Lämpötilojen ääriarvojen merkitys kasvillisuuden mallinnuksessa

• Main Author: Käkelä, Saara
• Other Authors: Helsingin yliopisto, Matemaattis-luonnontieteellinen tiedekunta, Geotieteiden ja maantieteen laitos, University of Helsinki, Faculty of Science, Department of Geosciences and Geography, Helsingfors universitet, Matematisk-naturvetenskapliga fakulteten, Institutionen för geovetenskaper och geografi
• Format: Master Thesis
• Language: Finnish
• Published: Helsingfors universitet 2014
• Subjects: Geography; Maantiede; Geografi; Arctic; Arctic Ocean; Gam; Climate change; Subarctic
• Online Access: http://hdl.handle.net/10138/42511

Due to global climate change there will be an increase in frequency of extreme temperatures. That will have an effect to species distribution regions. By adding extreme temperatures on a species distribution model based on the basic climatological variables we could improve the understanding of biological effect of climatic variability and extremes. The aim of this study is to develop a species distribution model for arctic-alpine and boreal vascular plant species and examine does the extremes improve predictions. The other question is, does vascular plant species from two different biogeographical distributions responds similarly to bioclimatic variables? A study area (26 000 square kilometre) is located in Northwestern Finland and it belongs to subarctic climate regime. A climate model is produced from the data of 61 climate stations from 1971–2000. Modeling was done with generalized additive model (GAM) by using geographical variables as explanatories (geographical location, elevation, the effect of the Arctic Ocean, lake and mire cover). Total sample of vascular plant species was 1182 of 1x1 km grids. The distribution model was done with three different statistical techniques (GLM, GAM, GBM). First a simple model was modeled with three baseline variables (a temperature of the coldest month, water balance, growing degree days) and then the extreme temperatures were added to compose a full model. The predictive power of the models was tested by calculating the area under the curve of a receiver operating characteristic plot (AUC) and the true skill statistic (TSS) for both models. Incorporating the extreme temperatures into the distribution model significantly improved the accuracy of distribution model of the both plant groups. The improvement was small but statistically significant. The relative importance of each predictor variable and the response of each bioclimatic variable to occurrence of species varied between the plant groups. The most significant explanatory variable to explain the arctic-alpine ...