Rapid progress on the vertebrate tree of life

Abstract

Background: Among the greatest challenges for biology in the 21st century is inference of the tree of life. Interest in, and progress toward, this goal has increased dramatically with the growing availability of molecular sequence data. However, we have very little sense, for any major clade, of how much progress has been made in resolving a full tree of life and the scope of work that remains. A series of challenges stand in the way of completing this task but, at the most basic level, progress is limited by data: a limited fraction of the world's biodiversity has been incorporated into a phylogenetic analysis. More troubling is our poor understanding of what fraction of the tree of life is understood and how quickly research is adding to this knowledge. Here we measure the rate of progress on the tree of life for one clade of particular research interest, the vertebrates.

Results: Using an automated phylogenetic approach, we analyse all available molecular data for a large sample of vertebrate diversity, comprising nearly 12,000 species and 210,000 sequences. Our results indicate that progress has been rapid, increasing polynomially during the age of molecular systematics. It is also skewed, with birds and mammals receiving the most attention and marine organisms accumulating far fewer data and a slower rate of increase in phylogenetic resolution than terrestrial taxa. We analyse the contributors to this phylogenetic progress and make recommendations for future work.

Conclusions: Our analyses suggest that a large majority of the vertebrate tree of life will: (1) be resolved within the next few decades; (2) identify specific data collection strategies that may help to spur future progress; and (3) identify branches of the vertebrate tree of life in need of increased research effort.

Figure 1

Cumulative phylogenetic information amassed for the last 16 years. The accumulation of sequences for vertebrates in GenBank (a), papers using the term 'phylogeny' or 'phylogenetics' in the Web of Science database (b) and phylogenetic resolution (measured as the proportion of nodes with at least 50% bootstrap support) in the vertebrate tree of life resulting from these research efforts (c). In all cases, the data are cumulative from the start of each analysis. Phylogenetic resolution is calculated as in Table 1. Trend lines are exponential in (a), and second order polynomial in (b) and (c).

Figure 2

Phylogenetic progress among major vertebrate lineages. Phylogenetic resolution as a proportion of the total possible nodes resolved (measured as the proportion of nodes with at least 50% bootstrap support) in high (a) and low-diversity (b) major vertebrate lineages. The black line in each panel shows the overall proportion of nodes resolved in vertebrata, calculated for all species pooled; note the different scale of the y axis in panels (a) and (b). The small decreases in resolution for the Crocodilia and lamprey are due to stochastic effects and the small size of the clades (see Additional File 1).

Figure 3

Scatterplot of dataset characteristics. Each circle represents the characteristics of the complete 2008 data matrix from each of 100 vertebrate clades. The number of species in the dataset (x-axis) refers to the number of species sampled for the clade, not the total described diversity of that clade. The number of character refers to the number of columns (nucleotides) for that matrix, rather than the total number of sequences sampled. The size of each circle represents the density of each dataset measured as the percentage of non-missing data.

Figure 4

Projections of future progress. Projection of the best-fitting trend line for species sampling, phylogenetic resolution and strongly supported phylogenetic resolution. Phylogenetic resolution refers to the proportion of nodes resolved in a 50% majority rule bootstrap consensus tree. Strong support refers to this same measure in a 95% majority rule bootstrap consensus tree.