Studies in the southern African species of Justicia and Siphonoglossa (Acanthaceae): seeds

The seeds of all species of Justicia (except J. thymifolia (Nees) C.B. Cl.) and Siphonoglossa in the southern African region (as defined by the Flora of southern Africa) were examined with the Scanning Electron Microscope. A wide range of different seed surfaces were seen, many of which could be correlated with other characters to further reinforce grouping of species into sections. However, some species that on other characters, would be placed together, had widely differing seed testas. Gross surface morphology varied from smooth to colliculate, irregularly rugose, reticulate with the ridges variously arranged, wheel-shaped (radiate or ammonite-like), with a central ridge or with multicellular barbed scales. The seeds were without hairs, unlike some American species sometimes considered to belong in Justicia (Graham 1989). Micromorphology also provided useful characters, with seeds being reticulate, papillate, minutely or deeply rugose or with one to many crystals visible. Critical-point drying was sometimes helpful in viewing microstructure.


INTRODUCTION
Very few SEM investigations of the seed surfaces of African members of the Acanthaceae have so far been carried out. Those with relevance to the present study are briefly reported on below. Munday (1980), in revising the southern African species of Monechma (Justiciae), found that, although all seeds had a smooth testa, there was variation in their size, shape and colour. In his paper on Anisotes (also Justiciae) Baden (1981b) found that the seeds often had a longitudinal ridge, and he described the texture as rugose, tuberculate, verrucose or smooth. In one species (A. guineensis Lindau) the seeds were glandular. This species was unusual in other ways also, such as having 4-porate pollen, and was placed by Baden in a separate monospecific section. Balkwill & Getliffe Norris (1984) used the fact that the seed surface in Hypoestes was asperous, tuberculate or smooth in their key to species.
Balkwill, Getliffe Norris & Schoonraad (1986) used the microsculpturing of the seed surface in their investigation of Peristrophe and concluded that its value at present was mainly at species level. The testa of the species of the genus was basically reticulate with minute papillae, and may or may not have tubercules bearing one, two or three layers of recoiled hooks. Details of the tubercule surface were also used. Hedren (1987) in his revision of Justicia capensis Thunb. and its tropical African allies, described the seed surface only as 'tuberculate' and did not use it to distinguish species. In his paper on the Justicia mollugo group in tropical Africa (1988), he described the seeds of most of the species dealt with, but did not use them in his key.
Balkwill & Getliffe Norris (1988) re-appraised tribal and subtribal limits within the Acanthaceae of southern Africa, and considered the lack of hairs on the seed coat to be definitive of Justiciae and some genera of Acanthoideae. However, Graham (1989) included species of Justicia with a variety of hair types, some even being tested by her and found to be hygroscopic (in her sect. Anisostachya). Graham's delimitation of Justicia was very wide, including a number of previously segregate genera, and she described some 20 seed types. In sect. Sarotheca some species had seeds with hairs having recoiled barbs similar to those seen by Balkwill et al. (1986) in Peristrophe. In sect. Vasica some of the Old World species had smooth testas, a character usually considered characteristic of Monechma, which she accepted as a distinct genus.  Where fresh fruiting material was available from the nursery (National Botanical Institute, Pretoria), seeds were also dehydrated, fixed and critical-point dried by the following method: 1, specimens were dehydrated either in series of ascending concentrations of alcohol or, more usually, in two changes of 2,2-Dimethoxypropane (DMP) for up to two hours; 2, they were then transferred to 100% acetone for five minutes;

Seeds
3, critical point drying was done with liquid CO, evaporated off at 40°C and 80-90 atmospheres (critical point for C 02 is at 31,3°C and 72,8 atmos.); 4, the dried specimens were stored in a desiccator with silica gel, or immediately mounted on aluminium stubs for coating.

All seeds were glow-discharge coated with metallic gold in an Eiko sputter coater. They were then viewed with an MSM 4 Hitachi-Akashi (desk top model) SEM at kv 15 and photographed.
Delimitation and names of sections are according to Graham (1989).

RESULTS AND DISCUSSION
Results of critical-point drying: it was noted in many instances that material that had not been critical-point dried showed structures of the testa such as radial cell walls, ornamentation of these walls, and crystals in the testa, better than fresh, critical-point dried material. This was due to the collapse of the outermost cell walls, which made it easier to see underlying structures. On the other hand, structures such as fine papillae were sometimes only visible when the material had been pre-treated before viewing. This was the case with J. anselliana (Nees) T. Anders. (Figure 1G), and it was felt that critical-point drying would be a valuable procedure for those species which, like7. anselliana, had 'amorphous' testas, as found in sects. Ansellia and Justicia p.p.
A wide variety of seed surface types was seen, sometimes peculiar to the species, but in at least some cases correlating with characters used to demarcate sections. the basis of other characters should probably be placed with J. crassiradix in subsect. Ansellia, do not show this semicircular pattern. Instead, the ridges are randomly placed on the seed surface. At higher magnifications, when they were viewed without having been critical-point dried, the seeds of all three species had an irregularly rugose surface ( Figure IE). When seed of J. anselliana was critical-point dried, the cells in the depressions between the ridges were found to each have a centrally situated papilla ( Figure 1G), while the cells on the ridge were without papillae, having instead a minutely verrucose surface. The papillae were longer than wide, with an acute or truncate apex. It was not possible to establish whether these papillae were present in the other two species, as fresh material was not available.  Figure 2C, D), which on other characters might be considered to occupy an isolated position among the southern African species of the genus, also has very atypical seeds. These are almost smooth, being only very slightly rugose even at high magnifications ( Figure 2D) and compare well with Monechma mollissimum (Nees) PG. Mey., which was also viewed for purposes of comparison ( Figure 2E, F). However, it was decided that, despite similarities in the seeds and capsules, J. bolusii could not be transferred to Monechma. It differs from Monechma in its 'spicate' inflorescence and its tricolporate pollen with entire margocolpi (Immelman 1989 Figures 3D-G; 8). These papillae are visible in material that was not pre-treated be fore mounting, as well as in critical-point dried material of 5. leptantha subsp. leptantha ( Figure 8A). Material of Siphonoglossa viewed without pre-treatment had an irregularly rugose testa.

Sect
In J. dinteri two kinds of capsule are regularly produced, a four-seeded normal capsule and a one-seeded smaller capsule with four irregularly toothed wings. The seeds from each type of capsule were compared, and the testa found to be similar, though the seeds in the one-seeded capsules were larger ( Figure 8B, D, F, cf. Figure 8C, E). One other species was examined, which on flower and pollen characters possibly belongs in sect. Hamieria. This was / capensis Thunb. (Figure 1A-C), which was placed in this section by Graham (1989). The seeds, however, are different from the papillate seeds of the other species in the section, being colliculate with a strongly reticulate micro-pattern. CONCLUSION Both macro-and micro-pattern of the seed testa may be considered of major importance in placing many species in their sections, and delimiting some sections in Justicia, but need to be used with caution as they do not always correlate with other features. They are also suggestive of a relationship between some species oi Justicia sect. Harnieria and Siphonoglossa.