Biogeographical Analyses and Applications

Abstract

Studies and applications in biogeography aim at understanding the causes and determinants of past, current, and future diversity distribution patterns. In this context, niche models are currently important tools. They make use of species georeferenced locations and sets of environmental variables to determine the potential suitable habitats of species. The derived results are applied for the conservation of biodiversity but their reliability strongly relies on the quality of the spatial information contained in biological databases. Biodiversity conservation effectiveness strongly depends on the identification of congruent areas for a set of diversity indicators. For example, congruent areas of high species richness for different biological groups.<br /> In the first part of this thesis a methodological framework is developed to evaluate the quality of biological databases. As a case example, a database of vascular plants in Benin, Ivory Coast and Burkina Faso is analysed. It consists of a total of 4,587 species collected in 2,931 different localities.<br /> Main criteria to evaluate database quality are the spatial configuration of collection localities, their spatial and environmental bias, and their floristic inventory completeness. It was shown that collection localities are unevenly distributed, forming strong clustered patterns and concentrated in the vicinity to cities, the coast, rivers, roads and protected areas. Collection localities represent only a narrow range of the environmental conditions in the study area. A gap selection index was created integrating density of collection localities, environmental bias and inventory completeness to represent those areas in need of more information and to guide future research activities. According to the gap selection index, only the Sahelian zone and surroundings of Comin-Yanga city in Burkina Faso, areas close to the coast and in the border with Liberia in Ivory Coast, and the region of the eastern Guinean forest in Benin are good represented.<br /> Different niche models strategies are tested to compare model performance between commonly used approaches, and those that seek to minimize the influence of spatial bias present in the occurrence records. The maximum entropy technique (i.e., Maxent) is used to model species ranges based on biased occurrence records and three types of environmental background information sets. These are random locations (’random background’), the locations of all collection localities except those of the target species (’target background’), and random locations weighed as a function of the gap selection index (’index background’). In average, target background based models perform better than random and index background based models. A visual examination confirms that only target background based models are able to correct for spatial bias. Species richness patterns are estimated to qualitatively describe spatial variations as a result of the three approaches employed Richness patterns strongly vary across the three approaches, but those based on target background models are more accurate to the known positive north-south richness gradient in the study area.<br /> In the second part of this thesis, the most comprehensive databases of vascular plants, amphibians and bats in West Africa are used to estimate geographical patterns of species richness and the range size rarity index (as a surrogate of species richness and endemism) at a half degree resolution. The relationship between abiotic factors and these two aspects of diversity are examined through the use of spatial auto-regressive techniques, indicating that elevation heterogeneity and temperature are the main determinants of variation in species richness and range size rarity patterns of the three taxa. Pair-wise correlation comparisons between the three groups for species richness and the range size rarity showed generally high and significant correlations (i.e. > 0.9). However, weak and even negative corelations were found when the comparison was applied at small geographical extents. Hotspots of the range size rarity index for the three groups together occupy 9.3% (i.e. 583.7 km) of the study area. 39.2% of this area is still covered by natural vegetation and 8.5% is covered by protected areas.<br /> The main finding of this thesis is that despite the large amount of information available to carry out research in biogeography, analysis and application should not be done without an objective and sound analysis of data quality. Only in this way sources of error and uncertainty can be identified, and proper modeling techniques can be applied. These are crucial steps that need to be followed before drawing conclusions and making decisions for the conservation of biodiversity.