KatharinaSchumannGoethe University Frankfurt, Institute of Ecology, Evolution and Diversity

Max von Laue Str. 13Frankfurt60438Germany

KatharinaSchumannassociatedPartyBlandineNacoulmaCo-ownerYvonneBachmannCo-owner2013-10-24In West Africa, rural households use several plant products from a wide range of species to meet their everyday needs and as a source of additional incomes (Kristensen and Balslev 2003; Taita 2003; Belem et al. 2007; Paré et al. 2010a). However, West African savannas are subject to drastic land use and climate changes (IPCC 2007; Wittig et al. 2007; Brink and Eva 2009; Ouédraogo et al. 2010) that are thought to change species composition and distribution (Da 2010; Heubes et al. 2011). There is an urgent need to better understand the effects of climate and land use changes on plant diversity patterns in savannas and their underlying factors. Species distribution models (SDM) present a robust methodology which has increasingly been used over the last two decades to relate biological data to environmental factors (e.g. Guisan and Thuiller 2005; Ferrier and Guisan 2006). Efforts have been made in understanding species distribution and plant diversity patterns in West Africa (Schmidt 2006; Da 2010; Márquez 2010; Heubes 2012) at regional or national scale, but there are no published data about the factors influencing these patterns at local scale. Nacoulma et al. (2011) showed that plant diversity differed between a protected and communal area. Thus, we hypothesized that land use is an important factor for species distribution at local scale. We investigated the distributions of protected and non-protected plant species and diversity patterns in- and outside of a protected area in a West African savanna (W- National Park, Burkina Faso). We aimed to assess the influence of land use and other environmental factors on the species distribution and diversity patterns.burkina fasocommunal arealand useplant diversityprotected areaprotected speciesspecies distribution modelingwest africaObtain permission from data set owner(s)W National Park and communal area, Burkina Faso1.752.2512.349711.666720072011altogether were modeled: 66 woody species and 129 Herb speciesSpeciesKhaya senegalensisSpeciesMitragyna inermisSpeciesAdansonia digitataSpeciesVittelaria paradoxaYvonneBachmannGoethe University Frankfurt, EU Project UNDESERT

Max von Laue Str. 13Frankfurt60438Germany

bachmann@bio.uni-frankfurt.de

Occurrence data of plant species were extracted from vegetation inventories performed in the study area between 2007 and 2011 (Schumann et al. 2010; Nacoulma et al. 2011; Schumann et al. 2011). Vegetation plots were placed within the study site following a stratified random design using ASTER satellite images from 2006 and 2007 with a spatial resolution of 30 m and soil maps in order to cover the main occurring habitat types (i.e. dry, medium, wet, and rocky conditions). 178 vegetation inventories were performed in the protected area and 212 in the communal area. Based on all described inventories, 2717 spatially unique occurrence points with 18818 species occurrence records of woody and herb species were available for the study site. The species occurrence points consisted in total of 594 species (441 herb species and 153 woody species). For modeling we used only species with more than 10 occurrence points. Species nomenclature followed Gautier et al. (2006). The environmental datasets used in the species distribution model Maxent 3.3 employed to predict the plant species distribution patterns were selected according to their ecological importance. Altogether seven environmental variables were applied in the model: mean annual precipitation (Worldclim-Database, HIJMANS et al. 2005), elevation, aspect, slope and potential incoming direct insolation (all four derived from the Aster Digital Elevation Model (DEM)) and soil/crown humidity (Kauth & Thomas 1976). Additionally the Normalized Differenced Vegetation Index (NDVI) was calculated to point up the differences between biomass production in the protected and the communal area. The NDVI and soil/crown humidity were derived from three nine-band ASTER images of the 11th of October 2006 (Earth Remote Sensing Data Analysis Center 2006). Processing of the images and of the environmental variables was done in ERDAS IMAGINE 9.1 (Leica Geosystems Imaging, LLc 2006), ArcGIS 10 (ESRI Inc., Redlands, USA) and SAGA-GIS 2.0 (SAGA User Group Association, Hamburg, Germany). After AUC value evaluation we analyzed 66 species distribution models for woody and 129 for herb species.

attached_fileNot authorized to access resourceOthertimeseriesField data were raised in Burkina Faso bei Schumann K. and Nacoulma B. between 2007 and 2011, the distribution models represent present species distributionotherField data were raised in Burkina Faso bei Schumann K. and Nacoulma B. between 2007 and 2011, the distribution models represent present species distribution