From species presences to abundances: Using unevenly collected plant species presences to disclose the structure and functioning of a dryland ecosystem

Species abundance data is essential to understand ecosystems structure and functioning and to support species and habitat conservation. However, most regional to global databases provide only presence or presence/absence data. The main aim of this paper is to develop a methodology to estimate plant species abundances from a presence/absence database using as a case-study the largest and one of the most diverse tropical dry forest of the world – the understudied Caatinga vegetation, that dominates in the drylands of Brazil. Plant data missed abundance estimations and derived from different sources, with uneven sampling efforts over space and time. Starting from the raw data, we considered only the presence records of terrestrial plant individuals identified to the species-level. Afterwards, we applied the re-sampling method to estimate species abundances thus obtaining database DB1. To deal with the uneven sampling effort along the study area and increase information reliability, we filtered DB1 in two ways: (i) we excluded re-sampling units with a lower sampling effort and produced the Database DB2; (ii) we excluded low occurrence species and build the Database DB3. The reliability of the databases was compared by calculating a measure of their completeness. DB1 had 789 species over 323 sampling units, DB2 retained 530 species distributed in 38 sampling units, and DB3 retained 48 species over 113 sampling units. In DB1 and DB2, despite the different number of species considered, the percentage of exotic (7%), endemism (14%), woody (44%), climber (12%), and herbaceous species (45%) was similar. DB3 included only native species (no exotic species) and displayed a higher percentage of endemism (29%) and woody species (79%), and a lower proportion of herbaceous species (21%) than DB1 and DB2. The databases obtained provide an important basis to improve Caatinga ecological knowledge and conservation: we suggest the use of DB2 to support conservation strategies, and of DB3 to support ecosystem structure and functioning studies. Moreover, the re-sampling methodology proposed to estimate plant abundances from presence data, dealing with uneven sampling efforts across large areas and over time, provides an important tool that may be used to obtain abundance data, often essential to the development of plant-based indicators of ecosystem structure and functioning, and to support conservation studies.