Floristic diversity and the landscape mosiac : an analysis of the vegetation of the island of Newfoundland

Includes bibliographical references (pages [122]-130)Research directions in vegetation ecology were reviewed with emphasis on the continuum concept, phytosociology, relationships between plant guilds at the landscape level, and the lack of hypothesis testing in studies employing ordination and phytosociological techniques. Continued research into compositional parallels among taxonomic guilds along measured environmental gradients was suggested to address hypotheses regarding vegetation structure in landscapes. Using a data base of 1280 releves from three major vegetation types across the Island of Newfoundland, a synthesis of previous floristic analyses of the island was performed. Correspondence between previous phytosociological classifications of releves and their ordination positions was assessed using principal coordinates analysis (PCoA). Overlap in the distribution of phytosociological associations was observed within vegetation types, and to a lesser extent across vegetation types; furthermore, no statistically significant discontinuities in floristic composition (assessed using the split moving window technique) were observed within or across vegetation types. These observations were supportive of the continuum concept, and indicative of the limitations of the phytosociological approach as a foundation for ecological studies on patterning in landscapes. Pattern diversity analysis was used to compare species richness and landscape complexity between the island and other boreal North American regions. Species richness for the island was lower but landscape complexity was similar to that of other regions, indicating that conclusions drawn from this analysis have broad applicability. A basis for the identification of causes of pattern was established by investigation of compositional similarities among vascular and cryptogam plant guilds along environmental gradient complexes on the island. Three correlation techniques were used: correlation of similarity matrices, correlation of releve positions on PCoA axes, and canonical correlation of PCoA axes. Strong correlations existed between guilds within and across vegetation types. Correlations were strongest in forest dominated regions and weakest in heath. The strength of correlations increased with increasing turnover of species along gradients (& diversity). Possible directions of causation for observed relationships included determination of the composition of one guild by the control exerted on local environmental conditions by the other guild, or independent determination of the composition of both guilds by regional abiotic factors acting through local environmental conditions.M.S. (Master of Science