androecium in diadelphous taxa (e.g., Glycine max (L.) Merrill [Crozier and Thomas, 1993
]) and pseudomonadelphous taxa (e.g., Pisum sativum [Tucker, 1989
], Psoralea pinnata L. [Tucker and Stirton, 1991
]).
Petal-stamen synorganization
Petal-stamen synorganization occurs in a variety of forms, most commonly as epipetalous stamens. Epipetalous stamens are commonly initiated on a floral apical meristem as primordia separate from petals, and the epipetaly develops through zonal growth below the corolla and stamens (e.g., Vinca rosea [Boke, 1949
], Synthyris, and Besseya [Hufford, 1995
]). This is similar in a very general way to the pattern of petal-stamen synorganization seen in the daleoid Amorpheae; the major difference appears to lie in the fact that in the asterids mentioned, the corolla is sympetalous and the androecium consists of free stamens, whereas in the daleoids, the androecium is synandrous and the corolla consists of free petals.
A very few papilionoid taxa outside the Amorpheae are known to have petal-stamen synorganization. A member of the Dalbergieae, Inocarpus fagifer (Parkinson) Fosberg, has a highly unusual sympetalous corolla (with strap-shaped petal lobes) on which ten epipetalous stamens are inserted, similar in general appearance to an asterid flower (Polhill, 1981a
). In a single species from the tribe Psoraleeae, Psoralea leucantha F. Muell. of Australia, the "corolla and staminal tube are fused for a short distance at the base" (Grimes, 1990
, p. 16). This description does not clarify whether the corolla forms a tube that is fused to the staminal tube, or the petals are free and inserted on the staminal tube (as in Amorpheae), nor can we say whether the fusion is postgenital or congenital. The clearest example of a similar situation in papilionoids is seen in Trifolium pratense L. (Picklum, 1954
; Retallack and Willison, 1990
). Picklum's cross sections show petal-stamen synorganization similar to that seen in the daleoids. Retallack and Willison also found a "staminal-corollary tube" but did not describe its morphology or development in sufficient detail to allow comparison with the daleoid stemonozone. We have found in another clover with a corolla-stamen tube, Trifolium macrocephalum (Pursh) Poir., that the petals peel away from the fused region and are not set in sockets, as in the daleoid flowers. This structural difference may be significant; however, further study is required to know if there are developmental differences as well. Though these cases of petal-stamen synorganization in papilionoids are interesting, they are clearly isolated by phylogenetic distance from Amorpheae (Kajita et al., 2001
; Pennington et al., 2001
) and are therefore unlikely to be homologous (i.e., similar by descent) to the daleoid synorganization.
Assessing homology in Amorpheae
Armed with a morphological description of the petal-stamen synorganization of daleoid flowers, the ontogeny of three flowers in Amorpheae, and an idea of the context in which these occur, we now turn our attention to three specific questions of structural homology. First and most important, (1) What is the stemonozone? Is it homologous to receptacle, to androecium, or are there other possibilities? Because the identity of the stemonozone is not immediately obvious, its presence has led to interpretations of floral structures that we wish to redress via the remaining two questions. (2) Do the prairie clovers have staminodes? That is, are the four lateral petaloid structures in these taxa modified stamens (Moore, 1936
; Faegri and van der Pijl, 1979
) or are they modified petals (Wemple and Lersten, 1966
; Barneby, 1977
)? (3) Does the presence of the stemonozone indicate that the daleoid taxa reinvented a papilionaceous flower (Barneby, 1977
)? Conversely, are the papilionaceous daleoid flowers structurally homologous to other papilionoid flowers?
Two approaches help us answer our questions of homology. By comparing across morphologies, we can describe similarities in special attributes and in relative position and evaluate possible intermediates (Remane's [1952
] principal criteria of homology [Riedl, 1978
]). By comparing morphologies through development, we can additionally describe similarities in timing, as well as positions at initiation and throughout development. None of these are absolute arbiters of structural homology. Rather, they are similarities that we would expect to observe if the genetic programs that underlie the structures were, indeed, inherited in common.
(1) What is the stemonozone?
Structures that are the products of zonal growth below multiple whorls of organs have long presented difficult homology problems. The stemonozone, above the hypanthium and below the corolla and the androecium, is just such a structure, similar to hypanthia (below corolla, androecium, and calyx), gynostegia (below androecium and gynoecium), or the walls of inferior ovaries (below corolla, androecium, calyx, as well as involving gynoecium). Of these, homology of the walls of inferior ovaries has received the most attention (Kaplan, 1967
; Kuzoff, Hufford, and Soltis, 2001
), concentrating on the fundamental question of whether they are homologous to shoots (receptacular) or to appendages (appendicular), a question relevant to our case as well.
Barneby (1977)
tentatively suggested that the stemonozone was an upgrowth of the receptacle. Positionally, it is below corolla and androecium, where we find receptacle tissue in other papilionoids. However, it is above (distal to) the insertion point of the calyx and distant from the base of the gynoecium, inconsistent with it being receptacular tissue. Special attributes of the stemonozone favor an appendage interpretation: the form and the histology of the structure are indistinguishable from androecium, as indicated in the common description of the flowers as having "epistemonous petals" (Barneby, 1977
). This is not to say that it is androecial tissue, but the form of the structure lends weight to an appendicular interpretation rather than receptacular. Remane's third criterion, the presence of intermediates, is more difficult to apply in this case because the question has not been asked for any other legumes with petal-stamen synorganization.
Kaplan (1967)
described two sets of criteria for interpreting an appendicular origin of the inferior ovary in Downingia bacigalupii Weiler and the receptacular origin for that in Pereskia aculeata Mill.: (1) the pattern of vasculature and (2) relationships between developmental events and processes. In Downingia, vascular bundles branch below the ovary; the wall of the inferior ovary therefore contains distinct bundles that lead to stamens and perianth parts. In Pereskia, vascular bundles branch above the ovary; some of these branches lead to perianth, some to stamens, and some loop down the inside of the ovary wall (recurrent vascular bundles) to the placentae. The positional difference in vascular branching is used to infer an important difference. The recurrent vasculature in Pereskia supports the idea that the receptacle has spread out and up, forming a concavity on which floral organs form. Nonrecurrent vasculature in Downingia supports the idea that the border between appendage and receptacle is below the ovary, so the ovary wall is appendicular.
Using the inferior ovary wall example as a model, we can develop expectations for the daleoid stemonozone. If the stemonozone is receptacular, then the plesiomorphic receptacle would have spread out and up, as in Pereskia. Moreover, if the vasculature reflected this change, then we would expect to see vascular patterns that include complexities such as recurrent bundles. Instead, we see vascular bundles that are separate, leading to the stamens and to the petals, branching below the stemonozone (Fig. 14). This configuration is similar to that in legumes without stemonozones and shows nothing that indicates upgrowth of the receptacle. Therefore, evidence from daleoid vasculature is more consistent with an appendicular interpretation, as in the case of Downingia. Thus, the special attribute of vasculature, in addition to general form as described above, supports an appendicular interpretation for the stemonozone.
If we are correct that the stemonozone is appendicular, then what sort of appendage is it? Key alternatives are androecium, corolla, both, or neither (i.e., a new appendage altogether). Position of the stemonozone argues against it being solely androecium or solely corolla, because it is below both corolla and androecium. If homology of the stemonozone is with both whorls, we might be able to observe histological differences among regions as seen in Eucnide aurea (A. Gray) H. J. Thomps. & W. R. Ernst (Hufford, 1988
). This may be the case for some Trifolium flowers that have petal-stamen synorganization, in which the petals easily peel away. However, in the daleoid stemonozone, there are no regional differences in tissue histology (data not shown), nor is there tissue that peels (Fig. 8); therefore, we have no evidence for a compound structure. Mature morphology of the stemonozone provides evidence for appendicular homology but no evidence to homologize the stemonozone with particular appendages.
Evidence from development was critical in the inferior ovary example described above: "The differences in the two types of inferior ovaries reflect a difference in the time when the floral bud becomes concave. In Downingia bacigalupii formation of the floral cup occurs in conjunction with organ initiation, whereas in Pereskia aculeata floral cup formation takes place after and independent of initiation of appendages" (Kaplan, 1967
, p. 1288; emphasis added). Because the floral cup forms after the organs form in Pereskia, Kaplan says that this floral cup is receptacular; coincident initiation of cup and organs results in an inference of appendicular affinity. This implies that we can use the timing of events to infer independence or nonindependence of events, and, further, that independence informs homology inference.
Kaplan was interested in the relationship between organ initiation and the development of synorganization. In the daleoids, however, this relationship tells us little: in all taxa studied, organ initiation occurs prior to any synorganization. Instead, it may be more important to focus on relationships among synorganizations. As discussed above, daleoid flowers have some synorganization in common with other papilionoids. Hypanthia are frequent in papilionoids, as are a variety of forms of staminal fusions. Are any of these ontogenetic events related to the daleoid petal-stamen synorganization?
One possibility is that the daleoid flowers have a developmental program homologous to, in modified form, the developmental program underlying the androecium in other taxa (Fig. 59, scenario 1). The fused portion of the androecium (distal to the stemonozone) is similar in development and in mature form to the fused portion of the androecium found in other papilionoids. Perhaps, in the daleoid flowers, the zonal growth involves not only the tenth stamen, but the petals as well.
|
A third interpretation is that the developmental program that produces the synorganization is novel (Fig. 59, scenario 3). This interpretation differs greatly from the previous two by invoking the addition of a developmental program, rather than a modification of a preexisting program (either the program that produces a hypanthium or the program that produces a sheathed androecium). As evidence against this interpretation, the zonal growth below the petals and stamens produces tissue that closely resembles androecium, which might be unexpected if the developmental program was entirely novel.
Therefore, based on the data at hand, we hypothesize that the stemonozone is a product of a modified androecial development program (Fig. 59, scenario 1). Comparison of development across several more taxa will allow us to investigate this hypothesis further. If, for instance, the timing of the zonal growth below the stamens and petals always follows close behind the onset of growth below the stamens (as in M. maritima) but is relatively uncorrelated with the growth of the hypanthium, the case for scenario 1 (Fig. 59) would be strengthened. Opposing evidence, i.e., correlation with hypanthial growth but not with staminal sheath growth, would support the model shown in Fig. 59, scenario 2. If the timing of the zonal growth below the stamens and petals is not correlated to any other process of zonal growth, then it may be possible to infer that the underlying developmental program is a novel addition, as in Fig. 59, scenario 3.
We conclude that the stemonozone is appendicular, but beyond that, we cannot say exactly to which organ or whorl of organs it is homologous. Instead, we favor the interpretation that the developmental process producing the stemonozone is a modified form of the process that produces the androecium. For the case of the stemonozone, perhaps for other synorganized structures as well, inferring an evolutionary source for its underlying developmental process allows a clearer idea of its homology than does focusing on the mature structure itself.
(2) Do the prairie clovers have staminodes?
The second homology issue we will address concerns a set of approximately 22 species of Dalea, subgenus Dalea section Kuhnistera (Lamarck) Barneby, previously known as the genus Petalostemon Michaux, or "prairie clovers." In this group of species, distributed from southern Canada to northern Mexico but most diverse in the U.S. Great Plains, the total number of organs is reduced relative to most of the genus Dalea. Furthermore, the identity of four of the petaloid organs has been questioned.
As described above (see description of Dalea candida in RESULTS), only five fertile stamens are present, not the common ten (or more rare nine). The corolla consists of at least one petal, the banner (medial adaxial) petal. In addition to the banner, four sterile laminate structures are present, but whether they are staminodes or petals has been debated (Moore, 1936
; Wemple and Lersten, 1966
; Faegri and van der Pijl, 1979
). The interpretation that these structures are staminodes is due, in part, to their positions between the free filaments of the five antesepalous stamens. These positions would be occupied by four of the five stamens that are present in other synandrous papilionoids but are absent in the prairie clovers. Also, all four of the structures are quite similar to one another, in most cases not distinguishable as "wing" and "keel" petals. However, the structures are petaloid: sterile, laminar, and frequently colored like the banner.
Applying Remane's (1952)
criteria, we find that the petaloid structures, at maturity, are positionally most similar to antepetalous stamens; they alternate with the sepals (as do petals) but they are on the same circumference as the stamens. However, early in development, these structures are in positions centrifugal to the antesepalous stamens (Figs. 48–51), which we would expect for petals. Special attributes of the structures include claws and lamina, in common with petals but not with stamens. These laminae form early in development (Fig. 51) and later enclose the stamens (Figs. 52–54), as do petals in other Amorpheae. Another special attribute lies in the attachment of the structures. They attach to the flower in distinct sockets (just as petals of other daleoid flowers), unlike the smooth, jointless attachment of stamen filaments. When calling the structures staminodes, Faegri and van der Pijl (1979)
and Moore (1936)
were essentially weighing the positional criterion more highly than the criterion of special attributes. Furthermore, they did not have key data on early development, so the positional criterion was applied only to the flower at maturity. Turning to the third criterion, we find that there are no cases that could be interpreted as intermediates for the staminode interpretation of the structures, i.e., there are no other known taxa in Amorpheae with staminodes. However, daleoid taxa have petals attaching at a variety of heights on the "staminal" column, from very close to the hypanthium (e.g., Psorothamnus emoryi; Fig. 20), to very close to the filament separation point (e.g., Dalea grayi [Barneby, 1977
]). This range of attachment heights represents a set of intermediate conditions between the laminar structures in section Kuhnistera and petals in other papilionoids.
Wemple and Lersten (1966)
examined mature flowers from all species of Petalostemon and many species of Dalea and interpreted the structures as petals. Their conclusion was based on the presence of the sockets and on the presence, in some species, of subtle differentiation among lateral appendage pairs, as in wing and keel petals. Our results, using developmental information and applying explicit criteria for structural homology, agree with that of Wemple and Lersten (1966)
, strengthening the case for the presence of petals in the prairie clovers, i.e., that the laminar structures are structurally homologous to petals and not to staminodes.
(3) Are daleoid flowers homologous to those of other papilionoids?
We will now address whether the flowers found in Dalea and Marina may represent an independent derivation of the papilionoid form, as tentatively suggested by Barneby (1977)
, or if they are structurally homologous to other papilionoid flowers. The criteria of position and special attributes are key. There are many attributes that Dalea and Marina share with other papilionoids. For most daleoid flowers, the number of organs in each whorl and their relative positions are the same as found in most papilionoids (five sepals, five petals, ten stamens, single carpel) although there are exceptions, as discussed above. The process of organ initiation demonstrated by the three daleoids studied is similar to that seen in other papilionoids (Picklum, 1954
; Sattler, 1973
; Tucker, 1984
, 1989
; Derstine, 1988
; Retallack and Willison, 1990
; Tucker and Stirton, 1991
; Klitgaard, 1999
). In most papilionoids studied, organ initiation is generally unidirectional (Tucker, 1984
), often with overlapping initiation among the whorls (Tucker, 1989
). Also, the adaxial petal of the daleoids is external in bud, producing cochleate vernation, just as in other papilionoids. Moreover, most daleoid flowers are strongly zygomorphic (i.e., the corolla is differentiated into a banner, two wings, and two keel petals), as is typical for papilionoids. These characteristics are powerful similarities because they are not widely distributed among eudicots; indeed, zygomorphic cochleate flowers distinguish the subfamily Papilionoideae (Polhill, 1981b
; but see Herendeen [1995]
for exceptions in Swartzieae). Daleoid flowers, therefore, share special attributes with other papilionoid flowers. However, similarity can arise via homoplasy. Indeed, Barneby's question was whether Dalea and Marina arrived at this general floral appearance via homoplastic evolution. Finding no major structural differences outside of the petal-stamen synorganization is reassuring but negative evidence.
Therefore, we must turn to the final criterion of structural homology: the presence of intermediate forms. For this we consider the flowers of Psorothamnus. Some species have a short zone of petal-stamen synorganization (e.g., P. emoryi) whereas others have no such zone (e.g., P. kingii). The flowers of those Psorothamnus species that lack petal-stamen synorganization are similar to those of other papilionoids whereas those with a short stemonozone are intermediate between the daleoids and typical papilionoids. In other respects, however, Psorothamnus flowers are similar to other papilionoids: same organ number, strong zygomorphy, and cochleate vernation. This bears substantially on the question of homology between daleoid flowers and other papilionoid flowers: by encompassing a series of intermediate forms, Psorothamnus provides evidence for morphological continuity in the zone of petal-stamen synorganization. Therefore, the third criterion is satisfied, and we infer that the flowers of Dalea and Marina are structurally homologous to other papilionoids and probably do not represent an independent derivation of the papilionoid facies, as tentatively suggested by Barneby. Reconstruction of the Amorpheae phylogeny will allow us to confirm homology, provided we can infer the ancestors of the group to have been papilionoid.
Conclusions
Structural homology is essential for understanding the evolution of form. In our study, evidence from the morphology of structures, at maturity and throughout development, and evidence from the timing of developmental events allow us to infer homology at several levels in the daleoid Amorpheae. We infer the daleoid stemonozone to be appendicular and to be related to the androecium developmentally, and we infer that the prairie clovers have five petals and five stamens, rather than one petal and nine stamens. Finally, we infer the daleoid flower to be structurally homologous to other papilionoid flowers. These inferences provide testable hypotheses for phylogenetic study of the group.
|
FOOTNOTES |
|---|
2 Author for reprint requests, current address: Section of Evolution and Ecology, One Shields Avenue, University of California, Davis, California 95616 USA (mmcmahon{at}mail.wsu.edu
)
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Bentham G. 1865 Leguminosae. In G. Bentham and J. D. Hooker [eds.], Genera plantarum, vol. 1, 434–600. Lovell Reeve, London, UK
Boke N. H. 1949 Development of the stamens and carpels in Vinca rosea L. American Journal of Botany 36: 535-547
