Abstract Begonia plants were regenerated from leaf ex- plants treated with increasing concentrations of the chemical mutagen nitrosomethylurea (NMU). In these plants, we evaluated three methods to assess the extent of variation: a qualitative, phenotypic assay (the percent- age of aberrant plants), a molecular assay (changes in RAPD patterns) and a quantitative, phenotypic assay (variation in a quantitative trait). The qualitative, pheno- typic assay required a large number of plants per treat- ment (approx. 100) and careful, skilled judgement. It was sensitive to physiological variation. The RAPD as- say was not sufficiently sensitive: even at the highest NMU concentration there were no changes in RAPD pat- terns. The quantitative, phenotypic assay gave the best results: it was simple, objective and sensitive, and re- quired few plants per treatment (approx. 30). Plants were also regenerated from different types of intermediate cal- lus, and their variation was assessed. The performance of the three assays was essentially the same as with plants obtained after mutagenesis with NMU. An intermediate nodular- or non-nodular-callus phase resulted in slightly or strongly increased variation, respectively. In contrast to NMU-induced variation, callus-related variation, as determined in the quantitative, phenotypic assay, ap- peared to be to a large extent transient since it decreased strongly after a second direct-regeneration step. An in- termediate callus phase resulted in 2.5% juvenile plants. This aberration, which might be related with changes in the methylation status of DNA, was not observed in NMU-treated plants. Keywords Auxins · Begonia · RAPD · Regeneration · Somaclonal variation Introduction Plants produced vegetatively in tissue culture may be different from the plants from which they originate be- cause of physiological changes (e.g. bushiness), genetic changes, loss of pathogens and/or loss of chimerical structure. The term ‘somaclonal variation’ has been in- troduced to describe genetic changes due to tissue cul- ture (Larkin and Scowcroft 1981). It should be noted that the difference between somaclonal and physiological variation is not an all-or-nothing phenomenon but a mat- ter of degree (De Klerk 1990). DNA methylation, for ex- ample, may be very long lasting and even transmittable through meiosis (Jorgensen 1994). The potential of somaclonal variation to generate new, desirable traits for breeders has been advocated (among others by Larkin and Scowcroft 1981), and al- though valuable cultivars have been obtained (Bouman and De Klerk 1996), the results have not lived up to ex- pectations. Since the highest level of somaclonal varia- tion occurs in plants generated via adventitious shoot or somatic embryo formation, in particular after an interme- diate callus phase (Yamagishi et al. 1996; Piccioni et al. 1997; Plader et al. 1998), somaclonal variation is a prob- lem in breeding and propagation techniques that involve adventitious regeneration. In micropropagation, somatic embryogenesis via suspension cultures may suffer from somaclonal variation. In breeding, undesirable variation may be introduced during genetic engineering or haploid culture. An assay to measure the extent of somaclonal varia- tion will assist research into the backgrounds of soma- clonal variation and, from a practical point of view, en- able the selection of cultivars and tissue culture proce- dures with little or no somaclonal variation. In this pa- per, we evaluate three putative assays using plants re- generated from explants treated with the chemical muta- gen nitrosomethylurea (NMU). We have also studied the effect of tissue culture conditions on somaclonal varia- tion. Communicated by H.F. Linskens H. Bouman ( ✉ ) · G.-J. De Klerk Centre for Plant Tissue Culture Research COWT, P.O. Box 85, 2160 AB Lisse, The Netherlands e-mail: han.bouman@lbo.agro.nl Theor Appl Genet (2001) 102:111–117 © Springer-Verlag 2001 ORIGINAL ARTICLE H. Bouman · G.-J. De Klerk Measurement of the extent of somaclonal variation in begonia plants regenerated under various conditions. Comparison of three assays Received: 30 January 2000 / Accepted: 14 April 2000