Bremekamp's systematic views

Bremekamp published various studies of Rubiaceae classification during a period of significant controversy over its familial and generic classification. Before molecular systematic data and techniques became available, classifications were evaluated based on their morphological documentation and their conceptual basis. Bremekamp's studies were detailed and influential (e.g., Bremekamp, 1966), especially with respect to generic delimitations, and his work was accepted by a number of workers, especially in Europe. His views were also sometimes controversial (e.g., Verdcourt, 1958), and his disagreements with other taxonomists cast doubt on other classifications. Modern molecular studies do not support most his groups, as outlined below for his classification of Psychotria, but study of his work gives us useful insights into morphological patterns in these plants, and historical views that are incorporated in today's Rubiaceae taxonomy.

Fig. 1.

Psychotria L. in Madagascar. A. Inflorescence of P. lucidula Baker; B. Short-styled flower form of P. polyphylla Bremek.

[A: Antilahimena 7943; B: Antilahimena 8848] [Photos: P. Antilahimena]

For Madagascar, Bremekamp (1958, 1963) recognized 44 species of Psychotria, 12 species of Apomuria, 8 species of Cremocarpon, 67 species of Mapouria, 2 species of Pyragra, 4 species of Psathura, and 6 species of Trigonopyren. He diagnosed these genera variously by fruit dehiscence, number of ovary locules and pyrenes, form of the pyrenes, and the presence and pattern of ruminations in the endosperm of the seed. Mapouria was a key genus in his classification, and was circumscribed based on his individualistic concept of Psychotria. Bremekamp also separated several species groups within both Psychotria and Mapouria, based on characters including leaf size, presence vs. absence in the leaves of bacterial nodules and acarodomatia, inflorescence arrangement and position, flower color, fruit size, pyrene form and details, and presence and pattern of endosperm rumination. Bremekamp's genera and species groups are useful today for identification, especially when both flowers and fruits of a plant are known.

Bremekamp's generic concepts of Psychotria and Mapouria and his taxonomic view that this overall group comprised a number of small genera were developed in detail in his study of the Madagascar flora (Bremekamp, 1958, 1963), but his basic taxonomic views were developed in his earlier study of the Surinam flora (Bremekamp, 1934). In the Surinam flora, he studied some characters not previously considered diagnostic in a flora where Psychotria was unusually complicated. In Surinam Bremekamp looked at relatively few species of several large genera, including Psychotria, and the species of those genera were notably different from each other. Here he separated groups of one to five similar species and treated them as distinct genera. Later authors working on the regional flora that includes Surinam, however, have found it difficult to clearly delineate morphological groups within Psychotria and the related Palicourea Aubl., or to place additional species into Bremekamp's generic classification (Steyermark, 1972; Taylor & Hollowell, 2016). In Madagascar, Bremekamp used a genus concept similar to that of his Surinam flora: here he again separated from Psychotria and Mapouria several groups of related species that are similar to each other and share an unusual or distinctive feature.

Molecular data do not support Bremekamp's genera

Nepokroeff et al. (1999) presented the first molecular phylogenetic analysis of world-wide Psychotria and related genera, and found the systematics of this group to be complicated and the existing classifications partially inaccurate in several ways. Their molecular data showed that Psychotria comprised several regional clades that were apparently distinct, and that some smaller genera were nested in them. No Malagasy species were included in their analysis.

Andersson (2002) then used molecular systematic data to analyze the group identified by Nepokroeff et al. (1999) as Psychotria and its closest relatives in more detail, and found a clade he characterized as Psychotria s.str. and noted that several other supposedly distinct genera were nested within it. He also surveyed morphological characters, and identified several features to diagnose Psychotria s.str. The features he used were based in large part on the work of Petit (1964, 1966). Andersson (2002) diagnosed Psychotria s.str. by pollen with simple apertures plus pyrenes that are “usually” plane or shallowly furrowed adaxially, longitudinally ridged abaxially, and lack PGS's. Separately (Andersson, 2001) he also used the presence of an alcohol-soluble pigment in the testa. The genera recognized by Bremekamp that he found to be nested within Psychotria s.str. were Mapouria, which was represented in his analysis by its Neotropical type species and two Malagasy species; Apomuria, represented by its African type species; and Cremocarpon, represented by two samples, neither of them the type: a Malagasy specimen at K that was identified only to genus, and C. rupicola (Schltr.) Bremek. from New Caledonia. The K specimen has not been seen in this study, but it is identified in that institution's online herbarium catalogue as C. lantzii Bremek.

Razafimandimbison et al. (2014) presented a broader molecular systematic analysis of Psychotria s.str. with a focus on the species from the Indian Ocean region, and included representatives of all of Bremekamp's (1963) segregate Psychotria genera from Madagascar. They found his segregate genera nested within several large, supported clades in Psychotria, and some to be polyphyletic. Razafimandimbison et al. (2014) adopted a broad view of Psychotria, and synonymized all six of Bremekamp's Malagasy segregate genera within it. They also found no support for any of Bremekamp's informal species groups in Mapouria and Psychotria, though their sampling was not adequate to test the monophyly of all of those. They analyzed only two morphological characters, bacterial nodules in the leaves and schizocarpous fruits, and found those both to be homoplasious; Bremekamp only considered the latter a diagnostic feature at the generic level. Posterior mapping of species characteristics on their molecular phylogram also shows that nearly all of the morphological characters Bremekamp used to separate genera and species groups are homoplasious in the overall group Psychotria. Most of the supported clades that Razafimandimbison et al. (2014) found within Psychotria were not correlated closely with biogeography, and none are (at present) diagnosable morphologically. Their sampling was too limited to provide a confident new systematic framework for this genus, either overall or in the western part of the Indian Ocean, as noted in their discussion.

Molecular systematic results are compared in more detail below for each of Bremekamp's genera in his treatment of Malagasy Psychotria.

Bremekamp's Malagasy genera and their modern identities

Below I review the morphological characteristics and current status of Bremekamp's (1958, 1963) segregate genera of Psychotria. The order of discussion starts with Psychotria and Mapouria, to help connect his classification to modern genus concepts and nomenclature, and then Bremekamp's order of presentation is followed for the remaining genera (the reader may prefer to read these in reverse order, because the last genera here are better delimited than the first and give better insights into Bremekamp's genus concept, while his Psychotria comprised mainly species that lack the apomorphic features of the other genera).

Psychotria

Morphology and identity of Bremekamp's Malagasy Psychotria

Bremekamp (1934, 1961, 1963) circumscribed Psychotria based primarily on pyrene characters, along with some other features he did not explicitly list. He diagnosed Psychotria (Bremekamp, 1934, 1963) by its pyrenes that adaxially have a flat face with two shallow longitudinal grooves that lie on each side of a medial ridge (Bremekamp described this diagnostic pyrene form as a shallow longitudinal excavation divided in two by a medial ridge, but as noted by Piesschaert (2001: 327–328), most taxonomists see the alternative condition, a plane face with two longitudinal furrows). Bremekamp's Psychotria (Bremekamp, 1934, 1961, 1963) was also diagnosed, in practice, by a lack of the features that diagnosed his other genera, i.e., consistently deciduous stipules (Mapouria), pseudoaxillary or axillary inflorescences (Notopleura (Benth.) Bremek., Ronabea Aubl.), capitate inflorescences with each flower involucrate (Cephaelis Sw.), multilocular ovaries (Psathura, Nonatelia Aubl.), dehiscent fruits (Cremocarpon, Pyragra), and pyrenes that are flat on the adaxial face (Mapouria, Gamotopea Bremek.). Bremekamp's diagnosis of Psychotria is problematic, however, because his diagnostic pyrene characters have been shown to vary without any apparent systematic pattern and there is no other feature that diagnoses his group by its presence (i.e., no apomorphy).

Petit (1964) and Piesschaert (2001) surveyed Psychotria pyrenes in Africa and world-wide, respectively, and found continuous variation in the form of the pyrenes' adaxial walls, and no correlation of pyrene form with any other characters or systematic groupings. Additionally, Bremekamp's characterization here is partially inaccurate, because the form of the adaxial pyrene wall varies within some species and even in different fruits of a single plant. Bremekamp's diagnostic feature for Psychotria is, thus, not diagnostic, and the species grouped in Psychotria by Bremekamp (1963) are heterogenous morphologically and difficult to characterize as a group. The only features that seem to be consistent within Bremekamp's Malagasy Psychotria are not unique to it: the mature fruits are blue or white, and Piesschaert (2001) found all the species included here have pyrenes that lack PGS's and have an alchohol-soluble red pigment. The characterization of a genus by its lack of the diagnostic features of other groups results in a classification that with genera variously diagnosed by shared derived features (synapomorphies) and genera that are residual, paraphyletic assemblages that have the ancestral features of the group. This was a common approach to classification in Bremekamp's time, but it is not in accordance with most current plant taxonomy. Modern systematic approaches to classification emphasize the consistent separation of groups that comprise entire lineages and have shared derived features (Razafimandimbison et al., 2014). Bremekamp (1963) recognized seven informal species groups in Psychotria based on leaf size and form, stipule form, drying color of the specimens, inflorescence form, and the presence of bacterial nodules. He designated these groups informally with numbers, and called each a “Group” in his key but a “Taxon” in his species treatment.

Fig. 2.

Psychotria L. in Madagascar. A. Long-styled flower of P. macrochlamys (Baker) A.P. Davis & Govaerts; B. Fruits of P. macrochlamys; C. Habit of P. lantzii (Bremek.) Razafim. & B. Bremer.

[A: Antilahimena 6104; B: Razanatsima et al. 1660; C: Ratovoson 1707] [Photos: A: P. Antilahimena; B: A. Razanatsima; C: F. Ratovoson]

Bremekamp's Malagasy Psychotria analyzed with molecular data Andersson (2002) did not study any Malagasy or Comoran species classified by Bremekamp (1963) in Psychotria. Razafimandimbison et al. (2014)'s molecular analysis included 13 species from this region that Bremekamp (1963) classified in Psychotria, and found these placed on two clades. The species here without bacterial nodules were grouped with species of Bremekamp's Mapouria, Apomuria, and Psathura.The species with bacterial nodules were grouped with other species of Apomuria and some nodulated African species; here the Malagasy species were grouped on one subclade, and separated from the Comoran species. Razafimandimbison et al. (2014) concluded that Psychotria as circumscribed by Bremekamp in Africa and the western part of the Indian Ocean was paraphyletic without the inclusion of Apomuria, Cremocarpon, Mapouria, Psathura, Pyragra, and Trigonopyren.The extensive morphological heterogeneity of Bremekamp's Psychotria is mirrored by the extensive polyphyly they found in this group. In the results of Razafimandimbison et al. (2014), all of Bremekamp's Malagasy Psychotria species groups are paraphyletic.

Mapouria

The name Mapouria as used by Bremekamp

Separation of Bremekamp's segregate genera from Psychotria required him to determine appropriate names for them, and their names depended on the identity of Psychotria. Bremekamp's analysis of Psychotria's identity was idiosyncratic, and as result so was the name he used for his largest segregate genus. Psychotria was described with one species, P. asiatica L., that was based on one paleotropical element from Asia and one neotropical element from the Antilles (Bremekamp, 1961; Petit, 1964; Davis et al., 2001). Because there were two elements, Bremekamp (1961: 317) concluded that Psychotria was a “nomen confusum” whose identity could not be determined. Bremekamp diagnosed his newly separated genera in part by stipule persistence, and he concluded that Psychotria as originally described probably had persistent stipules. He did not see any original material that corresponded to P. asiatica, but based his conclusion only on later use of that name by some other authors. Based on this, he then decided that P. asiatica did not agree with the genus he diagnosed by having deciduous stipules, so another name was needed for that newly separated genus. He then concluded that Mapouria was the oldest name in this group that demonstrably had deciduous stipules, and that Grumilea shared this feature and was not distinct from Mapouria. Bremekamp (1961) then used the name Mapouria for his newly separated Malagasy genus, and noted that using this name for this group was due to a possible major change in nomenclatural usage of the name Psychotria. He explained that if the type of Psychotria was eventually studied and confirmed to have persistent stipules, most of the species across the world with names in Psychotria should instead be called Mapouria. Bremekamp then recommended that in this case, for nomenclatural convenience, a different type should be chosen for Psychotria, and it should agree with his Mapouria so these numerous name changes would not be needed (he did not comment on how to resolve the nomenclature of the species with persistent stipules that he placed in Psychotria). Bremekamp's views here were contrary to our current nomenclatural practice (Turland et al., 2018), so his conclusions conflict with today's nomenclature and have created lasting confusion. Current nomenclatural practice prefers to find the identity of a name by direct study of original material, to confirm this identity before the name is used or a substitute name adopted, and to propose formal conservation of names that are problematic instead of just mentioning a possible problem.

Fig. 3.

Psychotria L. in Madagascar. A. Leaves with bacterial nodules of P. pachygrammata Bremek.; B. Leaves with retuse apices of P. biloba (Bremek.) Razafim. & B. Bremer; C. Unusually large green fruits of P. tolongoinensis A.P. Davis & Govaerts; D. Dense pubescence of P. rufovillosa (Bremek.) A.P. Davis & Govaerts; E. Blue fruits of P. glaucifolia A.P. Davis & Govaerts; F. White fruits of P. decaryi Bremek.

[A: Syde et al. 291; B: Bolliger 123; C: Razafindrahaja 242; D: Antilahimena 8184; E: Callmander 393; F: Rasoanindriana et al. 148] [Photos: A, C, D, F: P. Antilahimena; B: R. Bolliger; E: M. Callmander]

The identity of Psychotria was also reviewed contemporaneously but separately by Petit (1964), who did study the original material of Psychotria. He concluded that its two type elements are morphologically consistent and the Asian material should be considered the type of P. asiatica. He confirmed that both of the original elements have deciduous stipules, so Bremekamp's interpretation of this name was inaccurate. Davis et al. (2001) then reviewed the identity of P. asiatica in more detail, and formally accepted Petit (1964)'s conclusion as a lectotypification (the problem that Bremekamp noted with the characteristics of Psychotria has since been resolved by separating many of the heterogeneous Psychotria species into other genera and even tribes, e.g., Andersson, 2001; Razafimandimbison et al., 2014).

Apomuria

Bremekamp's Apomuria analyzed with molecular data

Andersson (2002) showed that the T-shaped endosperm intrusion is highly homoplasious in world-wide Psychotria, and concluded that Apomuria is not separable morphologically based on this diagnostic feature and formally synonymized it with Psychotria. Razafimandimbison et al. (2014) studied two species of Apomuria, both of which are nodulated species and neither of which is the type. They found these species separated from each other and grouped with species of Psychotria, and agreed with its synonymization.

Fig. 4.

Some morphological features of Psychotria L. in Madagascar. A–B. Apomuria, Psychotria biloba (Bremek.) A.P. Davis & Govaerts: A. Fruiting branch; B. Cross-section of seed (removed from pyrene). C–D. Apomuria, Psychotria punctata Vatke: C. Fruit; D. Cross-section of seed. E. Psychotria kirkii Hiern, cross-section of seed. F–G. Psathura, Psychotria borbonica (J.F. Gmel.) Razafim. & B. Bremer: F. Fruit; G. Cross-section of fruit, with four pyrenes. H–I. Psathura, Psychotria batopedina (Verdc.) Razafim. & B. Bremer: H. Fruit; I. Cross-section of fruit, with five pyrenes. J. Unusual stipules in Malagasy Psychotria L. (TAYLOR, in press). K–N. Trigonopyren, Psychotria tsiandroi Razafim. & B. Bremer.: K. Node near stem apex with developed stipule, unbroken and with glandular projections; L. Node below the node shown in K, with aging stipule that is falling off in fragments; M. Node in lower part of stem, with old stipule that has fallen off except for persistent basal part of sheath; N. Cross-section of pyrene. O–P. Trigonopyren, Psychotria bealanensis Razafim. & B. Bremer: O. Fruit; P. Cross-section of fruit, with two pyrenes.

[A: Du Puy et al. 757; B: Villiers 4881; C–D: Barthelet et al. 1474; E: Kuchar 23683; F–G: Blackburn 3200; H–I: Andriananrista 126; J: Antilahimena 2657; K–N: Bollinger et al. 240; O–P: Antilahimena 403]

Trigonopyren

Psathura

Cremocarpon and Pyragra

Habit and drying color

Psychotria species in Madagascar are mostly evergreen, erect, regularly branched shrubs and small trees (Fig. 2C), which is the common growth habit in this genus. A few Malagasy species appear to be deciduous. The “trash bucket” habit, with an unbranched main stem bearing well developed subsesssile leaves that accumulate detritis at their bases, is found in a few Malagasy Psychotria species from wet forest (e.g., P. simianensis (Bremek.) A.P. Davis & Govaerts). One Malagasy species is unusual (though not unique) in the genus in its low habit with prostrate main stems that produce adventitious roots (Taylor, in press).

Psychotria is generally distinctive in its reddish brown, dark brown, dark gray, or grayish green drying color, and herbarium specimens are frequently recognizable to genus by their color (Hamilton, 1989; Taylor, 2012). A number of Malagasy species dry with a clear green color that is uncommon in Psychotria elsewhere. Bremekamp (1963) frequently noted drying color in his treatment, and this detail is often useful for identification but this may vary with drying method and perhaps ecological or plant chemistry factors.

Leaves and stipules

Several species of Psychotria from Madagascar have unusual, deeply retuse leaf blades (e.g., P. biloba (Bremek.) Razafim. & B. Bremer, P. retusa (Bremek.) A.P. Davis & Govaerts; Fig. 3B, 4A) that are apparently not found in this genus elsewhere. In the upper portions of these leaves, the main part of the blade extends apically in two rounded lobes and the costa is shorter than these lobes. The secondary veins in the upper portion of the retuse blades are generally markedly curved and sometimes are closely set on the short costa. These leaf blades generally have obtuse to truncate tops when young, then as the leaf grows the upper part of the blade enlarges more than the costa. Species with this leaf form are found in both humid and seasonal vegetation. Only one species with retuse leaves was included in the analysis of Razafimandimbison et al. (2014), so whether this feature has arisen more than once remains to be tested.

Bremekamp (1960, 1963) noted the presence of bacterial nodules (i.e., bacteria leaf-galls; Robbrecht, 1988: 51–52) in some Psychotria species in Madagascar and the Comores (Fig. 3A), and this is an unusual feature in Rubiaceae that is found only in the African genus Sericanthe Robbr., some African species of Pavetta L., and some African and western Indian Ocean species of Psychotria. The bacterial leaf nodules in the Malagasy species of Psychotria are presumably similar in form, development, and function to those in African Psychotria (e.g., Pinto-Carbo et al., 2016). The bacteria in these nodules are nitrogen fixers and obligate symbionts of the genus Burkholderia Yabuuchi et al. Razafimandimbison et al. (2014) studied the systematic distribution of bacterial leaf nodules in Psychotria, and found at least two separate origins of this feature with both represented in Madagascar. Bremekamp documented variation in the form and position of the nodules: reticulated vs. simple to sparsely branched, and scattered in the lamina vs. only next to the costa. Nodules near the costa are sometimes few and difficult to see, and as noted above, Bremekamp overlooked these in Apomuria bullata.

Bremekamp (1963) also noted that acarodomatia (i.e., domatia; Robbrecht, 1988: 49–50) are often found on the leaves in Psychotria in abaxial axils of the secondary veins at the costa. He used presence and form of the domatia as a taxonomic character, but these are not always consistent in Psychotria. The presence of domatia sometimes varies within individual Psychotria species (e.g., Hamilton, 1989). Bremekamp recognized one domatium form, the crypt-type domatia (i.e., including pit-type, pocket-type, and crypt-type acarodomatia of Robbrecht) that is found in at least half of his Malagasy Psychotria species, but noted that tufts of trichomes borne on the flat surface of the lamina in the vein axils in other species might also be considered domatia (i.e., tuft-domatia of Robbrecht). The domatium forms separated by Robbrecht are not always distinct in Psychotria, so Bremekamp's keys can sometimes be problematic to use: crypt-type domatia may be densely pubescent inside, and axillary tufts of trichomes may be surrounded by low to developed ridges of tissue that could be considered a weakly formed (or reduced) crypt-type domatium or could be just a drying artifact.

Fig. 5.

Some morphological features of Psychotria in Madagascar. A–D. Cremocarpon, Psychotria boiviniana (Baill.) Razafim. & B. Bremer: A. Dehiscing fruit; B. Cross-section of seed (with pyrene wall removed); C. Dehiscing fruit; D. Fruit. E–H. Cremocarpon, Psychotria lantzii (Bremek.) Razafim. & B. Bremer: E. Fruit; F. Fruit partially dehisced, with carpophore and one mericap (behind carpophore); G. Carpophore persisting on pedicel; H. Cross-section of pyrene. I–M. Cremocarpon, Psychotria trichantha Baker: I. Fruit; J. Mature fruit smashed by pressing, internal view with two pyrenes; K. Mature fruit smashed by pressing, external view; L. Fruiting pedicel with fruit fallen and persistent fibers; M. Cross-section of seed (removed from pyrene wall). N–Q. Pyragra, Psychotria ankarensis (Bremek.) Razafim. & B. Bremer: N. Fruit just starting to dehisce; O. Fruit partially dehisced, with carpophore and one mericap (behind carpophore); P. Carporphore persisting on pedicel; Q. Cross-section of pyrene with seed removed.

[A, B: Barthelet & Mchanga 1543; C, D: Bremekamp (1958: fig. 20); E: Rabehevitra 4421; F, G: Rabehevitra 4386; I, L: Antilahimena 4478; J, K, M: Antilahimena & Edmond 3397; N, O, Q: Cheek 1436; P: Rakotonandrasana 1011]

Stipules vary widely in form and size in Psychotria, and often are useful taxonomic characters to separate species (e.g., Hamilton, 1989; Taylor, 2002). A wide range of stipule form is found in Psychotria in Madagascar, with some forms that are not known elsewhere so their description here expands our knowledge of morphological radiation in Psychotria. The stipules of Psychotria are most often deciduous and present only on the stem apex. Most of the species in Madagascar have interpetiolar stipules that are triangular or shortly bilobed. In some species, however, the stipules are partly fused into a well developed tube (i.e., sheathing), or fully fused into a conical cap (i.e., calyptrate; Taylor, 2002). Some Malagasy Psychotria species have an unusual form not reported from elsewhere: the stipules are persistent, well developed, and fused around the stem into a cupuliform tube (e.g., P. onivensis (Bremek.) A.P. Davis & Govaerts; Fig. 4J; Taylor, in press).

Inflorescences, flowers, and fruits

Both world-wide and in Madagascar, the inflorescences in Psychotria have generally green to whitened axes and bracts but vary notably in arrangement, position, and number of flowers. One aspect of inflorescence arrangement has been interpreted differently by different authors, which has created conflicts in species taxonomy and problematic keys for Psychotria: the interpretation of “pedunculate”. In Madagascar, Bremekamp (1963) made a careful distinction for the branched inflorescences in Psychotria as either pedunculate or sessile, and used this as a taxonomic character. In his treatment, a pedunculate inflorescence has one peduncle that supports all the axes and flowers, has small bracts at its top, and is subtended by well developed leaves, while a sessile inflorescence has three or more fasciculate peduncles, with these borne on a structure that has large bracts (or small leaves) at its top and is subtended by well developed leaves. Several authors have considered these two distinct inflorescence forms and distinguished Psychotria species based on this (Bremekamp, 1963; Sohmer & Davis, 2007), but most authors have not. These other authors have noted that the difference between these two inflorescence arrangements is only the size of the basal most inflorescence bracts, which is not a difference of actual arrangement, it is an interpretation of a variable character. These other authors have noted that variation in size of these bracts is continuous in some species, and considered Bremekamp's “pedunculate” and “sessile” to be variants of the same basic inflorescence arrangement and not species differences (e.g., Standley, 1938; Steyermark, 1972; Hamilton, 1989; Taylor, 2012).

Several Malagasy Psychotria species have an unusual inflorescence arrangement apparently not found elsewhere: apparently axillary inflorescences that are borne in both axils of stem nodes well below the stem apex. Bremekamp (1963) called these pseudoaxillary inflorescences, and separated species with these inflorescences in his Mapouria Group I. In general this is a good taxonomic character, but his interpretation differs from more common usage and there seems to be some variation in the development of this feature. The inflorescences of the Malagasy species do not agree morphologically with Robbrecht's (1988: 68) definition of pseudoaxillary inflorescences, which are found in various species of Psychotria. Pseudoaxillary inflorescences sensu Robbrecht are developmentally terminal on stems with sympodial growth, so after the inflorescence is formed, then the stem continues to grow from only one of the subtending axillary buds, and it produces another terminal inflorescence. This growth form produces inflorescences regularly situated in only one axil of each stem node. In contrast, the inflorescences of Bremekamp's Mapouria Group I are regularly produced in both axils of the the stem nodes and not at the stem apex. Bremekamp interpreted these inflorescences as borne terminally on axillary brachyblasts, or short lateral stems, produced from axillary buds (e.g., P. andapae A.P. Davis & Govaerts, P. paradoxalis (Bremek.) A.P. Davis & Govaerts). This agrees with Robbrecht's view that axillary inflorescences in Rubiaceae (i.e., those borne from both axils of a node and not at the stem apex) can be considered morphologically to be terminal inflorescences that are borne on contracted brachyblasts. Razafimandimbison et al. (2014) studied very few species of Mapouria Group I, so whether this inflorescence arrangement has arisen more than once remains to be tested.

The flowers of Psychotria generally have five calyx lobes, five corolla lobes, and five anthers, a condition usually called “five-merous” in Rubiaceae even though the ovary is nearly always bilocular. Not infrequently, one or a few flowers on an individual inflorescence are four-merous or six-merous in Psychotria and related tribes, but such species are generally characterized by their most common condition even when some variation is noted (e.g., Bremekamp, 1963; Steyermark, 1972; Taylor, 2012). A few Psychotria species in all regions have consistently four-merous flowers. Bremekamp used flower merosity to distinguish some species, but it is a problematic taxonomic character.

Flowers of Psychotria are generally distylous with two different flower forms, an arrangement sometimes also called heterostylous or heterodistylous (Robbrecht, 1988: 122–125). A distylous species has two floral forms, but individual plants of this species bear only one flower form. The two flower forms differ in the positions of the stigmas and anthers, which are separated spatially and are reciprocally positioned between the long-styled and short-styled forms (i.e., pin and thrum, respectively: Robbrecht, 1988; Hamilton, 1989). Confirmation of distyly requires finding both flower forms for the species. Most if not all of the Psychotria species in Madagascar are distylous, with the stigmas exserted on a well developed style and anthers included on short filaments in the long-styled form (Fig. 2A), and the stigmas included on a short style and anthers exserted on well developed filaments in the short-styled form (Fig. 1B; e.g., Robbrecht, 1988: 125, fig. 47; Hamilton, 1989: 74, fig. 8). Bremekamp used the presence vs. absence of distyly to distinguish some Psychotria species, but at least some species he considered monomorphic (i.e., with only one flower form) are now documented to be distylous by newer collections.

Terminology for the Rubiaceae calyx and ovary has varied widely among authors, based on differing morphological interpretations, resulting in some confusion. The flowers of Psychotria have an inferior ovary that is covered by layers of tissue that represent the fused calyx corolla, and androecium, or collectively a hypanthium. This ovary portion of the flower has sometimes been described separately from the calyx, but other times has variously been considered part of the calyx, a hypanthium, or part of the same structure as the tubular free part of the calyx. The free part of the calyx has also been described in various ways: some authors regarded it as separate from the ovary and called it variously the calyx (e.g., Standley, 1938; Robbrecht, 1988: 74), the calyx tube (Bremekamp, 1963), or the calyx limb (Taylor, 2012), while other authors considered the ovary and the free portion of the calyx as on one structure (e.g., Verdcourt, 1976; Sohmer & Davis, 2007). And, when the free part of the calyx limb ranges is lobed, some authors regarded this as one lobed structure but others (e.g., Bremekamp, 1963) described the lobes as separate from the tubular, unlobed, basal portion of the calyx limb. These various morphological interpretations are not inaccurate, but the exact calyx structure being described is not always specified in taxonomic treatments. This can generate confusion when different authors give different size measurements for apparently the same structure. Combining the measurements of the ovary and free calyx limb is particularly problematic, because the ovary changes markedly in size as the flower matures so the size of this combined structure can vary with developmental stage more than species identity.

Corolla color is generally white, cream, or pale yellow in Psychotria of all geographic regions (Fig. 1A, 3B), but some Malagasy species are unusual in their bright yellow to orange flowers (Fig. 1B, 2A). Bremekamp characterized several of his species groups of Mapouria by this feature. Flower color was not studied in the molecular analysis of Razafimandimbison et al. (2014), and cannot be posteriorly mapped on their phylogram due the many samples that are not identified to species.

Fruits, pyrenes, and seeds

The fruits of Psychotria are fleshy and indehiscent, and comprise the tissues of the inferior ovary and hypanthium and contain two to several pyrenes (Sohmer & Davis, 2007: 7, fig. 1). These Rubiaceae fruits have long been considered a kind of drupe, but they do not match the classic drupe (e.g., Prunus L., Rosaceae) so many authors call them “drupaceous”. As discussed below in the section on ecology, Psychotria fruits in Madagascar vary in size and color, and as noted above, some are schizocarpous.

Pyrene and endosperm characters have been used to diagnose the genus Psychotria, species groups within this genus, and other Rubiaceae genera by a number of authors (e.g., Müller, 1881; Bremekamp, 1934, 1963; Robbrecht, 1989; Piesschaert, 2001; Andersson, 2001, 2002), but these characters are variable and sometimes problematic taxonomically. Psychotria fruits contain two hemispherical or plano-convex pyrenes (Fig. 4O; Sohmer & Davis, 2007: 7, fig. 1), each with one seed, except a few Malagasy species have three to five ovary locules and pyrenes and their pyrenes are triangular in cross-section (Fig. 4G, I). The pyrene wall is an adherent layer of endocarp, and varies in texture from papery to hard or bony. The pyrene wall often opens along weak spots, the PGS's. The presence and arrangement of these PGS's have been considered important systematic characters in Psychotria (^Petit, 1964; Robbrecht, 1989; Piesschaert, 2001; Andersson, 2002). Piesschaert (2001: 436) surveyed this feature in detail in Psychotrieae and Palicoureeae, and found “[t]he basic PGS-pattern is consistent within a genus, but additional PGS's may occur [within that genus]”. Psychotria has been characterized as lacking PGS's on the pyrenes (Andersson, 2002; Piesschaert, 2001), so these pyrenes finally split open irregularly from the base at the raphal plug (Piesschaert, 2001: 321). There is wide variation in the presence and form of PGS's in Psychotria, however (Piesschaert, 2001; Davis et al., 2001). PGS's have not been surveyed well for Malagasy Psychotria species. This character is not well enough known to map on the phylogram of Razafimandimison et al. (2014).

The endosperm of Psychotria is entire (i.e., solid throughout) in some species but in other species it has holes and invaginations, called ruminations, as detailed above. Endosperm rumination varies widely in Psychotria world-wide. It has been surveyed in some detail regionally by Petit (1964) and Sohmer & Davis (2007), and broadly worldwide by Piesschaert (2001). The presence and pattern of endosperm rumination vary widely within Malagasy Psychotria. These endosperm details were considered by Bremekamp (1963) to indicate relationships among species. In particular, he distinguished between endosperm that is ruminated only on the adaxial surface of the seed and endosperm that is ruminated on both adaxial and abaxial surfaces. However, in Malagasy Psychotria the pattern and degree of rumination sometimes varies within a species, some endosperm ruminations are small and may be overlooked if the seed is not studied carefully (e.g., P. aegialodes (Bremek.) A.P. Davis & Govaerts), and Bremekamp's distinction between the adaxial and abaxial surface is difficult to locate on seeds with obtuse to rounded junctions between these surfaces. Andersson (2002) studied this character in his analysis, and found it homoplasious. Posterior mapping on the results of Razafimandimbison et al. (2014) of presence vs. absence of any endosperm rumination and of rumination in the form of a T-shaped intrusion also suggests these features are homoplasious.

Fig. 6.

Distribution of Psychotria L. in Madagascar.

Each circle represents a specimen locality for a specimen of this genus. The map is based on c. 2700 specimen records that can be found in the Tropicos (2020) database.

Diversification and ecology

Psychotria is found in most habitats in Madagascar (Fig. 6), with its highest species diversity in evergreen humid vegetation. This genus is also well represented in seasonal habitats in western Madagascar, and is even found within the subarid southwestern area. As noted above, most species of Psychotria worldwide are evergreen but a few species in all geographic regions, including Madagascar, are at least facultatively deciduous (e.g., P. boenyana (Bremek.) A.P. Davis & Govaerts). This aspect is not well documented for most Malagasy species, but adaptation to dry habitats may be a factor in the diversification of Psychotria here.

Psychotria has a relatively large number of species in Madagascar for the area of the island, compared to both Psychotria in continental regions and other Rubiaceae genera here. Some other Rubiaceae also show notable endemic species radiations within Madagascar (e.g., Danais Comm. ex Vent.: Buchner & Puff, 1993; Puff & Buchner, 1994; Gaertnera Lam.: Malcomber & Taylor, 2009; Taylor et al., 2014; Pavetteae, De Block, 2018; De Block et al., 2018). Psychotria and Gaertnera have the largest documented species radiations radiations and are similar in habit, inflorescence form, flower biology (including distyly), size, and color, fruit size and form, and apparently generalist pollination and dispersal syndromes. Generalist modes presumably provide access to a relatively wide range of pollinators and dispersers, and reduce reproductive limitation due to these factors. Malagasy Psychotria species mostly have white, cream, or yellow corollas with tubes 2–6 mm long, which could agree with a generalist pollination mode.

Most of the largest genera in Rubiaceae are distylous (e.g., Palicourea Aubl., Palicoureeae; Ophiorrhiza L., Hedyotideae; though certainly not all, e.g., Ixora L.). Distyly promotes outcrossing within a population, which may be a factor in systematic diversification. Distyly is sometimes lost in peripheral populations (Sobrevila et al., 1983; Hamilton, 1990; Sakai & Wright, 2008; Consolaro et al., 2011), and in species with very limited ranges (Hamilton, 1990). Most Malagasy Psychotria species are confirmed to be distylous, which suggests they have robust populations.

Psychotria's fruits are dispersed by animals, which eat the fleshy portion and pass the pyrenes unopened but scarified through their digestive tract. The hard walls of the pyrenes protect the embryo inside, but also can prevent it from emerging. The pre-formed germination slits and enlarged raphal plugs found in many species are assumed to be release mechanisms for the embryo, but this does not actually seem to have been studied nor has any relationship been investigated between PGS development and form and type of fruit disperser. A few species in Africa and Madagascar have purple-black fruits borne on unusual fruiting pedicels that become elongated, swollen, fleshy, and brightly colored (e.g., P. rubropedicellata (Bremek.) A.P. Davis & Govaerts), presumably to attract dispersers and perhaps as an accessory part of the fruit that is also eaten. Similar fruits and swollen pedicels are also found in some Malagasy species of Chassalia Comm. ex Poir., which perhaps share the same dispersers.

The exocarp color of Psychotria fruits apparently signals to dispersers that a fruit is ripe, and may change from green to bright red within hours. Mature Psychotria fruits worldwide are generally 3–8 mm in diameter with a thinly leathery, red or orange exocarp and a juicy, colorless or whitened mesocarp, and are mostly dispersed by birds. Red and orange are the most common fruit colors in Madagascar (Fig. 2C, 3A), but black, blue, and/or white mature fruits are also found; some white fruits are an intermediate stage and later ripen to blue (Fig. 3E), but some species do have white mature fruits (Fig. 3F). Different fruit colors may attract different dispersers, and this may be a factor in diversification of this group here. A few fruits of Malagasy Psychotria are unusual in their relatively large size, 15–20 mm in diameter, fleshy texture, dull brown or perhaps green mature color (e.g., P. nossibensis A.P. Davis & Govaerts), and in some species an unusually large and lax infructescence. These are similar to fruits of other groups that are known to be eaten by lemurs. The Malagasy Psychotria species with relatively large fruits have inflorescences and corollas of average size, so these large fruits do not appear to result from a general size increase in reproductive structures of the species. As detailed above, a few species of Malagasy Psychotria have unusual dry, schizocarpous fruits, for which the dispersal mode is unknown.

Biogeography

The molecular phylogeny of Psychotria in Madagascar was analyzed by Razafimandimbison et al. (2014), with a broad sampling from across the Indian Ocean region. They found the Malagasy species all nested within world Psychotria, and three separate colonizations of the island. One colonization appeared to be from Africa, while the others could be from Asia, the Pacific, or Africa. Many of the Malagasy Psychotria species they studied were grouped with Psychotria species from Madagsacar, the Mascarenes, the Seychelles, and the Comores. All but one of the species with bacterial nodules were grouped on another lineage, with nodulated African species of Psychotria. Another, larger group of Malagasy Psychotria species were grouped with African and Comoran species on a clade that also had Asian and Neotropical species. Razafimandimbison et al. were not able to identify individual species groups and radiations within Malagasy Psychotria. Eventually, delimiting such radiations along with their morphological features and fine-grained geographic patterns within the island help understand why there are so many species of Psychotria here.