CGRG Bibliography of Canadian Geomorphology
Search Results
Author : Goring, S.J.; Lacourse, T.; Pellatt, M.G.; and Mathewes, R.W.
Date : 2009.
Title : How sensitive are pollen-based climate models to large-scale vegetation change? An example from Marion Lake, British Columbia.
Publication : CANQUA–CGRG Biennial Meeting. May 3-8, 2009. Simon Fraser University, Burnaby Campus, Burnaby, British Columbia.
Issue : Programme and Abstracts Volume.
Page(s) : 80.
Abstract
Paleoclimate reconstructions using pollen assemblages obtained from modern sediments may be confounded by the effects of anthropogenic change. At Marion Lake, within the UBC Research Forest in southwestern British Columbia, strong evidence of logging appears in a short (38 cm) lake sediment core, represented by rapid increases in proportions of Alnus and Pteridium and decreases in proportions of Pseudotsuga and Cupressaceae pollen. Determining whether large-scale vegetation change will have an effect on reconstructed climate at the site is important. Longer term paleoclimate reconstructions using pollen may be especially sensitive to changes resulting from landscape level disturbance processes. While this is the case, it is often difficult to predict the effect that these events will have on climate reconstruction. We use pollen proportions from the Marion Lake short core, along with a network of 539 modern pollen samples from lacustrine sediments in British Columbia and adjacent Washington state to develop pollen-based climate models using downscaled climate data from the ANUSPLIN model. These pollen and climate data are combined using a number of techniques including the Modern Analogue Technique (both weighted and unweighted), Partial Least Squares, Weighted Averaging, Weighted Averaging Partial Least Squares, Non-Metric Multidimensional Scaling/Generalized Additive Model, and randomForest. Model output from each technique is then combined to produce a model of predicted climate at Marion Lake throughout the sampling period, from approximately 1865 - 1974. These data are compared to weather records from the lower mainland of British Columbia from 1874 – 1980. Our results show that the pollen-based climate reconstruction is not biased by the effects of logging within the UBC Research Forest. We attribute this lack of effect to the range of seral stages sampled in the modern pollen network. The British Columbia Modern Pollen Database effectively covers a breadth of seral stages among Coastal Western Hemlock sites that correspond to the variability in the Marion Lake core. This indicates the potential for this regional pollen database to properly distinguish changes in climate, such as those in the early-Holocene, from local or regional vegetation turnover resulting from short-term disturbance events.
Bibliography of Canadian Geomorphology