604 
Telopea Vol. 6(4): 1996 
Discussion 
The overall consistency indices and retention indices for the trees located are low. 
This is consistent with rampant convergence and parallelism, which is evident from 
the cladograms, showing the relatively low number of unique synapomorphies within 
this group. However, the basic pattern in the relationships was retrieved by all 
analyses, despite variations in the number of species included, variations in the 
characters used, etc., suggesting that the basic patterns located are robust. 
A problem associated with the analysis of large data sets of often 'messy' data is that 
calculation times are slow, thus limiting the possibility of doing statistical tests based 
on character manipulations. In addition, such data sets often contain many possible 
arrangements of the taxa, which can limit the ability of the search algorithms to find 
the most parsimonious solutions. There are two sampling approaches to dealing 
with these problems: the first is to include as many species as possible, yet not be 
able to search the data set adequately; the other is to use a smaller set of representative 
species, but possibly miss important character combinations. The results of the first 
analysis almost certainly are not maximally parsimonious, while the results from the 
second analysis may represent an overly simplified picture. 
Keeping these caveats in mind, the following interpretations can be made from 
these results. 
Danthonia 
The distinction between Danthonia and the Rytidosperma clade is strongly supported. 
This node was retrieved by all character and taxon manipulations, was unaltered by 
successive weighting, and obtained a bootstrap support value of 79%. 
Danthonia, as circumscribed here, is defined by the presence of cleistogenes among 
the culm sheaths and by a base chromosome number of 18. With this definition, it 
includes 23 species, with nine in South America, eight in North America, and three 
from Europe to the Himalayas. Although none are native to Australia, Danthonia 
decuinbens has been introduced in Tasmania and Victoria, where it is rather rare. 
Zotov (1963) separated Rytidosperma from Danthonia by lemma indumentum and 
hilum shape. In Rytidosperma the lemma indumentum is tufted or patterned, while 
in Danthonia it is either evenly scattered on the back, or in marginal lines. The hilum 
in Rytidosperma is punctiform; in Danthonia it is linear. To these characters Veldkamp 
(1980) added ciliate lodicules and larger caryopses as typical of Rytidosperma. The 
two taxa also differ in basic chromosome number, with 2n = 24 in Rytidosperma and 
36 in Danthonia. 
Jacobs (1982) critically analysed the characters used by Veldkamp (1980), and showed 
that there are exceptions to every character used by Veldkamp and Zotov. The lemma 
indumentum character is possibly the most difficult to quantify, as the distinction 
between the marginal strip of hair and the marginal tufts of a tufted indumentum is 
in many cases subjective. The hilum shape/length and lodicule indumentum 
characters have some striking exceptions in Danthonia, according to the data presented 
by Jacobs. However, from the data available, the exceptions are not correlated: species 
with bristly lodicules generally do not have short, punctate hila. Jacobs is particularly 
worried about the placement of D. cirrata, D. secundiflora and D. rhizomata. Yet in all 
three the lodicules are generally glabrous; and Tomlinson (1985) did not record 
bristles or microhairs for these species; only D. secundiflora has an oblong, short 
hilum; and D. cirrata has 72 chromosomes (either a tetraploid based on 2n = 36, or a 
hexaploid based on 2n = 24). D. unispicata has bristly lodicules, but the linear hilum 
