A dissertation submitted to the Faculty of Science, University of the Witwatersrand, Johannesburg, in fulfilment of the requirements for the degree of Master of Science. Johannesburg, 2014. / The persistence of rare plants is an important dimension in the conservation of biodiversity. Consequently an improved understanding of the nature and determinants of plant rarity and its relation with vulnerability to extinction could provide a basis for “proactive conservation” instead of the present day tendency for conservation actions to be “reactive”.
In this dissertation I explore the relation between plant rarity and anthropogenic pressures (land transformation and use), biophysical factors, and plant traits in KwaZulu-Natal Province. Rarity was defined as the product of plant species abundance (population size) and its distribution (extent of occurrence). A number of a priori hypotheses regarding plant rarity were developed from the literature and these were then tested on a sample of plant species from KwaZulu-Natal. Species were selected in a stratified random manner to include species from different levels of threat and rarity or commonness. As the interest of this study was KwaZulu-Natal, only KwaZulu-Natal records were used for the analysis. Although the study suffered from a paucity of data particularly on the biological traits and behaviour of each species I was able to explore rarity in terms of seed dispersal distance, stress tolerance, habitat specificity and ecological niche width. I also explored potential island effects based on a species affinity to isolated erosional land surfaces and the anthropogenic effects of utilization and land transformation.
To get an initial insight into relations, rarity was compared with each explanatory variable independently prior to using a multiple regression analysis approach aimed at understanding the potential interactive effects of suitable variables on rarity. Three different analytical techniques were used to provide a more robust understanding of the variable associations. These included Regression tree analysis (CART Salford Systems Inc., USA) and two generalized linear regression approaches; Generalized Linear Modelling (GLM) and Generalized Additive Modelling (GAM).
All three multiple regression methods indicated that niche width had the strongest influence on rarity. Although Land Surface was shown to be the second strongest variable this, according to the GLM and GAM analyses, was due to a
positive correlation between species with no affinity to land surface and species commonness. Visual representation of the regression tree analysis showed inconsistent partitioning of this variable throughout the tree indicating that land surfaces are not good predictors of rarity.
Although the relation between Rarity Index and Habitat Transformation is not linear it was shown to be significant (p <0.1(p=0.0549)) after “smoothing” in GAM analysis. A smoothing curve on the bivariate analysis and the regression tree analysis indicated that species start to become rare after approximately 36% of their habitat is transformed.
While GLM and GAM showed little or no relation between life history, dispersal distance, habitat specificity and rarity, the regression tree selected habitat specificity as the third most important splitter in the tree and dispersal distance was selected as a primary splitter for species with a niche width of greater than four. These differences observed in the three multiple regression analyses highlight the value of using more than one method to explore relations in ecological data.
Considering all three analyses Niche Width is the strongest determinant of Rarity in KwaZulu-Natal, followed by Habitat Transformation and then Habitat Specificity. This improved understanding of the determinants of rarity will enhance our ability to prioritise plant species for conservation action.
Key Words: ecological niche width, habitat specificity, habitat transformation,
rarity, seed dispersal distance, stress tolerance, human use.
| Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:wits/oai:wiredspace.wits.ac.za:10539/15077 |
| Date | 31 July 2014 |
| Creators | Church, Brigitte |
| Source Sets | South African National ETD Portal |
| Language | English |
| Detected Language | English |
| Type | Thesis |
| Format | application/pdf |
Page generated in 0.0018 seconds