Aim Species distribution models and geographical information system (GIS) technologies are becoming increasingly important tools in conservation planning and decision-making. Often the rich data bases of museums and herbaria serve as the primary data for predicting species distributions. Yet key assumptions about the primary data often are untested, and violation of such assumptions may have consequences for model predictions. For example, users of primary data assume that sampling has been random with respect to geography and environmental gradients. Here we evaluate the assumption that plant voucher specimens adequately sample the climatic gradient and test whether violation of this assumption influences model predictions.Location Bolivia and Ecuador.Methods Using 323,711 georeferenced herbarium collections and nine climatic variables, we predicted the distribution of 76 plant species using maximum entropy models (MAXENT) with training points that sampled the climate environments randomly and training points that reflected the climate bias in the herbarium collections. To estimate the distribution of species, MAXENT finds the distribution of maximum entropy (i.e. closest to uniform) subject to the constraint that the expected value for each environmental variable under the estimated distribution matches its empirical average. The experimental design included species that differed in geographical range and elevation; all species were modelled with 20 and 100 training points. We examined the influence of the number of training points and climate bias in training points, elevation and range size on model performance using analysis of variance models. ResultsWe found that significant parts of the climatic gradient were poorly represented in herbarium collections for both countries. For the most part, existing climatic bias in collections did not greatly affect distribution predictions when compared with an unbiased data set. Although the effects of climate bias on prediction accuracy were found to be greater where geographical ranges were characterized by high spatial variation in the degree of climate bias (i.e. ranges where the bias of the various climates sampled by collections deviated considerably from the mean bias), the greatest influence on model performance was the number of presence points used to train the model. Main conclusionsThese results demonstrate that predictions of species distributions can be quite good despite existing climatic biases in primary data found in natural history collections, if a sufficiently large number of training points is available. Because of consistent overprediction of models, these results also confirm the importance of validating models with independent data or expert opinion. Failure to include independent model validation, especially in

Aim To explain the relationship between topography, prevailing winds and precipitation in order to identify regions with contrasting precipitation regimes and then compare floristic similarity among regions in the context of climate change.Location Eastern slope of the tropical Andes, South America.Methods We used information sources in the public domain to identify the relationship between geology, topography, prevailing wind patterns and precipitation. Areas with contrasting precipitation regimes were identified and compared for their floristic similarity. ResultsWe identify spatially separate super-humid, humid and relatively dry regions on the eastern slope of the Andes and show how they are formed by the interaction of prevailing winds, diurnally varying atmospheric circulations and the local topography of the Andes. One key aspect related to the formation of these climatically distinct regions is the South American low-level jet (SALLJ), a relatively steady wind gyre that flows pole-ward along the eastern slopes of the Andes and is part of the gyre associated with the Atlantic trade winds that cross the Amazon Basin. The strongest winds of the SALLJ occur near the 'elbow of the Andes' at 18°S. Super-humid regions with mean annual precipitation greater than 3500 mm, are associated with a 'favourable' combination of topography, wind-flow orientation and local air circulation that favours ascent at certain hours of the day. Much drier regions, with mean annual precipitation less than 1500 mm, are associated with 'unfavourable' topographic orientation with respect to the mean winds and areas of reduced cloudiness produced by local breezes that moderate the cloudiness. We show the distribution of satelliteestimated frequency of cloudiness and offer hypotheses to explain the occurrence of these patterns and to explain regions of anomalously low precipitation in Bolivia and northern Peru. Floristic analysis shows that overall similarity among all circumscribed regions of this study is low; however, similarity among superhumid and humid regions is greater when compared with similarity among dry regions. Spatially separate areas with humid and super-humid precipitation regimes show similarity gradients that are correlated with latitude (proximity) and precipitation.

A major problem in establishing effective protocols for conserving Madagascar's biodiversity is the lack of reliable information for the identification of priority sites in need of protection. Analyses of field data and information from herbarium collections for members of the plant family Pandanaceae (85 spp. of Pandanus; 6 spp. of Martellidendron) showed how risk of extinction assessments can inform conservation planning. Application of IUCN Red List categories and criteria showed that 91% of the species are threatened. Mapping occurrence revealed centres of richness and rarity as well as gaps in Madagascar's existing protected area network. Protection of 10 additional sites would be required to encompass the 19 species currently lacking representation in the reserve network, within which east coast littoral forests are particularly under represented and important. The effect of scale on assessments of risk of extinction was explored by applying different grid cell sizes to estimate area of occupancy. Using a grid cell size within the range suggested by IUCN overestimates threatened status if based solely upon specimen data. For poorly inventoried countries such as Madagascar measures of range size based on such data should be complemented with field observations to determine population size, sensitivity to disturbance, and specific threats to habitat and therefore potential population decline. The analysis of such data can make an important contribution to the conservation planning process by identifying threatened species and revealing the highest priority sites for their conservation.

Few studies have attempted to quantify the reduction or document the floristic composition of forests in Madagascar. Thus, we focused specifically on deforestation and plant diversity in Madagascar's eastern littoral community. We used a data set of approximately 13,500 specimen records compiled from both historical and contemporary collections resulting from recent intensive inventory efforts to enumerate total plant species richness and to analyze the degree of endemism within littoral forests. Change in littoral forest cover from original to current extent was estimated using geographical information systems tools, remote sensing data (satellite imagery and low-elevation digital photography), and environmental data layers. Of the original littoral forest only 10.3% remains in the form of small forest parcels, and only 1.5% of these remaining fragments are included within the existing protected-areas network. Additionally, approximately 13% of Madagascar's total native flora has been recorded from these forests that originally occupied <1% of its total land surface, and over 25% of the 1535 plant species known from littoral forests are endemic to this community. Given the ongoing pressure from human settlement along Madagascar's eastern coast, protection of the remaining forest fragments is critical for their survival. Fifteen of the largest intact littoral forest fragments we identified, collectively representing 41.5% of remaining littoral forest, are among priority sites recommended to the government of Madagascar for plant conservation and incorporation into the protected-areas network.

Abstract. Using the data published in the Catalogue of the Flowering Plants and Gymnosperms of Peru, we analyzed the elevational distributions of 5323 species reported as endemics from that country as a whole, for 10 families with the highest number of endemic taxa in Peru, and the distribution patterns of these species according to life form. We calculated the density of endemism (number of endemic species divided by area  1000) and absolute number of endemic species among life forms and families, along an elevational gradient. Overall densities of endemics were 10-15 times higher at mid-elevation (2000-3500 m) than in the Amazonian lowlands (0-500 m). Absolute numbers of endemics peaked at 1500-3000 m for herbs, shrubs, and epiphytes, while trees, vines, and lianas showed maxima in the lowlands (0-500 m); yet densities of endemics for all life forms peaked at 1500-3000 m. Among the 10 families with the highest number of endemics, densities of endemics peaked at mid-to high elevation (1500-4500 m), but showed much disparity in the elevational distribution of absolute numbers of endemic species. Finally, the percentage of endemic species to total species is highest for herbs, shrubs, and epiphytes. Given that less than 10% of the land area for each of the montane zones (2000-4500 m) is protected compared to 13.5-29.9% in the lower elevations (0-1000 m), we recommend that priority be given to increasing the size of protected areas at mid-to high altitude in the Andean slopes to grant further protection in zones with the highest density of endemics. We also recommend that more emphasis be given to collecting and studying non-trees, since most endemic species belong to that class.