We studied genetic variability of 100 isolates of Claviceps purpurea by usi
ng randomly amplified polymorphic DNA (RAPD), an EcoRI restriction site pol
ymorphism in the 5.8S ribosomal DNA (rDNA), the alkaloids produced, and con
idial morphology, We identified three groups: (i) group GI from fields and
open meadows (57 isolates), (ii) group G2 from shady or wet habitats (41 is
olates), and (iii) group G3 from Spartina anglica from salt marshes (2 isol
ates). The sclerotia of G1 isolates contained ergotamines and ergotoxines;
Gt isolates produced ergosine and ergocristine along with small amounts of
ergocryptine; and G3 isolates produced ergocristine and ergocryptine, The c
onidia of G1 isolates were 5 to 8 mum long, the conidia of G2 isolates were
7 to 10 mum long, and the conidia of G3 isolates were 10 to 12 mum long. S
clerotia of the G2 and G3 isolates floated on water. In the 5.8S rDNA analy
sis, an EcoRI site was found in G1 and G3 isolates but not in G2 isolates.
The host preferences of the groups were not absolute, and there were host g
enera that were common to both G1 and G2; the presence of members of differ
ent groups in the same locality was rare. Without the use of RAPD or rDNA p
olymorphism, it was not possible to distinguish the three groups solely on
the basis of phenotype, host, or habitat. In general, populations of C. pur
purea are not host specialized, as previously assumed, but they are habitat
specialized, and collecting strategies and toxin risk assessments should b
e changed to reflect this paradigm shift.