J OURNAL of the A DELAIDE B G OTANIC ARDENS AN OPEN ACCESS JOURNAL FOR AUSTRALIAN SYSTEMATIC BOTANY flora.sa.gov.au/jabg Published by the STATE HERBARIUM OF SOUTH AUSTRALIA on behalf of the BOARD OF THE BOTANIC GARDENS AND STATE HERBARIUM © Board of the Botanic Gardens and State Herbarium, Adelaide, South Australia © Department of Environment, Water and Natural Resources, Government of South Australia All rights reserved State Herbarium of South Australia PO Box 2732 Kent Town SA 5071 Australia J. Adelaide Bot. Gard. 18(2): 107-160 (1998) NOTES ON HIBBERTIA (DILLENIACEAE) 2. THE H. ÁSPERA - EMPETRIFOLL4 COMPLEX H.R. Toelken State Herbarium of South Australia, Botanic Gardens of Adelaide, North Terrace, Adelaide, South Australia 5000 Abstract The Hibbertia aspera group has been re-assessed and the characteristics of the hairs in particular examined in some detail. Eleven species and five subspecies are recognised, illustrated and keyed out on vegetative as well as floral characters. The following new taxa are described: H. appressa, H. aspera DC. subsp. pilosifolia, H. decumbens, H. empetrifolia (DC.) Hoogland subspp. radians and uncinata, H. hirta, H. hirticalyx, H. notabilis, H. rhynchocalyx and H. truncata. The new combination, H. cinerea, is published for Pleurandra cinerea R. Br. ex DC. and H. pallidiflora Toelken is added to the flora of south-western Victoria. Introduction This species complex was selected to demonstrate, in the first of a number of publications planned as a revision of the species of Hibbertia in southern and eastern Australia, the problems encountered in the taxonomy of this genus. Although several previous attempts were made to re-assess the classification of this complex, the group shows the neglect typical of the taxonomy of Hibbertia, and even more clearly the lack of understanding of species in the genus. It is not clear why the taxonomy of this genus, often with such spectacular flowers, has not engendered the same great interest and popularity that collectors have shown for it. It seems the early history at least in part accounts for this lack of appreciation. De Candolle (1817) recognised six species in this complex of which one, Pleurandra ovata, had been described by Labillardière (1806). Sprengel (1827) formally published Sieber's Pleurandra astrotricha. J.D. Hooker (1855) was the first to express concerns about the similarity of species and regarded the complex (as P. ovata sens. lat.) as "an extremely variable plant, whose forms I have felt repeatedly inclined to separate however fmd no difference whatever in the flowers and fruit of all these varieties". He thus reduced P. scabra to varietal rank of P. ovata followed by Mueller (1862) who reduced most of the known species to his Hibbertia billardierei, which was to replace Pleurandra ovata when transfered to Hibbertia. While the latter only remarked that some "forms of P. ovata may be distinguished as varieties", Bentham (1863) described five varieties mainly based on leaf shape and tomentum. A similar trend developed at generic level as Mueller (1862) also reduced the genera Adrastaea DC., Candollea Labill. and Pleurandra Labill, to synonymy of Hibbertia, and Bentham continued by amalgamating Hemistemma DC., Ochrolasia Turcz. and Hemistrephus Drummond with Hibbertia but retaining Adrastaea, Candollea and Pachynema. Finally Gilg (1893) recognised only the genera Hibbertia and Pachynema in the Dillenioideae-Hibbertiae in Australia, but some of the names were retained at infrageneric level. This status at genus and species level was maintained for almost one hundred years (Gilg & Werdermann 1925; Stebbins & Hoogland 1976, in discussion) except that Druce (1917) added to the confusion by publishing the illegitimate combination H. ovata (Labill.) - non H. ovata Steudel (1845). Domin (1928) then transferred al! Bentham's varieties to this species. Very few new taxa were described until Wakefie/d (1957) resuscitated, as the result of his original research and especially field observations, two species in this complex, namely H aspera with stellate-tomentose undersurface of the leaves, and H. astrotricha for those 107 H.R. Toelken J. Adelaide Bot. Gard. 18(2) (1998) plants with mainly few hooked simple hairs on the undersurface. He introduced for the first time the importance of different hair types found in this group, but unfortunately he did not follow it throug,h the whole range. Nor does he seem to have been aware of the difference in juvenile plants (cf. juvenile to adult developmental stages) at least in some species of the group. Hoogland (1974) maintained this concept but established that the oldest name of the latter species should be H. empetrifolia. He had consulted for the first time in about 150 years the de Candolle types but from his preliminary identifications and manuscripts (CANB but at present on loan in AD) there are no indications that he intended recognising more species in this complex. Even with Wakefield's initiative for original research, publication of species remained restricted to some obviously new taxa for local floras largely described out of context from existing species (e.g. Conn, 1990; Reynolds 1991; Toelken 1996). Even these species were largely delineated on local material. It is extraordinary that nobody re-assessed the number of stamens, a character mentioned by Labillardière (1806) in the protologue of P. ovata, and used by de Candolle (1817) to separate P. scabra var. fi from P. ovata, yet J.D. Hooker (1855) could "find no difference whatever in the flowers and fruit" between the various forms which he did not formally recognise. This is but one example where superficially similar plants belong to different species. Without a large number of specimens, it would seem, earlier botanists could not appreciate the range of variation of individual species and were consequently unable to discern distinguishing characters. It is also important to note in the delimitation of taxa that putative hybrids between taxa were found in field observations to be extremely rare, if at all present in some groups of Hibbertia, and Stebbins and Hoogland (1976, p. 150) report similar findings. Fortunately the attractive flowers maintained a constant stream of collections over time so that the present re-evaluation is based on a broad range of material which allowed a better overview of the wide range of local variation usually found in most of the species. As a result the geographic variants can now be individually examined as well as seen as part of the variation of each species. Local variation may seem to link species but that is only part of the whole picture. For instance, the hairs on the leaves of plants of H. empetrifolia north of Sydney are often extremely similar to those of H decumbens (cf. H empetrifolia subsp. uncinata) but their stamen characteristics still distinguish them. It seems that previous investigators were often misled by convergent evolution of some characters. The large amount of material consulted here allowed such an overview, but also showed that much more discerning collecting is still needed to record and assess as much as possible of the local variation. The author made some field observations but in the H. aspera group the work was mainly restricted to South Australia and only some species in Victoria were examined in their natural habitat. One is continually overwhelmed by the variation found which has often not been recorded previously. It is hoped that immediate publication of completed groups of species will encourage feedback and hopefully provide an incentive for more discerning collecting, observations on the still not appreciated diversity of the floral biology and general basic ecological observations, so that in the short time alotted for the presentation of a flora write-up a similar standard can be achieved to that of treatments of other plants already published in the Flora of Australia. Characters A wide range of characters is used in this re-evaluation of a group of closely related species, but only a few need further explanation. Many of these will also be applicable to other groups of species but the present discussion refers only to the H. aspera group. 108 J. Adelaide Bot. Gard. 18(2) (1998) Hibbertia aspera group (Dilleniaceae) Habit The habit of the species of this group is plastic as in most taxa of Hibbertia. The plants usually produce wiry to slightly woody branches, which, when young and actively growing, produce branches of the first and second order with very much elongated internodes and usually larger leaves. Older plants have quite a different habit because of elaborate branching, some of which is di- or trichotomous (as discussed under inflorescence below) and a decrease in the size of the leaves and internodes. Plants tend to have a decumbent habit but the cane-like branches often scramble into other vegetation. Records of plants up to 2 m or more are known for H. cinerea, H. hirticalyx and H. truncata. The report of 4 m high plants of H. pcdlidiflora (Toelken 1995), is an extreme case. These plants often develop a dense cover on top of others. Although suckering is common among some Hibbertia species, it was only observed in H. pallidiflora and H. truncata, but the decumbent branches of H. empetrifolia were often found rooting at the nodes where they were touching soil. Since plants of these species often occur in groups it is likely that vegetative spread is much more common than has been previously recorded. Juvenile developmental stages The morphology of juvenile organs on seedlings often differing drastically from adult growth has been recorded for many plants (e.g. Lubbock 1892). In Hibbertia such characters are not only found in seedlings but they are sometimes retained for some time, and often intermediate forms link them to adult stages. Juvenile traits may reappear with new growth, such as, coppicing branches after burning. In the extreme case of H. pallidiflora, juvenile growth is in some areas maintained for a long time and/or juvenile characteristics re-appear with the first leaves of every new branch produced. The stages in the development were found best shown in the changes of the indumentum, especially since in this study particular emphasis was placed on an understanding and subsequent use of the hairs as a tool for identification of different taxa. The full range of variation of particularly the leaves was examined and illustrated, but the general patterns found suggested that available herbarium material is incomplete and, it is hoped, that more discerning collecting will supplement lcnowledge before a systematic arrangement of the species is attempted Therefore only the two extremes, the juvenile (not 'seedling' because . plants examined were often more advanced but it does include intermediate stages) and adult characters are formally described, and an intermediate range can then be extrapolated from the following general patterns. Whenever intermediate stage/s were available, one was illustrated but these are not necessarilly at a similar stage in different taxa. They were included to assist with an understanding of the complete range of character statesfotincl in a taxon. Reasons for a longer or shorter retention of juvenile/ intermediate characters can at present not be assessed and seem to be different for every species, and sometimes even between local forms within species. With limited herbarium material and field studies available for the present study it was assumed here that the stages of, for instance, the hooked simple hairs to stellate hairs on the undersurface of the leaves followed similar patterns in all the species with such adult leaves, although some of these juvenile stages have not been recorded for all these species. Question marks were inserted in the illustrations to show where stages are presumed unknown, as opposed to those species where apparently these developments do not take place either because the species concerned is presumably primitive, e.g. H. decumbens, or, in the case of H. empetrifolia, seem to have retained some of their juvenile characters (cf. affmities below). The following stages are usually observed: 109 H.R. Toe[ken J. Adelaide Bot. Gard. 18(2) (1998) I. The indumentum of branches with leaves showing juvenile characters tends to be similar to that of adult branches except that it is usually more sparse and often the individual stellate hairs have fewer branches. Juvenile leaves tend to be more obovate and smaller than adult leaves from fast growing branches (with long internodes), but they are generally comparable to those fi-om older branches. The indumentum of the adaxial leaf surface shows usually a range from simple hairs more or less antrosely inclined in juvenile leaves to usually some stellate hairs with few antrorse branches (often unequally long) usually mixed with simple hairs in the intermediate stages to usually more or less dense stellate hairs with about equal branches radially arranged in the adult stage. Simple hairs are conunonly found in, or restricted to a strip along the flanks and the depression above the central vein. The indumentum of the abaxial leaf surface (excluding the flanks and central vein) varies from usually only hooked simple hairs (in some local forms these hooked hairs are absent or almost so, in others the occasional scattered stellate hairs are found interspersed) to increasing numbers of stellate hairs with short similar branches in between hooked simple hairs in the intermediate stages to usually stellate-velutinous to -woolly with long equal branches of the stellate hairs often somewhat depressed and strongly overlapping in the adult stage. A few larger stellate hairs with pronounced tubercles are found in some species, but they are usually of the same soft texture. If flowers are present on material with juvenile or intermediate characters then they are usually leaf-opposed and rarely with subwhorled leaves below the peduncle (cf. inflorescence, affinities below), except in H. pallidiflora, where the different stages are often not clearly separated. In the latter species juvenile and intermediate characters occur, are maintained or recur for as yet unlcnown reasons (cf. Toelken 1995). Similarly, juvenile characters have often been observed on specimens of H. hirticalyx from Tasmania, but not one specimen from Victoria showed leaf characters other than adult. While in H pallidiflora at least some of the variation seem to be ecologically induced, there are indications that those of the two forms of H. hirticalyx are genetically fixed, but in both cases generalisations should be avoided. Considering the known variation of the shape and size of leaves as well as the indumentum, it is obvious that Bentham's concepts could not provide natural delimitations of the taxa. This study of the range of variation of the indumentum in each species is not only important for the delimitation and subsequent identification of these species, but also gives, when provided with a predictable range of characteristics, some indications on the affinities of the various taxa to one another (cf affinities below). These are, however, only generalisations as a basis for further discussion in a group plants that has been neglected for a long time. Indumentum (Vestiture) The hairs covering various organs of the plants are at times so varied in the species of sect. Pleurandra that a special descriptive system has been developed to maximise the use of these characters yet allow for observed variation. The hairs are here described under the broad entry of vestiture for the types and shapes of hairs, their variation, distribution and 110 J. Adelaide Bot. Gard. 18(2) (1998) Hibbertia asperagroup (Dilleniaceae) stratification. A separate indication of the denseness and texture, e.g. pilose, is provided only for young individual organs in order not to be misleading because it refers only to the visible simple hairs overtopping at first, but as they usually wear off they expose the velutinous cover of short stellate hairs on older leaves and branches of, for instance, H. rhynchocalyx. Westiture' is used here to describe the detail in contrast to `indumentum', which as used by Hewson (1988) also includes indications of texture and denseness, so that the two terms as used here are not interchangeable. This division of characters allowed a greater versatility to describe the large variation found in some species of Hibbertia without duplicating information for each organ. In a few cases some hair characters needed accentuating in connection with certain organs at the risk of an apparent inconsistancy in the overall treatment The absence of such information in other taxa should be interpreted as negative. Each character of the vestiture is, like any other taxonomic character, only described when its range of variation (usually including its juvenile range) on a specific organ of a certain taxon has been evaluated. As a rule the vestiture of a few key organs, viz, branches, upper and lower leaf surfaces and the calyx lobes were examined and described, but there are also references to specific features of other organs. More information and especially field observations is still needed as many of these genotypic and phenotypic variations, e.g. differences of vestiture between juvenile and adult leaves as in H pallidiflora, showed that these are probably affected by environmental factors causing, for instance, a prolonged retention of these juvenile characters, which in turn could result in misidentification of material. HAIR TYPES The basic distinction between simple and stellate hairs is sometimes not obvious, because when simple hairs form from adjoining epidermis cells the swollen basal tubercles often join so that the structure should be called a stellate hair. Similarly the branches of stellate hairs often do not seem to develop, or wear off easily and the apparent contradiction to stellate hairs with 1-3 branches is determined by comparison with the size, shape and deflection of the surrounding hairs. The tubercle is an arbitrary term often used in descriptions of hibbertias for a more or less swollen base of simple and stellate hairs. In addition to the tubercle, simple and stellate hairs on the upper leaf surface of species of the H. aspera group are usually surrounded by base cells, which are somewhat bulging epidermis cells with thickened walls. These hairs are somewhat reminiscent of those found in the Boraginaceae, but in the H. aspera group are never coarse. The base cells together with remains of the tubercles are the paler spots usually observed on older leaves where the hairs have largely worn off. Hooked simple hairs which are orracinnnily found on the upper surface of leaves of sonle foims are usually without base cells. The simple hairs are apparently always unicellular from an epidermal cell and often with a basal swelling or tubercle. They are straight or with a hooked apex, and often longer than stellate hairs, so that the frequently used terms `pilose' or 'hirsute' refer to the visible single hairs. While straight simple hairs are more or less antrorsely inclined, rarely curved, hooked simple hairs are usually erect or rarely retrorsely inclined particularly on the calyx. Stellate hairs in Hibbertia are formed by a clustering of simple unicellular hairs (fig.1A), the bases of which are more or less connate to form a longer or shorter tubercle. They are very variable and the alternatives are retained in the following sequence in all descriptions in order to remain consistently comparative in the wide range of combinations of characters as required (fig.1B): 111 H.R. Toelken J. Adelaide Bot. Gard. 18(2) (1998) persisting/ wearing off except basal tubercles, erect/ multiangulate/ de- pressed, radial/ antrorse/ retrorse stellate hairs (with number of branches) with small/ broad/ stalked tubercle base. A simple terminology was maintained in order to keep it self- explanatory. "Erect" and "de- pressed" do not replace Hewson's (1988) "porrect" (though the odd hair on plants of the H. aspera group may produce a straight erect central hair) and "rotate" or "peltate" (branches of hairs in this group are rarely in one plain) respectively, and some of Hew- son's terms may be used in subsequent papers in this series. Basically they are multiangulate stellate hairs and "erect" refers to hairs usually with only a few del: ansBors1298 branches all of them at 75 or more Fig. I. Hairs. A, H. cinerea: stellate hairs are composed of several from the base; "depressed" to hairs adjoining simple hairs x50; B, diagram of juvenile leaf of H. with all branches less than 35 to 40 aspera subsp. aspera showing above few simple hairs above from the base. The tenn "stalked multiangulate antrorse stellate hairs (few often unequal branches), stellate hairs" (Toelken, 1996) was and below few multiangulate radial stellate hairs (many, often similar branches) rarely overlapping with hooked simple hairs used to avoid confusing them with x20. (A, R. Alcock 625, AD; B, C. T. White BRI 10502). dendritic hairs, because their branches spread from more or less the same point and the length of the tubercle is relative to that of other stellate hairs on that specific organ. Considerable variation of stellate hairs occurs between specimens of the same species and between leaves of juvenile or adult plants. Stellate hairs usually have similar branches, but occasionally there are distinctly unequally long ones found especially when there are larger stellate hairs over smaller ones as in H. cinerea. These unequal branches can reach extremes in cases where one or two branches may extend to more than twice as long as the other branches, and may on superficial investigation be confused with single hairs over stellate ones. Although some sequences of the above terms might seem to indicate progression to more complex conditions, and this might be observed in particular examples, generalisations should be avoided as there is, for instance, evidence for increasing as well as decreasing number of branches of stellate hairs within the species of this group. In the above descriptions the following words are added in order to provide a simple image of the three dimensional stratification of the hair cover similar to that of vegetation (cf. fig. 1B) over/ similar/ under together with overlapping (free being usual) accepting that the distribution of the different types is a random scatter unless specific accumulations are mentioned. This simplification expressed useful descriptive information of the hair cover which could otherwise only be obtained by unjustifiably cumbersome measurements. 112 J. Adelaide Bot. Gard. 18(2) (1998) Hibbertia aspera group (Dilleniaceae) Leaves The shape and size of the leaves often varies considerably from young to older branches but can equally vary on the same plant predominantly due to the amount the margins are recurved, which depends on the environment or techniques of drying specimens. These are therefore generally not useful characters in this group of species. The description and illustration of juvenile and adult leaves and especially their indumentum represents extremes observed (cf. juvenile to adult developmental stages). The extent of the variation found seems to vary with environment, as particularly shading seem to have a retarding influence on the development of adult leaves, but the duration of expression of the juvenile leaves is also variable in different taxa (cf. subpecies of H. aspera), and may be repeated in some species, e.g. H pallidiflora, with every new axis developing. Discolourous leaves are not entirely due to their indumentiun, as leaves of H empetrifolia, which often has glabrescent older leaves (above and below), are usually paler below. Inflorescence and flowers The concept of an inflorescence has not been developed in Hibbertia although accumulations of flowers have been recorded in some groups of species in the genus. Flowers in Hibbertia are commonly single, terminal and often become leaf-opposed by sympodial growth (e.g. H. decumbens, fig.6A). Flowers maydevelop repeatedly at intervals of three or four intemodes (e.g. H hirta). Two or three leaves below the peduncle of most of the species of this group are not separated by intemodes so that they are here referred to as subwhorled. Sympodial growth from their axils often results in di- or trichotomous branching, and only in the latter case can one observe the terminal position (leaf-opposed) of the flower. If, however, one of the three axillary buds has not developed into a branch then the flower will appear to be axillary to that subwhorled leaf. Some forms particularly of H aspera and H. empetrifolia produce flowers on short shoots without fully developed leaves, sometimes in the axils of leaves along the main branches. These and normal flowering branches then produce one to three fascicled axillary branches (sympodial growth) with vestigial leaves (cf. fig. 4A), which often drop off early, so that the cluster of flowers thus formed appears to be a few-flowered inflorescence especially as the peduncle of each flower often remains intact for more than a year. They are, however not referred to as inflorescences because, unlike a raceme or cyme, or even the complex inflorescences described for some of the species of Dillenia (Comer 1978), there are several nodes of vegetative growth, however short, between subsequent flowers. The same applies to the dense clusters of flowers with special bract-like leaves in the H sericea complex (Toellcen, in preparation). Troll's definition of an inflorescence as "the shoot system which serves for the formation of flowers and which is modified accordingly" (Weberling 1987) could cover some forms of this species complex, but the concept of a single terminal flower is here preferred so that one can show the different developments within the genus more clearly. Since the single terminal flower was accepted as the basic unit, the characteristic elongated intemode between the subfloral bract and first leaf is here called "peduncle"; the intemode between the bract and the flower is then referred to as "anthopodium" (Briggs & Johnson 1979), in preference to "pedicel", which has been used for so many different conditions in literature (Conn, 1995). The characteristic subulate bracts always subtend the calyx in the H aspera group. They are without recurved margins and their length relative to the calyx was found a useful 113 H.R. Toelken J. Adelaide Bot. Gard. 18(2) (1998) character to distinguish some of the species while in others, e.g. H. empetrifolia, it is variable. The buds of most species are broadly ellipsoid but in the case of H empetrifolia they are narrowly ellipsoid and the base of an open flower is not abruptly constricted but more or less stepped. This is due to a reduction of the number of ovules from four laterally attached to the placenta along the suture of the carpel to two, more or less basal ovules in the latter. In the process the position of the ovules changes from parietal (horizontal), or almost so in the case of four ovules, to basal (erect) when two ovules are produced. Calyx and corolla In contrast to the petals the calyx lobes are usually basally connate. In the H. aspera group the outer three are commonly different from the inner two in shape, indumentum and the presence of a more or less pronounced central ridge, while in many other groups it is the outer two and inner three calyx lobes that are similar. The petals vary greatly in size, the shape of the claws, and their apices can vary from truncate, emarginate to deeply lobed, sometimes even within a population of a species. The shape of the base of the petals can, in the H. aspera group, be distinctive as it is cuneate (cf. fig. 7A) in the species with fewer than eight stamens and tends to be more rounded to obovate in species with ten or more stamens. It remains a character of limited use on herbarium material. Stamens The species of the H. aspera group can be divided into those with (9) 10-12 (-15) stamens and 4-8 (9) in H. aspera and H empetrifolia. The anthers of the group with more stamens are oblong (fig. 2D) and (1-) 1.2-1.5 (-1.8) mm long while those of the latter are usually oblong-oblanceolate (fig. 2E) and (0.6-) 0.7-0.9 (-1.1) mm long in herbarium specimens. The length of the anthers in the group with many stamens does not vary as much as in the others where one or more of the central ones are usually enlarged as part of the pollination syndrome. In some species the anthers, again especially the central ones, have short pointed appendages from the connective (fig. 2D). The filaments are usually connate for more than half their length in H hirticalyx. Even in this species there is some variation but in others there is even more. As in most Hibbertia species with stamens in one bundle (sect. Pleurandra) their filaments are variously basally connate but in local forms the amount of fiision of all or bundles of them varies, and is constant in only a few species. In the H. aspera group it was not found to be a reliable rharacter_ The solid filaments and the closely packed anthers are more or less curved forward (fig. 2A,B) to cover the styles so that just the stigmas protrude horizontally. This suggests a specialised pollination syndrome unique to species of the H aspera group except for H pallidiflora which has tubular petals (Toelken 1995). Flowers in this group are therefore more zygomorphic than in other species of sect. Pleurandra. Fruit The shape of the fruit also varies according to whether two or four seeds develop. As the ovary enlarges the original villous indumentum becomes sparser so that one can distinguish the normally stellate hairs but in H appressa they are simple. There are usually two to four ovules (rarely up to six in H. truncata) in each carpel and the fewer there are the more basal the placentation usually becomes. Rarely more than two 114 J. Adelaide Bot. Gard. 18(2) (1998) Hibbertia aspera group (Dilleniaceae) seeds per carpel mature and the position of the attaclunent is also often displaced at maturity, so that little importance was attached to these characters. del: acxsorgt Fig. 2. Flowers and fruit. A, B, H. truncata: A, front view of rotate flower with stamens curved forward as commonly found in the H. aspera group x3; B, side view of androecium curved over the gynoecium to place apex of anthers above stigmas x3. C. H. pallidiflora: flowering and fruiting branch showing unique pendulous flowers with tubular petals (free petals arranged in a cylinder) shorter than calyx (one lobe removed) x1.5. D, H. rhynchocalyx: oblong anther with appendage terminal to connective x30. E, H. empettifolia subsp. radians: oblanceolate anther with flared apex x30. FI, H. aspera subsp. aspera: the aril on seeds of the same plant varies in size and may develop broad lobes to finger-like processes x6. (A, B, H.R. Toelken 9251, AD; C, H.R. Toelken 8567, AD; D, iB. Williams 603, NSW; E, H.R Toelken 8681, AD; FI, Price NSW 10293). The seeds vary considerably in size and shape depending on the number developing in the same carpel as well as on environmental conditions as in many cases sudden heat waves etc. seem to produce premature ripening of the seeds. While most seeds are shiny black they become brown to light brown under those conditions. The aril in this group of species develops from an upper fleshy collar of the funicle a membranous sheath of varying size and lobing (fig. 2FI) even within the species. When four ovules develop the sheath usually does not develop well in the contact zone. The aril also seems to be smaller in prematurely ripened seeds, so that this variable character cannot be used to distinguish species within the H. aspera group but the well developed membranous sheath will distinguish members of this group from others in the genus. Affinities The species have for convenience been alphabetically arranged but the limited information available allows some speculation on the affinities of the different taxa reenoTliqed 115