BLUMEA 22 (1975) 175—195 Comparative anatomy of the Crypteroniaceae sensu lato 1 G.J.C.M. van Vliet and P. Baas Rijksherbarium, Leiden, Netherlands Summary The and nodal of and leaf, twig, anatomy Alzatea, Axinandra, Crypteronia, Dactylocladus, Rhynchocalyx described detail. This delimited and Van is in family, as newly by VanBeusekom-Osinga Beusekom, appears the wood anatomical in to be very heterogeneous, a conclusion supported by diversity reported another The nodes be unilacunar with of the paper. may simply unilacunar, a complete girdling trace, common gap in Further (or split lateral) plus median trace type, or trilacunar. Cortical bundles may be present addition. is in cuticular stomatal of diversity present e.g. texture, type, hypodermal development, arrangement vascular bundles in petiole and midrib, mechanical support of the veins, the crystal complement, foliar sclereids, and cork origin. evidence character of all and close mutual of Anatomical only supports the Myrtalean genera, a affinity Axinandra and Crypteronia. These genera, together with Dactylocladus show several features characteristic wide for some Melastomataceae, which family with its anatomical range would also cover most of the anatomical diversity of Lythraceae, Sonneratiaceae, Oliniaceae, Alzatea, and Rhynchocalyx. Rhynchocalyx ap- to be closer in its to and Melastomataceae than to the other mem- pears anatomy some Lythraceae, Oliniaceae, bers of Crypteroniaceae. Alzatea shows affinities with all families mentioned, but remains problematic with its in families. The anatomical trilacunar node notknown to occur the other complex pattern ofoverlapping of Crypteroniaceae s.l., Melastomataceae, Oliniaceae and Sonneratiaceae may be ranges Lythraceae, , interpreted between families. as evidence ofintimate relationships these Introduction the of Beusekom delimitation of the In present issue Blumea, Van proposes a new family Crypteroniaceae, uniting Alzatea, Axinandra, and Rhynchocalyx with Crypteronia number of and Dactylocladus. These five genera share a supposedly primitive relic char- of acters, absent or rare occurrence in other Myrtalean families, implying that mutual between affinities the individual genera are not necessarily very close. Furthermore, they are also characterized by a combination ofmore specialized features by which the delimi- of the factual base tation the Crypteroniaceae in new sense appears to have a comparable to that of Melastomataceae and Sonneratiaceae. The naturalness of the family remains, however, open to some doubt. A detailed anatomical study was therefore undertaken to intra contribute to our understanding of and inter-family relationships of these genera, and and to compare the results on macromorphology (Van Beusekom, 1975) pollen with those of wood and this morphology (Muller, 1975) anatomy (Van Vliet, 1975, paper) and leaf, twig, and nodal anatomy (this paper). The of all five has been reviewed complicated taxonomic history genera by Van Beusekom. Since the anatomical evidence clearly supports a Myrtalean affinity for all of in the wood and in leaf genera (presence vestured pits intraxylary phloem and twig) literature only families from this order, mentioned in the as putative relatives (e.g. Melastomataceae, Lythraceae, Sonneratiaceae) will be considered in the comparisons. leafand of the s.l. Hallier Data on twiganatomy Crypteroniaceae are very scanty. (1911) and described the of Alzatea. Nie- Stant (in Lourteig, 1965) briefly vegetative anatomy 176 BLUMEA VOL. 22, No. 2, 1975 denzu (1892) recorded some anatomical characters for Crypteronia, and Sprague & from Metcalfe (1937) did so for Rhynchocalyx. Metcalfe & Chalk(1950) in compiling data the older literature only included some information on Axinandra and Crypteronia. the Reference to these publications will only be made as far as is relevant for discussion of the results. present attention is Apart from a detailed histological study of leaf and twig, special given to the course of vascular bundles in the nodal region. MATERIALS AND METHODS available for data will be Only herbarium specimens were this study. Collecting given for each species. Since the specimens used in this study were also used for the revision herbaria (Van Beusekom-Osinga & Van Beusekom, 1975) no of origin will be cited. and Sections macerations were prepared according to standard techniques (Baas, 1970; sections unembedded ofnodal and Bongers, 1973). Serial were made of twigs large parts serial of internodal regions. These sections were only bleached and studied in semi- permanent glycerin mounts. All nodal regions were studied at some stage of secondary thickening. Specimens used for the study ofnodalanatomy are marked with an asterisk. far of the The anatomical terminology employed, as as uncommon use, is same as ex- elsewhere & & plained (Jansen Baas, 1973; Bongers 1973; Van Staveren Baas, 1973). RESULTS AND DISCUSSIONS Nodal anatomy (Fig. 1—4) the literature the term node has been different this study the In applied in ways. For definition the node that of the which the of following is employed: is part stem in course the vascular bundles from the leafto the stele (or vice versa) produces a deviation of the vascular pattern in the stem. For practical reasons the course of vascular bundles will be described as descending from the leaf into the stem. This should not be taken as a devel- data the ofthe vascular bundles opmental interpretation, since no on ontogeny in Crypter- available. The classification ofnodal based on the number of and oniaceae are types gaps useful tool in traces (Sinnott, 1914) provides a descriptive (Howard, 1974). However, the such that this classification be some Crypteroniaceae pattern is a complex one cannot employed for all genera or species. the sections In all Crypteroniaceae leaves are oppositely arranged. Serial through one of with the nodal region therefore afford the study vascular connections of two leaves node stele. It appeared that even within a single differences may exist between the vascular the of the of patterns at opposite sides stem. More evidence variability within a species was found in the few instances that several nodes per species or specimen were studied. descend from the leaf. The median bundle fuses In Crypteronia three traces large im- mediately with the stele in all species. In C. griffithii and C. macrophylla (fig. la) lateral stele the leaf traces girdle around the and fuse with those from opposite to enter a com- the lateral condition described Howard mon gap, conforming to split or common gap by for (1970) several families e.g. Rhizophoraceae. In C. cumingii an additional centric bundle from each lateral the stele level below (lateral anastomosis) splits off trace to enter at a just in the the medianbundle In C. or same gap as (fig. ib). paniculata var. paniculata (Kochum- men 97773, fig. ic) the situation is comparable to that in C. cumingii but the girdling lateral the would traces fail to enter the stele after fusion on place where one expect a common gap. On one side of the node the lateral anastomosis was found to fuse with the median G. J. C. M. van Vliet & P. Baas: Anatomy of Crypteroniaceae 177 Idealized three-dimensional of nodal C. similar to Fig. 1—4. diagrams types. 1. Crypteronia 1a. griffithii, Dactylocladus; 1b.C. cumingii;1c. C. paniculata var. paniculata (KochummenKEP. F. 97773). — 2.Axinandra; A. A. 2a. A. beccariana (similar toCrypteronia paniculatap.p.); 2b. alata; 2c. zeylanica.— 3. Alzatea verticillata. — 4. Rhynchocalyx lawsonioides. l.t. c.b. = cortical bundle; e.g. = common gap; c.t. = complete girdlingtrace; l = lateral anastomosis; = lateral trace; m = median trace. No. 178 BLUMEA VOL. 22, 2, 1975 bundle above the level of its entrance into the stele. In the other three specimens of the C. paniculata laterals form a complete horizontal girdling path showing no special from in locus of fusion; the lateral anastomoses both leaves enter the stele the same gaps the Axinandra as the median traces (similar to situation pictured in fig. 2a for beccariana). The fact that in Crypteronia these lateral anastomoses vary in their connections with the and stele from a three-traced condition at the median gap (C. paniculata var. affinis, partly in C. paniculata var. paniculata and C. cumingii) to a situation where these anastomoses their have own gaps at a lower level (partly in C. paniculata var. paniculata and C. cumingii) makes it impossible to fit them into the nodalclassification of Sinnott. Their absence in other and the above makes difficult two Crypteronia species reported variability it to assess the systematic value of these lateral anastomoses. In the formal descriptions the different conditions will be referred to follows: Nodes with will be described as a common gap as with laterals median these from having a common gap split plus one trace (to distinguish where several lateral traces and be types gaps may present as reported by Howard, 1970). The nodal will be referred to a trace pattern as in C. paniculata as complete girdling plus median bundle. Lateral if will be mentioned in addition. one anastomoses, present, Axinandra the nodal all similar that In patterns are to in Crypteronia paniculata var. affinis (fig. 2a—c). Cortical bundles may, however, be present in addition. These bundles be collateral if centric sections may or, sufficiently large, (amphicribal). In transverse through the internode these bundles are situated at the four corners, sometimes in wing- in radial with like extensions, and occasionally arranged rows the bundles decreasing in size towards the periphery. In the latter instance the bundles are always fused with each other just below the node. In the nodal region the cortical bundles descend from the lateral traces at the points of insertion of the lateral anastomoses and bend into the outer cortical region, passing through the layer ofcork tissue. In Axinandra zeylanica the cortical bundles end in the outer cortex at various distances from the node.In A. alata no complete internodes were sectioned but the fact that both above and below the lateral trace a (the is is cortical continuous same?) cortical bundle attached, suggestive ofa bundle system over several nodes(fig.