EVOLUTIONARY RELATIONSHIPS

Taxonomically important character states are generally not morphologically complex in the Pottiaceae, e.g. leaf margin recurvature, presence or absence of a stem central strand or a hyalodermis, red color reaction to KOH, etc., see also discussion of Rohrer (1985, p. 230), for a similar problem with characters in the Hylocomiaceae. Character states are apparently easily re-evolved (or perhaps de-suppressed). This contributes much homoplasy or taxonomic “noise.” For many of these character states, functions, if any, are not clear, but their appearance may be pleiotropic to complex physiological functions.

The occasional presence of a twisted peristome in genus Ditrichum (Ditrichaceae; cf. discussion by Robinson 1968) is difficult to interpret as indicative of either primitive or advanced status for this character in respect to the base of the Pottiaceae tree. There is, apparently, evidence for reticulate evolution in at least some mosses (Wyatt et al. 1992). On the other hand, the Pottiaceae has many morphological and anatomical characters, and an analysis based on these many characters ought to provide a testable evolutionary hypothesis or at least point to an acceptable classification.

The Pottiales

Fleischer (1920) defined the Pottiales in the commonly accepted modern sense as: small, mostly turf-forming, upright plants; leaves radially arranged on the stem, always with a costa, seldom much-elongate or setaceous, of one layer of cells; upper laminal cells parenchymatous; sporophyte acrocarpous, seldom cladocarpous (i.e. borne on a short side branch); capsule mostly upright and smooth, commonly cleistocarpous; peristome single or absent, of 16 teeth, the dorsal layer thicker than the ventral, usually papillose, sometimes with a basal membrane. Families included were Calymperaceae, Syrrhopodontaceae, Encalyptaceae, and Pottiaceae. Robinson (1971b) on the other hand placed the genera of the Pottiales in a much larger Dicranales, a judgment the present study tends to support.

Shaw et al. (1987, 1989) have indicated that Buxbaumia (Buxbaumiaceae) and Diphyscium have a haplolepideous peristome. The relationship of Buxbaumia to Diphyscium (Diphysciaceae) is clear in its capsule shape and peristome (cf. Taylor 1962, p. 205; Crum & Anderson 1981). Diphyscium is similar to families of the Pottiales (see also Shaw et al. 1987, p. 68) in its gametophyte morphology, without unique characters (the bistratose laminal cells are staggered, overlapping each other as in Timmiella). The basal membrane of the pleated peristome does not have the tesselations of that of the high basal-membraned species of Pottieae, but the ornamentation is similarly branching-spiculose (note that Shaw et al. 1989 showed evidence that the haplolepideous peristome is homologous to the cilia of Bryum-type peristomes, not the endostome segments). There seems to be no problem with the inclusion of both the Buxbaumiaceae and the Diphysciaceae (including Diphyscium, Theriotia and Muscoflorschuetzia) in the Pottiales along with Calymperaceae and Encalyptaceae (the latter includes Bryobartramia, but see cogent remarks on the non-haplolepideous character of the peristome of Encalypta by Edwards 1979, p. 342). The leaves of Buxbaumia aphylla Hedw. are deep red in KOH. The three species of Diphyscium seen in this study had laminae reddish orange in KOH, although the cell walls of the laminae of young leaves were often uncolored or merely vaguely pinkish. The cladograms of Mishler (1986b) and Mishler and Churchill (1984) show the Buxbaumiales as the closest sister group for the Bryales.

Another family worthy of classification near the Pottiales (but see Vitt 1984) is the Timmiaceae, of which the single genus, Timmia, has an inner peristome strikingly like the high basal-membraned, fenestrate peristome typical of many Tortula (s. lat.) species, except that the inner peristome of Timmia is more coarsely granulose and does not show the 2:3 primary peristomial layer to inner peristomial layer ratio pattern characteristic of the Pottiaceae and the Haplolepideae in general as discussed by Edwards (1979). The gametophyte has many features otherwise characteristic of the Pottiaceae: ligulate leaves with ventrally mamillose and dorsally flat cell walls, sheathing leaf base; upper laminal cells isodiametric and bulging ventrally but only weakly convex dorsally; and the costal transverse section with two stereid bands, hydroid strand, guide cells, and ventral and dorsal epidermises well differentiated. In addition, the laminal KOH reaction is uncolored in young leaves (laminae clear green in toto, the walls are uncolored) with reddish orange blotches in patches, especially on the leaf base (T. megapolitana Hedw.), but pink in mature leaves. Akiyama and Nishimura (1993, p. 191), on the other hand, felt that the Timmiaceae was only remotely related to the Bryineae on the basis of branch development analysis.

Encalypta species examined (E. ciliata Hedw., E. mutica Hagen, E. procera Bruch, E. rhaptocarpa Schwaegr., E. spathulata C. Müll., E. streptocarpa Hedw. and E. vulgaris Hedw.) in the course of this study have KOH color reactions characteristically bright yellow (becoming red-orange in older leaves) in the upper lamina and deep red in the basal cells, and species of Encalyptaceae have inner peristomes with high basal membranes (Horton 1982). Edwards (1979) indicated, however, that Encalypta, because it did not have the characteristic 2:3 pattern of peristomial development, was probably not a member of the Haplolepideae, while Shaw et al. (1987, p. 68) indicated a close phylogenetic relationship between the Encalyptaceae and Buxbaumiaceae based on their own work and that of others. In any case, a high basal membrane is probably not an advanced character in the Pottiaceae, since four acrocarpous families with papillose or ventrally mamillose, isodiametric upper laminal cells (Buxbaumiaceae, Diphysciaceae, Encalyptaceae and Timmiaceae) have high basal-membraned inner peristomes. Of these families, all have either reddish orange or deep red laminal KOH color reactions (in Encalypta restricted to the basal cells), which indicates that a yellow KOH reaction is an advanced character.

Edwards' (1979) indication that the Ptychomitriaceae (two stereid bands, yellow in KOH) belong with the Haplolepideae is agreeable on general morphological grounds: none of the major gametophyte features would be out of place in the Pottiales, while the sporophyte is also similar to those of the Pottiales, excepting only the large, plicate, mitrate calyptra of Ptychomitrium.

Species of Calymperaceae examined reacted to KOH with a yellow color. Edwards (1980a) characterized the Calymperaceae as having hyalocysts abruptly differentiated from the green upper laminal cells and having resorption pores, teniolae (a strong, mostly intramarginal leaf border) present, and propagula borne on the lamina. Only the presence of resorption pores is constant, according to Edwards. All the characters noted by Edwards (1980a) are present in the Pottiaceae. In combination, however, these characters are rather different than those of the Pottiaceae. Reese (1987b) listed several advanced characters of the Calymperaceae (Pottiales), including tropical environment, corticolous habitat, well-defined cancellinae (basal “windows” of hyaline, inflated laminal cells) with superficial pores, reduced or absent peristomes, broad and sheathing leaf bases, and propagula usually present. The differences between Calymperaceae and Pottiaceae are discussed at length by Reese (1987b) and Reese and Zander (1988). Hypodontium (q.v.) is here included in the Pottiaceae although previously long accepted as Calymperaceae (see also discussion of Calymperastrum); Reese (1987b) felt that Hypodontium is intermediate between Calymperaceae and Pottiaceae.

Regarding evidence of gametophytic characters, certain families with somewhat similar gametophytes and haplolepideous peristomes (Calymperaceae, Encalyptaceae, Dicranaceae, Ditrichaceae and Seligeriaceae) are shown by Kawai (1968) to have both one and two stereid bands, and costal morphology with clear differentiation into dorsal and ventral epidermises and medial guide cells. Other families gametophytically similar to the Pottiaceae (Timmiaceae, Ptychomitriaceae) have two stereid bands. Other, less closely related families (Mniaceae, Polytrichaceae) have one or two stereid bands. The Orthotrichaceae, which often have the general appearance of Pottiaceae, have mostly a single stereid or substereid band and poor costal differentiation into cellular layers. The Bartramiaceae, on the other hand, have a single band and good cellular differentiation, as have certain genera of Grimmiaceae. Pleurocarpous families appear to have internally fairly undifferentiated, thin costae. Given the often rather high degree of costal differentiation of closely related families and of such relatively primitive families as Polytrichaceae (see cladogram of Mishler 1986b), one might expect that the Pottiaceae and related families were derived from ancestors in which two stereid bands and possibly a high basal membrane, too, had already been evolved. Additional evidence against convergence is that certain pottiaceous taxa normally having a single stereid band (species of Pseudocrossidium, Tortula s. lat., Streptocalypta) may develop a few stereid cells between the ventral epidermis and the guide cells, an exact positioning more indicative of partial desuppression (cf. Basile & Basile 1984) of a ventral stereid band than of de novo elaboration of one. Thus, the line of costal evolution in Pottiaceae probably began with taxa with two stereid bands and progressed through loss of the ventral band to those with a single strong stereid band. The fact that some taxa that usually have two stereid bands may have only one band in occasional collections of small stature is probably not apropos here, since genera with only one band may be quite large in stature (e.g. Crumia, Hennediella, Syntrichia and Tortula) in some species and have their own reduction series involving sporophyte characters and stem length but without much variation in costal morphology. Reduced sporophytes are correlated with small gametophyte stature (but cf. Willia).

Cladogram 1, discussed at length below, of both the genera of the Pottiaceae and haplolepideous and diplolepideous genera selected for having gametophytes similar to those of the Pottiaceae, with scored character states restricted to those found in the Pottiaceae and using Polytrichum (Polytrichaceae, which has many of the anatomical and morphological characters of the Pottiaceae) as outgroup, shows most non-pottiaceous genera derived from lines of the Pottiaceae. This may not show the relationships of non-pottiaceous genera to each other, since non-pottiaceous characters are not used, but it at least indicates likely relationships based on shared apomorphies of these genera with the Pottiaceae. The distance in steps of non-pottiaceous genera from the root of the tree indicates the degree of weighting of particular pottiaceous or non-pottiaceous (and not used) characters that would be necessary to bring these genera down to the base of the tree; it also indicates that certain genera (e.g. Ceratodon and Encalypta) high in the tree have a large number of advanced characters that can only doubtfully be accommodated by weighting. It will be important to study details of the peristomes of all taxa if sufficient additional characters are to be found to justify viewing the Pottiaceae as presently constituted as monophyletic.

Intrafamilial relationships of the genera

It will be apparent from the discussions presented here that there are probably far more genera worthy of recognition than are presently accepted; however, much more work will be necessary to adequately identify and describe these additional genera. In the absence of a massive cladistic evaluation at the species level, there are two obvious places to look for unrecognized genera; first, as transformation series within large genera, such as Trichostomum (q.v.), and, second, as rather different species brought together in small “wastebasket” genera, such as Gyroweisia. If clusters of closely related species were recognized as genera or sections, there would be at least double the number of supraspecific taxa presently accepted. On the other hand, segregation of such clusters would leave an overabundance of species not assignable to any segregate and remaining in the original genera by default.

“Good” characters may be variable. Many species in the Barbuleae vary in production of a second stereid band (usually in well-developed plants or larger leaves). Thus, it is the potential for development of a second stereid band that provides this traditionally major character. A complication is the occasional production of a second stereid band in some species otherwise clearly belonging to the Pottioideae. One of the aims of this study was to examine the possibility that separating genera at the suprageneric level on the basis of presence or absence of a second stereid band may mask relationships of closely related species losing a second stereid band in a reduction series. Certainly small plants of Didymodon and Gymnostomum, among other genera, have only one stereid band. It was found, however, that there were few (e.g. Calyptopogon and Streptopogon) clear-cut overall morphological similarities between species traditionally distinguished by numbers of costal stereid bands that might be inferred as being due to a close phylogenetic relationship marred only by lack of a second stereid band, i.e. the number of stereid bands is a valuable character.

A “separate evolution of gametophytic and sporophytic characters” (cf. discussion by Rohrer 1988) is often invoked to explain incongruent character assemblages. Although sporophyte characters have much utility at the suprafamilial level in the mosses, only a few such characters in the Pottiaceae were found important at the generic level (cf. treatments of Aschisma, Trachycarpidium and related genera) in the Pottiaceae. Many character states once widespread among ancestors of present-day mosses may now be suppressed or lost. In the Pottiaceae, few uncommon characters, if any, are found in one genus and not also in another, apparently distantly related, genus. Striking convergent evolution is apparent in a large number of cases. This may even include the one most obvious “unique” character in the family, the twisted peristome of many of the genera. Although Mishler (1985b, p. 389) could identify no apomorphies for the Pottiaceae, the twisted peristome may be considered a synapomorphy for the family if genera of apparently reduced and simplified morphology, including shortening (and subsequent straightening) or loss of the peristome are considered derived, or if such genera are considered relicts of larger lineages with members having twisted peristomes now extinct. A twisted peristome, in any case, is present in widely divergent pottiaceous genera and not in others, and it is present in two species of Ditrichum (Ditrichaceae), namely D. tortipes (Mitt.) Par. and D. ambiguum Best (cf. Crum & Anderson 1981, Grout 1927 and Robinson 1968). These two species of Ditrichaceae (discussed further under treatment of Barbula) have a distinctly twisted peristome much like that of the Pottiaceae, and the laminae of these species have yellow KOH reactions. These two species or the genus Ditrichum itself may actually belong near Barbula sect. Hydrogonium. One might note, in this respect, that Distichium (Ditrichaceae) has a pottiaceous peristome, albeit mostly untwisted.

Apparent convergences may be genetically complex: one might cite the production of a propaguliferous leaf apex in Leptodontium proliferum, Tortella fragilis, and Syntrichia angustifolia with associated elongation of the upper marginal laminal cells, and the fleshy capsules of Tridontium (here placed with the Scoulerioideae of the Grimmiaceae) and Dialytrichia; but more usually such convergences involve apparently simple traits, such as basal margins of lamina serrulate in Eucladium verticillatum, Molendoa hornschuchiana and occasional collections of Gymnostomum aeruginosum; the lack of a ventral stereid band in Calyptopogon, Pseudocrossidium, Streptopogon and Tortula; systylious capsules in Hymenostylium and species of Hennediella (taxa previously recognized under Desmatodon).

Although certain families of mosses (e.g. Archidiaceae, Ephemeraceae) are rightly characterized by relatively small sporophytes with short setae and cleistocarpous capsules, the assumption that these characteristics are conservative in the Pottiaceae has proven to be unacceptable to certain modern authors, e.g. those dealing with the Weissia-Hymenostomum-Astomum complex. In the Pottieae, Pterygoneurum is an example of a genus with so distinctive a gametophytic character (the ventral costal lamellae) that species with sporophytes with peristomes and species without peristomes have long been acceptable within that genus. Elsewhere in the Pottieae, species with eperistomate capsules have occasionally been thrown by previous authors into genera (e.g. Tortula) that are generally peristomate but with gametophytes that are fairly heterogeneous, and the complexity of these groups of disparate species has discouraged further work by traditional methods. Such reduction is evident in other acrocarpous families; for instance, Fife (1985), in his treatment of the genera of the Funariaceae, stated that “The dominant evolutionary trend in the Funariaceae has been the progressive simplification of the sporophyte.” He associated large stature, small spores and “elaborate morphological structures for the regulation of spore release” in patchy environments through wind dispersal, being most suitable for opportunistic species, and at the other end of the spectrum small stature, large spores and lack of peristomes or even opercula is associated with growth in habitats widely and continuously or at least periodically available, often characterized by disturbance or flooding. A parallel is evident in the Pottiaceae, although the gametophytes of the Pottiaceae are not nearly as uniform in morphology as those of the Funariaceae. This last supports my apprehension of several independent reduction series. On the other hand, Mishler and Churchill (1987) emphasized that “postulating reduction series¼depends on the use of outgroup comparison, not appeal to an a priori evolutionary scenario¼,” a caveat addressed below in the phylogenetic analysis.

Longton (1988) reviewed the literature on life history strategies in desert-dwelling bryophytes and noted that “most bryophytes are poikilohydric: they have little control over rates of water uptake or loss, but in compensation the gametophytes of many species can tolerate severe and prolonged cytoplasmic desiccation.” He found several trends, among them: “¼towards reduced gametophyte life-span, early and prolific production of spores and/or asexual propagula, and monecism in sexually reproducing species: shifts from K- to r-selection and from tolerance to evasion of environmental stress” and “¼towards reduced duration of habitat availability but with habitats tending to recur predictably within a community. Parallel trends towards increase in spore size and loss of feature[s] promoting spore dispersal in regularly sexually-reproducing species.” The apparent morphological reduction series seen in some genera of Pottiaceae (notably Hennediella, Tortula and Weissia) is possibly associated with increased fitness in the environments of arid lands.

Many nomenclatural novelties are presented here, most being just transfers for long-unstudied exotic species necessary to reflect modern generic constructs, but many others involve recognition of probable evolutionary tracks often signalled by unique character states or character state combinations.

A beginning at a new synthesis based on recognition of apparent transformation series that may cut across traditional generic limits is indicated in the discussion of Trichostomum, Weissia and related genera, and the Phascum-Pottia-Desmatodon-Tortula complex. An alternative view emphasizing the taxonomic utility of peristome morphology in the Pottiaceae is presented at length by Hagen (1929, p. 8–13), probably as a reaction to Dixon's (1924) extensive generic lumping in the family. Reduction of the sporophyte in Pottiaceae (and other acrocarpous families colonizing dry areas) may be assessed as an evolutionary adaptation correlated with two environmental influences: (1) capsules may be exerted to their detriment above the boundary layer of stagnant humid air for considerable periods of time (sporophytes generally take one year to mature, cf. discussion by Zander 1979d), and (2) fitness of mosses with massive diaspores, such as large spores and cleistocarpous capsules, or with diaspores developed close to the substrate (”atelochory” in the terminology of van der Pijl 1972, and “precinctiveness” of Carlquist 1966, 1974) may be improved in patchy environments (cf. discussion by Zander 1979f).

Phylogenetic evaluations: discussion

Phylogenetic opinions of past authors were apparently based on one or more of a combinations of “marker characters,” being unique or rare shared characters, of presumed great phylogenetic weight, to distinguish genera and higher taxa. The mass of non-weighted characters were apparently assumed to be considerably less important and prone to abundant convergence and parallelism. One must evaluate, however, the assumption that morphological convergence between taxa in a particular (especially stressful) environment may involve the evolution quite naturally of a combination of many characters the existence of which in clearly unrelated taxa in the same environment must otherwise be imagined as improbably fortuitous. (Thus the particular low weighting of 22 reduction-related characters in the cladistic analysis below.)

Taxonomic evaluation of intrageneric relationships is here mostly based on general phenetic similarity; when transformation series are clear, however, the generic limits as given presumably reflect probable intrageneric phylogenetic relationships, i.e. grouping those taxa most readily derived from common ancestors through evolution of traits that are unique or rare in the immediate group. In the case of what appear to be extant little-different descendants of ancestral species, the decision whether to keep closely related species representing a branching series together in one genus or assign them (perhaps arbitrarily) to their respective, different phyletic lines is here usually made in favor of the latter. Thus, Pseudocrossidium crinitum is quite closely related to the genus Barbula (differing mainly in the shape of the section of its dorsal stereid band and in the sheathing perichaetial leaves), yet because it may be construed as a stem species for the Pseudocrossidium transformation series, it is placed with that genus.

It may also prove fruitful to examine the Barbuleae (as was begun by Saito in his lucid treatment of the distinguishing characteristics of Barbula and Didymodon) and identify transformation series that begin in genera with well developed peristomes and end in those with reduced or absent peristomes. This may result in some splitting of the well-peristomed genera and possible disappearance of some eperistomate genera, e.g. Gymnostomum and Gyroweisia. A typical transformation series is described in a synopsis of Pseudocrossidium (Zander 1979f) in which a south to north transformation series from the Andes to the Arctic showed that P. revolutum and P. hornschuchianum belonged in the genus even though lacking many of the more salient characters (but possessed of the distinctive costal section anatomy).

Unfortunately, in this light, revisions of single genera of the Pottiaceae will commonly not reveal complete transformation series. If such series are found to be worthwhile as contributing to the clarification of generic limits, future revisionists must resign themselves to careful examination of several genera at once to distinguish the full range of potential series. At present, the genera of the Pottiaceae are in some cases neither “natural” nor instructive of possible phylogenetic relationships, and, in fact, many present generic definitions seem to cut across natural lines. Exceptions are probably those genera with sporophytes that have unusual (rare or autapomorphic) features or modifications (mainly of the peristome) different than what may be interpreted as mere reduction. These include Aschisma (exothecial cells in palisade-like rows and stomates absent), Leptodontium (and other taxa with unvaryingly smooth teeth), Steptotrichum (elaboration of internal lateral cell walls interior to the teeth), Leptodontiella and Trachyodontium (teeth cleft into several rami), Uleobryum (and other genera with pustulate exothecial cells), and Weisiopsis (16 spaced teeth).

In the absence of a species-based cladistic study, genera are here characterized either by an observed decided lack of species of transitional morphology between two groups (a phenetic gap) or by some obvious division of evolutionary directions (non-rigorous evaluation of salient advanced characters). It is, of course, most satisfactory when both obtain, such as is the case with some of the smaller, isolated genera like Tetrapterum. Transitional species between apparent major evolutionary thrusts, however, are disconcertingly common. It is clear that cladistic evaluations are capable of generating more rigorous (based on defined methodologies) and more reasonable classifications than an “omnispection” technique grouping taxa with general overall resemblance. This study attempts a classification based on cladistically derived lineages of genera as these are presently delimited.

An emphasis on using only presumed derived characters (Hennig 1966) as a basis for phylogenetic analysis has been extensively rationalized recently, and has been used in studies of the mosses (e.g. Bremer, 1981; Cao & Vitt 1986; Churchill 1981, 1985; Frahm 1991b; Koponen 1968, 1973, 1982; Mishler 1985b,c, 1986a,b; Mishler & Churchill 1984, 1985b; Rohrer 1985, 1988; Waters et al. 1992, and others). On the other hand, cladistics has become a new orthodoxy (see also Heywood 1983 and Robinson 1986a) that, in sweeping clean, has discounted the value of past classifications; even taxonomy is eliminated from one recent definition our field: “¼phylogenetics and biogeography, together referred to as systematics” (Erwin 1991)! Judging from the good match of the three traditional Pottiaceae subfamilies (Trichostomoideae, Barbuloideae and Pottioideae) with the results of the present study, formal parsimony is reflected to a significant extent in past attempts at the creation of a simplest classification that takes into account the available data with emphasis on shared characters or character state combinations, even if the effort does not involve a large data set, base relationships only on derived characters, and use an exact algorithm. It is expected that the generic assignments of infrageneric taxa will be supported to a considerable extent by future species-level cladistic studies, except in genera pointed in the taxonomic treatments as clearly heterogeneous (e.g. Barbula, Didymodon, Hyophila, and Trichostomum).

Plausible phylogenetic relationships are satisfactorily inferable within identified transformation series, but the direction of evolution is sometimes far from obvious. A series might start in the middle, at either end, or in two or more places. Assuming, however, that similar complexes of many apomorphic characters are more likely to be derived from shared ancestors than evolved anew for each set of individuals supports an interpretation of most transformations as reduction series. This does not imply that reduction is a process that occurs more easily than elaboration, but that an explanation of a transformation series as reduction requires the assumption of fewer parallel evolutionary events than does elaboration.

There are two ways to view “splitting” and “lumping.” The first is to see these as two different ways of deviating from good taxonomy. The second, and better approach, is to see these as analytic and synthetic methods of dealing with information, splitting thus being appropriate at exploratory stages and lumping at revisionary stages.

Many of my publications attribute rather wide variation to certain species and species complexes (e.g. Didymodon fallax complex, Didymodon vinealis complex, Hymenostylium recurvirostrum complex, Molendoa sendtneriana s. lat.). For example, see charts of variation in Hymenostylium (Zander 1977c, Zander & Eckel 1982), which demonstrate independently varying character states and are surely evidence of considerable genetic variation. It should not be surprising that such intraspecific phenetically distinctive swarms exist and can be expected to exemplify complex intraspecific genetics inasmuch as genetic expression in mosses is expected to be essentially genotypic in the gametophytic generation. Additionally, the considerable variation within species or species complexes confounds easy identification. Often, characteristics that may not always be present, but are nevertheless distinctive, may be helpful. For instance, peculiar to the Didymodon vinealis group is the usual absence of the quadrate ventral costal cells at the extreme leaf apex, resulting in a short, boat-shaped groove bottomed by epapillose elongate cells. Hymenostylium species commonly have longitudinally elongate medial laminal cells, but this is not always the case.

Species are here seen as inferential populational units composed of individuals sufficiently similar to form a probable biologically predictable unit, and when this is not clear (such as when species are known from single gatherings or when taxonomy is doubtful), a species is put forward as representing a presumed “basic taxonomic unit.” Genera are more or less clear-cut apparent phyletic series of species or groups of such phyletic series, or simply groups of similar species when no evolutionary series are evident. Unusual characters are viewed here with some diffidence: the lens-like central thickening of the exothecial cells of Byroceuthospora, Uleobryum and Trachycarpidium is a clearly advanced feature, yet this is not a complicated character and may well have been developed through evolutionary convergence. If these three genera were seen, however, as a single suprageneric taxon, the quite different morphology of the gametophytes would require that they be considered remnants of a once larger, complex group with taxa of intermediate morphology now extinct. Since these are arid-land genera with austral affinities, such a view is an acceptable possibility and the cladistic analysis below supports it.