478 
Telopea 8(4): 2000 
Evidence supporting the recognition of new families 
Members of the same plant family can be expected to form a monophyletic group 
separate from related families; they should also share synapomorphic features by 
which the family can be characterised and contrasted with other families. Criteria of 
relationship and morphology must both be met. 
In the present case, in brief, molecular evidence requires the exclusion of Hopkinsia and 
Lyginia from Restionaceae. Morphological evidence gives no basis for these genera 
being placed in Anarthriaceae and similarly fails to give a basis for uniting them in a 
single new family. The analyses of morphological and DNA data in this case produce 
contradictory results, but these appear to indicate a degree of homoplasious and 
convergent morphological evolution, rather than implying substantially different 
phylogenies. 
The justification for recognising the new families Hopkinsiaceae and Lyginiaceae 
depends on establishing that: 
(1) Hopkinsia and Lyginia are not correctly placed within the Restionaceae; 
(2) synapomorphies shared with Anarthriaceae are not sufficient for them to be 
included in that family; and 
(3) they are not appropriately placed together in a single new family. 
(1) Exclusion from Restionaceae 
ie major evidence for exclusion of Hopkinsia and Lyginia from Restionaceae comes 
rom analysis of DNA sequence data. Cladistic parsimony analysis of DNA sequences 
° *■ e chloroplast gene rbcL and of the trn L intron, together with the tniL-lniF 
intergenic spacer, does not associate Hopkinsia and Lyginia with other genera referred 
to Restionaceae but shows a robust clade of these two genera with Anarthria 
(Mg. 1 modified from Briggs et al. 2000). This clade has 100% jackknife support in 
separate cladistic analyses of each of the two regions of the chloroplast DNA, as well 
as ot the combined data. In addition, the clade is marked by two indels in the trnL 
m on, an insertion of five bases and a deletion of three; these indels show no 
tiomoplasy among other members of the Poales studied. This molecular evidence 
comes w io y rom chloroplast DNA but is supported by the highly conserved rbcL 
gene, as v\ e as a less conserved region of DNA that is now being widely used in 
phylogenetic studies (e.g. Bayer & Starr 1998) and has given moderately well resolved 
rees in our studies of Poalean families. The consensus trees from the combined DNA 
r . f , aCeS , e ' ,r ' a c l fl de as sister to the Restionaceae clade (Restionaceae with 
l . ? e ^! L 3C -?f e ' ®ta from rbcL alone similarly distinguish these and the Poaceae 
tee eaciwi l ,, jackknife support) but places them (and several other families) 
at an unresolved polychotomy. 
t / le reSU ^j f' orn studies (Fig. 1) as giving a clear indication that 
in a rMH Hopkt ™\ md h ^ mw form * clade. These genera, however, fail to associate 
!IiIrf an !i‘ 1S , 0nnl0rph0lft8ic ' 11 data ( pi 8- 2 ' modified from Linder et al. 2000); 
Restinmr ”‘rru ^‘ nm , a ' ld Hopkinsia emerge successively at the base of 
Centrnlpn H 6 cladograms also show contrasting positions for the 
noTdwiK?!>H ea ° f that iamii V are being further investigated and are 
Dositions nf r m 10 P resent P a P er - Similarly, the two trees show differences in the 
are beimr § en fra within Australian Restionaceae and these discrepancies 
arise from h f U ' P ' ^. uc ' 1 a contr asting result for Hopkinsia and Lyginia could 
perhans durinp 0 ^ .'T'r' 5 3n convergent change in morphological characters, 
morDholn(rinl^ eVO U '° n L , n ? C f r s ‘ mi ' ar environmental conditions. Constraining the 
P g c tree so that Hopkinsia and Lyginia associate with Anarthria, rather than 
