The elusive Juncus minutulus: a failure to separate tetra- and hexaploid individuals of the Juncus bufonius complex in a morphometric comparison of cytometrically defined groups.

Screening of nuclear genome size was carried out on ca 2400 plants from over 120 mainly Central- European localities of the Juncus bufonius group. Besides the diploid level, corresponding to known diploid species (in this case J. ranarius, J. hybridus and J. sorrentini), two polyploid cytotypes were detected, conforming with the tetraploid and hexaploid levels treated by some authors as separate species: J. minutulus and J. bufonius s. str. The relationship between nuclear DNA content and the number of chromosomes was verified by chromosome counting. Polyploidy, as opposed to agmatoploidy can, therefore, account for the karyological variation. The 2C values of diploid, tetraploid and hexaploid individuals were ca 0.65, 1.18±2.8% and 1.84±1.6% pg 2C DNA, respectively. No other cytotype or statistically significant variation in nuclear genome size was found. To asses the utility of hitherto published morphological characters distinguishing J. minutulus from J. bufonius s. str., measurements of seven floral and three vegetative quantitative characters were obtained (no less than 10 measurements per flower, 30 per plant) for 358 mature plants of known ploidy level from 47 localities. Diverse ordination and clustering techniques did not indicate the presence of any grouping in the dataset. Canonical discriminant analysis and stepwise variable selection indicated that inner tepal length followed by mean capsule width and mean capsule length were the most useful characters for identifying the two ploidy levels; however, the estimated 10-fold cross-validation error rate of a simple k nearest neighbour classification analysis was 0.45. Other analyses corroborated this result. No new morphological character that would allow successful separation of tetraploids from hexaploids was discovered. This provides independent support for the opinion of some previous authors that J. bufonius L. is best treated as a single variable species comprising two cytotypes that are inseparable using hitherto suggested diagnostic characters until convincing proof to the contrary is available.