Building a Robust, Densely-Sampled Spider Tree of Life for Ecosystem Research

Phylogenetic relatedness is a key diversity measure for the analysis and understanding of how species and communities evolve across time and space. Understanding the nonrandom loss of species with respect to phylogeny is also essential for better-informed conservation decisions. However, several factors are known to influence phylogenetic reconstruction and, ultimately, phylogenetic diversity metrics. In this study, we empirically tested how some of these factors (topological constraint, taxon sampling, genetic markers and calibration) affect phylogenetic resolution and uncertainty. We built a densely sampled, species-level phylogenetic tree for spiders, combining Sanger sequencing of species from local communities of two biogeographical regions (Iberian Peninsula and Macaronesia) with a taxon-rich backbone matrix of Genbank sequences and a topological constraint derived from recent phylogenomic studies. The resulting tree constitutes the most complete spider phylogeny to date, both in terms of terminals and background information, and may serve as a standard reference for the analysis of phylogenetic diversity patterns at the community level. We then used this tree to investigate how partial data affect phylogenetic reconstruction, phylogenetic diversity estimates and their rankings, and, ultimately, the ecological processes inferred for each community. We found that the incorporation of a single slowly evolving marker (28S) to the DNA barcode sequences from local communities, had the highest impact on tree topology, closely followed by the use of a backbone matrix. The increase in missing data resulting from combining partial sequences from local communities only had a moderate impact on the resulting trees, similar to the difference observed when using topological constraints. Our study further revealed substantial differences in both the phylogenetic structure and diversity rankings of the analyzed communities estimated from the different phylogenetic treatments, especially when using non-ultrametric trees (phylograms) instead of time-stamped trees (chronograms). Finally, we provide some recommendations on reconstructing phylogenetic trees to infer phylogenetic diversity within ecological studies.