Comparison between Parsimony Analysis of Endemicity (PAE), Endemicity Analysis (EA), and an alternative coding of Three-Distribution Statements based on hypothetical distributions

Abstract

Areas of endemism are identified by a variety of methods, none of which is universally accepted. Performance of each method depends upon the variables chosen. Here, we compare Parsimony Analysis of Endemicity (PAE), Endemicity Analysis (EA), and a new coding method that we propose, Three-Distribution Statements (3DS). We rate performance based on the ability to identify hypothetical predefined patterns that represent non-conflicting, nested, and overlapping areas of endemism. Additionally, we also compared properties commonly used in analyses, such as shape and size of the area and the number of taxa involved. We found that 3DS has the best performance in retrieving predefined areas. EA is the only method that resolved a completely overlapping pattern, but it also found spurious patterns. Resolution with PAE always had intermediate precision and efficiency and so is not the best option for analysis of endemism. We recommend the use of 3DS together with EA as the best available option for hypothesizing areas of endemism.