Orlovich, Drinnan & Ladiges, Floral development in the Metrosideros group 
717 
The three united stamens opposite each petal in Tristania neriifoUa probably reflect 
the lack of space available at the time of primordia initiation, followed by enough 
expansion in the antesepalous regions to keep every third pair of stamen primordia 
separate, but the lack of availability of early ontogenetic stages precludes a definite 
conclusion. Thaleropia queenslandica (Fig. 21) has even fewer stamens, and these are 
clearly inserted directly on the flank of the invaginated apex and are evenly and 
widely separated by subsequent expansion of the flower. 
Although the diversity in form of mature flowers can be adequately described in 
terms of discrete morphological units such as stamen fascicles, this is not always the 
most appropriate approach for the best understanding of the mechanism of 
evolutionary change or the circumscription of systematic characters. Flowers, like all 
morphological structures, are the result of complex ontogenetic processes that have 
spatial and temporal components. Differences in mature flower structure usually 
result from differences in ontogeny, and so factors affecting the spatial and temporal 
relationships of floral organ initiation and development best describe the key events 
that lead to this diversity. 
The genera of Myrtaceae examined here are good examples of this. Lophostemon 
confertus has a well developed pre-staminal bulge that is quite large, allowing the 
insertion of in excess of one hundred stamen primordia. However, constriction of 
the base of the FSB in a static, non-expanding area of the flower does not allow for 
maintenance of integrity of one hundred expanding stamen primordia, which merge 
and result in a fascicle of united filaments. The comparatively earlier inception of 
stamens in Tristaiiiapsis laurina leads to fewer stamens developing on a smaller pre- 
staminal bulge. The pre-staminal bulge (and antepetalous area) is still enlarging in 
that critical period when stamen primordia are expanding, so the filaments remain 
separate and there is little or no common filament tissue in the fascicle. Lophostemon 
Inctifluus is intermediate. The development of the pre-staminal bulge is sufficient to 
result in a fused fascicle of filaments, but stamen initiation is sufficiently early to 
prevent a vertical groove in the pre-staminal bulge. Consequently, each fascicle has 
fewer stamens than L. confertus, and is not cleft. In Xanthostemon and Lyskarpus 
stamens appear very soon after petals and there is no independent development of 
a pre-staminal bulge, and no fascicle. If there is insufficient expansion of the flower, 
especially in early stages, there is no room for more than a few stamens to form, e.g., 
Tristania neriifoUa. Although the morphological differences in mature flowers manifest 
as variation in number and arrangement of parts, their cause is differences in timing 
and duration of organ initiation on floral apices growing at different rates. 
Because changes in spatial/temporal factors are potentially variable and reversible, 
evolutionary change from one condition to the other in either direction is possible. 
If a species has potential to produce a pre-staminal bulge, but this is inhibited by 
early stamen initiation directly on the apex, then there will be no fascicles. However, 
a delay in stamen initiation would allow the pre-staminal bulge and fascicles to 
form (Figs 22, 23). It is a good illustration of how features that are 'potential' by 
virtue of one aspect of the genotype (i.e., pre-staminal bulge) can be modified by 
results of another aspect (i.e., timing of stamen initiation). The potential to produce 
the pre-staminal bulge may not have been lost; it is just precluded by a change in the 
timing of the ontogenetic processes. A reversion in the timing would allow the pre- 
staminal bulge, and hence fascicles, to manifest again. 
The rigid framework of the classical concept of angiosperm flowers as consisting of 
four successive whorls of organs, i.e., sepals, petals, stamens and carpels, does not 
provide a category of organ to which an entire fascicle of stamens can be homologised. 
Consequently, stamen fascicles have usually been interpreted as a modification, both 
ontogenetic and phylogenetic, of a single stamen (see Leins 1964, 1975). While this 
