ASSESSING THE LIMITS OF RANDOM AMPLIFIED POLYMORPHIC DNAS (RAPDS) IN SEAWEED BIOGEOGRAPHY

As judged by comparison with other molecular data sets, random amplifi ed polymorphic DNA (RAPD) data are robust in identifying large-scale b iogeographic populations that range from hundreds to thousands of kilo meters apart. As the geographical scale is shifted downward, however, RAPD data often fail. This is because RAPD data are inherently ''noisy '' as a result of technical artifacts and reproducibility problems ass ociated with non-independence of bands, ''missing'' bands, and the pre sence of de novo bands, all of which contribute to scoring errors in t he data set. To estimate the contribution of these error factors in al gal phylogeographic studies, segregation of RAPD bands in tetrasphorop hytic and gametophytic parents, their natural and synthetic offspring, and self-cycled tetrasporophytes were compared in Lophocladia trichoc lados (Mertens in C. Agardh) Schmitz and to a limited extent in Digene a simplex (Wulfen) C. Agardh. Wide-ranging biogeographic populations o f D. simplex were compared as were mixed populations of tetrasporophyt es and gametophytes. Results show that nested priming can lead to some nonindependence of bands but that this probably does not significantl y contribute to scoring error. Southern analysis using individual RAPD bands as probes revealed that up to 16% of visually nondetectable ban ds are actually present but that the random distribution of the error contributes uniformly across the data set. Non-parental (de novo in of fspring) and parental (not present in offspring) bands may contribute substantially to the scoring error in tetrasporophytes, gametophytes, and self-cycled tetrasporophytes. The presence of tetrasporophytes and gametophytes in a sample is not important in large-scale phylogeograp hic studies but does affect within-clade variation at smaller scales. We conclude that the overall level of error remains roughly constant a t probably between 5 and 10%, which is not a problem at large biogeogr aphic scales where the phylogenetic signal is strong. Finally, some un expectedly large abberations in RAPD banding patterns among life stage s in L. trichoclados were observed that cannot be explained by methodo logical artifacts alone clue to comparisons with synthetic offspring c ontrols. The possibility that carpospore amplification may not always involve a simple mitotic process is discussed.