Under & Verboom, Generic limits in the Rytidosperma complex 
605 
is 75% of the length of the caryopsis. D. domingensis approaches Rytidosperma most 
closely by its bristly lodicules (Jacobs 1982), but Conert (1975) describes the lodicules 
as glabrous, raising the possibility that one of the authors was working with 
misidentified material. However, its chromosome number is 36 (Davidse & Pohl 
1972). It is necessary to decide whether these character distributions should be 
regarded as exceptions to the rule, or whether they should be interpreted as indicating 
a more complex pattern of relationships between Rytidosperma and Dautlionia. If the 
exceptions were to some extent correlated, then they would probably indicate that 
Dauthouia is nested within Rytidosperma. However, the exceptions occur singly, and 
Danthonia species with one Rytidosperma character do not have any others. We suggest 
that this indicates homoplasy, rather than a distinct pattern. We would therefore 
regard them as exceptions, rather than as intermediates between the two genera, 
and on a phylogenetic analysis they should appear as autapomorphies. Thus we 
interpret the data as indicating two distinct lineages: one for Rytidosperma and allied 
genera, and one for Danthonia. 
The position of the Himalayan species Danthonia cachemyriana (including D. exilis 
and D. jaquemontii) and D. schneideri are not clear; Tomlinson (1985) suggested a 
relationship to Karroochloa, but there has never been a formal placement of these 
species outside Danthonia. The analysis of all species aligns them to the Rytidosperma 
s.l. clade rather than to Danthonia. In D. cachemyriana the lodicules have microhairs, 
the lemma indumentum is distinctly tufted, and the hilum 30% of the caryopsis 
length. D. schneideri also has lodicules with bristles and microhairs (Veldkamp, pers. 
com.), but the lemma indumentum is poorly tufted, with marginal strips of hair, and 
the hilum is 60% of the caryopsis length. However, despite these results we do not 
feel satisfied that we have seen enough good quality, convincing material to make 
the formal transfer of these species from Danthonia to Rytidosperma s.l. 
All Danthonia species described from Malesia by Veldkamp (1979, 1993) should be 
included in the Rytidosperma s.l. clade; Veldkamp's descriptions and our own studies 
of some of the species indicate this clearly. 
Plinthaiithesis, Schismus, Karroochloa and Notochloe 
This is a robust clade, which is retrieved by all analyses, and which is robust to 
bootstrapping. This clade is defined by a reduction in the degree of lobing of the 
lemmas, a reduction in the development of the lemma awns, the absence of a tufted 
indumentum on the abaxial surface of the lemma and the absence of bulliform cells. 
Our analyses indicate that Plinthanthesis is more closely related to Notochloe and 
Schismus than to either Danthonia or Rytidosperma. Although Plinthanthesis and 
Notochloe have very different spikelet morphologies, the spikelet morphology of 
Notochloe is unique, and so uninformative on its relationships. But the two genera 
share the presence of islands of clear cells between the vascular bundles. This is the 
only occurrence of this character in the danthonioid grasses (although it does occur in 
Phragmites, a member of the Arundineae s.s.), and so may be considered to be a good 
synapomorphy for them. In addition, this grouping is supported by the molecular 
analyses of Barker (1995). The two genera are associated with Danthonia by the mesic, 
moniliform leaf-blades and linear hila, but there is no direct morphological support 
for the molecular association between Danthonia and Plinthanthesis. 
The association of Schismus and Tribolium with Plinthanthesis and Notochloe is 
interesting. Tribolium is endemic to southern Africa, while Schismus has most of its 
species there, with one species found around the Mediterranean. The group has no 
representatives in South America, unless the molecular data supporting the link to 
Danthonia are corroborated. 
