A BIOSYSTEMATIC STUDY OF RUMEX SECTION RUMEX IN CANADA AND THE UNITED STATES

A biosystematic study of Rumex section Rumex was carried out in order to resolve problems that have arisen since the last revision of the group in 1937. The introduction of new species and range extensions of others have resulted in the misidentification of many species. Two areas of infraspecific confusion were evident: (1) in R. maritimus and (2) in R. occidentalis. Also, little was known about the nature and extent of hybridization within the section in North America. The study utilized herbarium specimens from the major North American herbaria along with field observations and the study of cultivated material.;The investigation of the breeding biology of the species indicated high levels of pollen fertility and seed set which can be used as a baseline from which supposed hybrids can be recognized. Observations based on the flowering times of the species indicated that temporal barriers are of little importance in preventing gene flow between the species.;A review of the cytological literature on the species is presented. The chromosome numbers can vary from 2n = 20-200, with most species being represented by the diploid, tetraploid and hexaploid levels. Pairing in meiosis of the species is normal.;Eleven interspecific hybrids were found to occur in North America; the pollen fertility and seed set in those hybrids examined are both extremely low. Observations made on meiotic material of one of the hybrids and information from the literature indicates abnormal pairing. A close assessment of hybrid morphology is used in indicating the putative parents.;The usefulness and reliability of the various vegetative and fruit characteristics used to discriminate the species are discussed.;Numerical techniques are utilized to resolve the infraspecific problems and in examining interspecific associations. R. maritimus is now recognized to consist of one introduced and three native varieties separable on tubercle size and shape. In R. occidentalis, two varieties are recognized separable on valve length. Canonical analysis did not suggest groups of species that could be easily delimited from each other or that corresponded directly to existing subsections.;A key to the species is given and species descriptions and distributions within North America are presented. Where possible type material of the species names was examined and it is shown that the name R. britannica a Linnaean name should be reinstated for the native species referred to as R. orbiculatus in recent works.