Vieses especiais em dados de ocorrência de peixes de água doce do Brasil

Abstract

Spatial sampling bias, i.e., the predominant sampling effort in some regions in detriment of others, affects the quality and representativeness of biodiversity data. These biases can be generated by many factors, such as sampling in areas near to urban centers, presence of roads, lack of financial resources for more spatialized samplings, preference for samplings nearby or within conservation areas, preference for a particular species and/or place, among others. Although the knowledge about freshwater fish diversity has increased rapidly, there is still an knowledgement gap of their occurrences, and studies about sampling biases in aquatic environments are scarce. Thus, we evaluated if the sampling of freshwater fishes is biased due to factors such as accessibility, research infrastructure and preference for protected areas. Here, we compiled the occurrence records of 3358 species of Brazil (27.3% of the worldwide biodiversity) and modeled the sampling effort (collection events) in relation to the proximity to research centers, proximity to and density of roads, proximity to protected areas and populational density, considering both the whole Brazilian extension and each country region individually. We found that the sampling effort is spatially biased, more concentrated in the South and Southeast regions of Brazil. In addition, for Brazil, the most important factors to explain the variation in sampling effort were, in a decrescent order of importance, the proximity to research centers, populational density, density and proximity to access routes and proximity to protected areas. Despite variations in the importance of factors explaining sampling effort among regions, the proximity to research centers was significant for all regions. This result reinforces the relevance of the internalization and regionalization of research infrastructure to increase knowledge about biodiversity.