11.) Pubescence on adaxial surface of external tepals. 0, sericeous; 1, glabrous. 
12.) Red trichomes on floral structures. 0, present; 1, absent. 
13.) Staminal glands. 0, present; 1, absent. 
14.) Sporangia of stamens of whorl IIT. 0, four; 1, two. 
15.) Pedicel in fruit. 0, partially turbinate; 1, entirely turbinate. 
16.) Tepals in fruit. 0, entirely persistent; 1, partially persistent; 2, deciduos. 
17.) Hypanthium interior. 0, sericeous; 1, glabrous. 
Figure 26 shows the result of the analysis with all characters unordered. Figure 27 shows 
the result of setting the change in number of sporangia as irreversible (to root the tree), actually 
the only character change that could be visualized in a defined evolutionary trend at the species 
level. 
Considering the outcome from both cladograms (figures 26 and 27), it is suggested that 
the subdivision of the American species into subgenera based on the number of sporangia in the 
anthers of whorl III may be wrong. Figure 26 shows at least two independent events for the 
change in this character; from the node where C. se/lowianum derives up to C. quadrangulum, 
and the clade conformed by C. heterantherum, C. napoense, and C. formicarium. Figure 27 
shows at least three different points for the reduction of the number of sporangia in whorl III (see 
taxa mentioned before). On the other hand, the trend of four to two sporangia has been 
considered to occur independently in different lines across the family (Rohwer, 1994), a matter 
which is also appreciated in the surveys attempted here with the genera. Furthermore, the change 
from four to two sporangia has been recorded also in some Asian species of Cinnamomum 
(Kostermans, 1985). Therefore, the subgenera Heteranthera and Euphoebe created by Mez (1889) 
243 
