
 
UPGMA dendrogram based on combining data grouped 32 
S. rotunda
 
accessions  into  five  different  groups  at  76.2%  similarity.  Cluster  I 
contained 22  
 
accessions including 1, 5, 2, 3, 4, 13, 14, 15, 16, 17, 22, 18, 20, 23, 
28, 29, 30, 31, 21, 19, 24, and 27. Cluster II had two accessions 11 
and  12.  Cluster  III  comprised  accessions  25  and  26.  Cluster  IV  had 
only one accession 32. Cluster V consisted of five accessions including 
6, 7, 9, 10, and 8 (Fig 4). 
The  Mantel  test  and  Pearson's  correlation  (
r
)  exhibited  a  moderate 
coefficient  of  0.543  (p<0.0001)  between  ISSR  and  RAPD  markers 
whereas genetic matrics of RAPD or  ISSR makers and integrated data 
showed  a  strong  correlation  with 
r
  =  0.930  and  0.814  (p<0.0001), 
respectively  (Table  4).  The  results  indicated  that  the  efficiency  of 
genetic  variation  analysis  of  RAPD  to  32 
S.  rotunda
  accessions  is 
higher than that of ISSR. 
Principal component analysis (PCA) based on the combined RAPD and 
ISSR data displayed the genetic relationships of 
S. rotunda
 accessions 
in  two-dimensional  space  accounted  for  45.71%  and  17.17%  of  the 
total  genetic  varia  difference.  The  PCA  data  separated  32 
S.  rotunda
 
into  major  clusters  that  appeared  similar  to  the  tree  diagram 
generated from the cluster analysis (Fig 5). Accessions 1, 12, 11, 13, 5, 
14, 15, 3, 4, and 2 were grouped into one cluster whereas accessions 
16, 19, 17, 21, 23, 18, 30, 31, 22, 29, 28, and 20 were in the same 
cluster. Accessions 6, 7,  8, 9, and 10 were gathered together in  one 
cluster,  and  accessions  24,  25,  26,  and  27  were  in  one  group. 
Unsurprisingly,  accession  32  was  isolated  into  one  group  from  the 
remaining accessions (Fig 5).