Matrix models in population ecology
Matrix theory is a powerful tool in the study and description of populations. This study commences by discussing the theory relevant to ecological studies, before reviewing recent uses of matrix models in population dynamics. Stage-structured and multistate matrix models are then developed to describe populations of Cirsium palustre in the U.K. and of Carduus nutans in New Zealand. These models were parameterized using field data and provide information on transient dynamics, long-term growth rates, reproductive contributions and eventual stable distributions. Sensitivity and elasticity analyses allow the robustness of the models and the relative contributions of the various life history stages to the population growth rate to be assessed. In the case of C. nutans, a major weed pest ofpasture land, this allows ontogenetic states vulnerable to control by mechanical, chemical or biological means to be targeted. These models are further refined to include the effects of environmental stochasticity. For C. palustre, lack of information on the existence and persistence of a seed bank limited the model construction, and so field and greenhouse experiments were conducted to provide additional information about seed viability and the factors affecting seedUng recruitment and survival. In addition, a theoretical critique of the use in comparative ecology of elasticity analysis to detect trade-offs in life history strategies is presented. Simulation and analytical studies indicate that this approach is not valid. Finally, the use of matrix models in evolutionary demography and management strategy formulation is discussed. As a whole, this thesis illustrates the strength and versatility of matrix models used in population ecology.