Sample Data and Training Modules for Cleaning Biodiversity Information

Abstract

The recent appreciation of rapid global losses of biodiversity has created a growing demand for quick and reliable access to high-quality primary biodiversity data, essential for conservation and evolutionary science, among other applications. As a consequence, biodiversity databases have become increasingly available, allowing large-scale assessment of patterns and processes influencing the evolution of life on Earth. However, data quantity is often compromised by low data quality, and, even though working with plenty of records could be tempting, building ecological models with inaccurate data mislead researchers in driving conclusions about reality. The most common types of errors in biodiversity data are those related to georeferencing, which vary from obvious to barely noticeable, making them challenging to recognize. Fortunately, Geographic Information Systems (GIS) have increasingly allowed identification of georeferencing mistakes. Here, we provide a hands-on exercise for data cleaning that allows easy and prompt detection of inaccurate information. We focus on the use of GIS software for fixing problems related to the geographic coordinates of the data, such as changes in the order or the sign of latitude-longitude values, and records placed outside the study region, or outside continents.