Studies in Annonaceae . VIII . A cladistic analysis of Tetrameranthus

Summary The small genus Tetrameranthus (five species) stands isolated within the Annonaceae. A cladistic analysis was carried out using macromorphologicalcharacters in order to find possible apomorphies and to attempt a phylogenetic reconstruction. In the “best” cladograms there appear two subsets, one formed by T. duckei, T. macrocarpus, and T. pachycarpus, the other by T. laomae and T. umbellatus. Both are supported by a number of apomorphic character states. Any other conclusions remain speculative.


Introduction
All attempts to find allies for Tetrameranthus, so far, led to nothing or to contradictions at the most.Therefore a logical step now was to subject Tetrameranthus to a cladistic analysis in order to find out apomorphic character states, and to attempt to give a phylogenetic reconstruction.Only macromorphological data could be used, since a complete set ofdata of other kind is not available for each ofthe five species at this moment.Cladistic analysis is performed by means of a method that integrates the concepts of compatibility and parsimony (Zandee, 1985).A description and application can be found in Roos (1986) and Zandee and Geesink (1987).
The Data-Matrix Within Tetrameranthus several aspects of variation may be noticed: -arborescent vs. shrubby habit; four out of the five species now known are trees; T.   duckei is a shrub to a small tree (note: in the revision (Westra, 1985) T. macrocarpus is erroneously described as "Shrub to tall tree"); 1 Institute of Systematic Botany, Heidelberglaan 2, 3584 CS Utrecht, The Netherlands.Recently, a taxonomic revision of the genus Tetrameranthus was published by Westra   and collaborators (Westra, 1985).That paper essentially deals with presentation of data (including two newly described species), it does not make any statements on possible relationships other than accepting the genus as a member ofthe Annonaceae.This is because Tetrameranthus has a unique combination of characters and stands rather isolated within the family.Principal among these characters are: 1) leaves arranged in a spiral, 2) the perianth in whorls of four, and 3) flowers subtended by a verticil of four bracts; for more data the reader is referred to Westra (1985).
Various positions for Tetrameranthus have been proposed.In Fries's (1959) compre- hensive survey of the Annonaceae it is placed as the sole member ofa tribe Tetramerantheae in subfamily Annonoideae.Walker (1971), in a classification based on his palynological studies, places the genus in the Uvaria tribe ofthe Malmea subfamily.It should be remarked in this context that there is a considerable discrepancy between palynological data by Hesse and Waha (1984) and those by Walker.Most recently, karyological investigations by Mora-  wetz (1986b) show that Tetrameranthus could have a closer connection with the African genus Uvariopsis and a looser connection with the neotropical Guatteria tribe, rather than forming a monotypical tribe or being part of a heterogeneous Uvaria tribe sensu Walker.MAY 1988   -indument of comparatively large stellate hairs (trichomes = rays mostly £0.5 mm long) in T. duckei, T. macrocarpus, T. pachycarpus vs. minute stellate hairs (trichomes mostly SO.I mm long) in T. laomae and T. umbellatus; -rather large fruits with a diameter (in sicco) exceeding 3.5 cm in T. macrocarpus and T. pachycarpus vs. distinctly smaller fruits not exceeding 2.5 cm in diameter (in sicco); -a prominent to flat primary vein on the adaxial side in T. laomae and T. umbellatus vs. an impressed primary vein onthe adaxial side in T. duckei, T. macrocarpus, T. pachycarpus-, -flowers single or flowers (up to) five in umbels in leaf axils.The latter condition is only found in T. umbellatus, although T. duckei may exceptionally develop two flowers together; -peduncle well-developed, the bracts well beyond the base of the inflorescence vs.
peduncle reduced, the bracts at the base of the inflorescence, tiny and partly missing.The latter condition is found in T. laomae only; -anthers with a protruding, conical apical prolongation of the connective in T. duckei, T. macrocarpus.T. pachycarpus vs. with a flat, cushion-shaped top in T. laomae and T. umbellatus.
Together with some less conspicuous characters which need not to be mentioned at all in detail here, these characters form the base of a binary data-matrix (Table 1).
List of code numbers used in the data matrix and cladograms indicating character states in Diameter of (dried) fruit becoming >3.5 cm 1 0 0 1 0 21.
Diameter of (dried) fruit remaining <2.Choice of the Outgroup For outgroup selection, the following features were especially taken into consideration: -leaves in a spiral.Leaves in Annonaceae are placed in two alternating rows (1/2) in essentially all representatives but Tetrameranthus.Among the rare exceptions (fide Fries, 1959) are a few species of Artabotrys and possibly Annona crotonifolia, where leaves on some shoots have been found in three rows (1/3).A further exception is Disepalum anoma- lum where, similarly, a phyllotaxis of 2/5 has been observed; -tetramerous flowers.Such flowers may exceptionally be found (usually among the regular flowers) on individuals in probably all species of Annonaceae that normally have trimerous flowers.Deviations of this kind are nothing unusual, and occur in other families of flowering plants as well.Truly tetramerous flowers are met with, apart from Tetramer- anthus, in Asimina tetramera and in Reedrollinsia (only in R. cauliflora); -stellate trichomes.Within the family, predominantly simple (to occasionally furcate) hairs are found.Stellate (and/or scaly) hairs are encountered, however, in a variety of not necessarily related genera and species in Annonaceae.
Out ofthose, the number of perianth members per whorl seems to offer the best handhold.
There are no species (known) in Annonaceae outside Tetrameranthus that have leaves in an arrangement other than 1 /2 over the whole plant (in the cases mentioned only some of the shoots on a plant are involved).The indument, on the other hand, seems to leave too wide a variety of choice, and so does not lead us anywhere.
For practical reasons, Reedrollinsia had to be dropped, since it is too incompletely known at the moment.This leaves Asimina tetramera as the only candidate out of all applicants for the job, even though populations of this species certainly do not have exclusively tetramerous flowers (Krai, 1960).
Asimina tetramera is a shrub to about 3 m tall with axillary, single flowers with a perianth consisting of one whorl of four or three sepals and two whorls of four or three petals, and with (rather) large, fleshy monocarps with several seeds of 1-2 cm long.It occurs in eastern peninsularFlorida (Krai, 1960).Note thataxillary inflorescence position and comparatively large monocarps with large seeds (1-2) are (also) characteristic of Tetrameranthus!Recently it was found that karyologically the African genus Uvariopsis shares a great deal with Tetrameranthus (Morawetz, 1986a(Morawetz, , 1986b; see also introduction to this paper).
Morphologically there seems to be little similarity between the two, however, at least no more than between Tetrameranthus and the bulk of annonaceous genera.The tetramerous corolla in a single whorl (!) in Uvariopsis might be regarded as made up of two dimerous whorls rather than a single truly tetramerous whorl; the number of sepals, after all, is two!

Character Analysis and Discussion
The outgroup choice just discussed cannot be but a provisional one, as complete data with regard to other candidates are still lacking.Therefore it was decided to have separate analyses: first, with the five species assigned to Tetrameranthus only, and, secondly, with the addition of Asimina tetramera as the outgroup.
Ten of these cladograms fit the data equally well, the other two (Figs. 1.1 and 1.2) only fall behind because of one more ambiguity remaining in the character state pairs 16-17 and/or 20-21, due to lack of an outgroup.
When Asimina tetramera is introduced as an outgroup, only six completely resolved cladograms result from an analysis using partially monothetic sets as building blocks for cladograms.One of these stands out as the best regarding the balance between fitting (15) and contradictory (7) character states.It corresponds with Fig. 1.3, but it has A. tetramera now in the position of global outgroup (Fig. 2.1).In fact, it is the only cladogram out of these six that features A. tetramera in this position.All other grouping opportunities for as sistergroup.Only those apomorphic character states mentioned in the text are given.For further explanation, see text.(See Fig. 2 for key to symbols used.)