Among the millions of invertebrate species, the genomes of insects (particularly fruit flies of the genus Drosophila) and nematodes (from the genus Caenorhabditis) have come under the closest scrutiny. Now it is time for annelids - the segmented worms - to reveal their DNA sequences and gene structures. Last November, Raible and co-authors reported in Science the initial investigation of 30 genes from the marine annelid Platynereis dumerilii 1. And it revealed a big surprise. The sequences of the annelid proteins were found to be more closely related to their human orthologs than to the insect and nematode orthologs. Moreover, among the species compared, the exon-intron structure of P. dumerilii genes was also most similar to that of humans: the human and the marine worm genomes have the highest number of introns per gene (7.8 for annelid and 8.4 for human and other mammalian genes) and more than 60% of annelid introns divide protein-coding sequences at exactly the same positions as human introns. By comparison, insects have 2.4 to 5.4 introns per gene and the plant representative Arabidopsis thaliana has 4.4, whereas fungi have the broadest spread from 0.0075 to 6.8 2. Thus, as far as shared introns are concerned, P. dumerilii is more similar to humans than to any insect or nematode.

The similarities in intron numbers and positions between P. dumerilii and humans does not imply that annelids should be combined with the vertebrates into a sister clade and distanced from nematodes and insects. Despite known uncertainties in the exact positioning of the segmented worms on the animal evolution tree (reviewed in 3), nobody has ever grouped annelids with vertebrates. Molecular evolution is an intricate nonlinear process that can be interpreted in many different ways and it cannot be inferred from a set of equations. Conflicting facts and opposing opinions are common in the field and several alternative phylogenetic trees have been proposed for the animal kingdom. In their short article, Raible et al. 1 present only one animal phylogeny, whereas a subsequent comment by Kumar and Hedges 4 in Cell points out other well-recognized possible alternative relationships between flies, worms and humans. There has been a long and fierce debate about the phylogenetic relationships between arthropods, nematodes and vertebrates 5. The 'Ecdysozoa hypothesis' groups arthropods and nematodes into a monophyletic clade and distances them from the vertebrates. The alternative 'Coelomata hypothesis' considers arthropods to be more closely related to the vertebrates than to the nematodes. We are still very far from resolving this dilemma. Even the whole-genome phylogenetic analyses of Drosophila, Caenorhabditis elegans and humans have not brought much clarity because, depending on the algorithms used, support can be found for both the Coelomata hypothesis 6 and the Ecdysozoa hypothesis 7. Undoubtedly, the results in the Science article by Raible et al. 1 will serve as important additional, yet non-decisive, evidence in these endless debates on the origin of species.

The new data of Raible et al. 1 further complicate our perception of evolution. We used to think that humans have significantly advanced in complexity compared to the various worm phyla. It is, however, unclear where this complexity is encoded within our genome. Homo sapiens has only 21% more protein-coding genes than the microscopic nematode worm Caenorhabditis elegans, whose entire neural system is composed of 300 neurons 8. Moreover, Raible et al. 1 have now shown that evolutionary changes in human protein sequences have occurred more slowly than in insects and nematodes. Thus, our proteins cannot be superior to those of invertebrates. Many biologists used to think that vertebrates had perfected their gene structures by acquiring thousands of new introns which, in turn, increased their protein diversity via alternative splicing. Yet according to Raible et al., the last common ancestor between vertebrates and annelids had nearly as many introns as humans. So, evolution has hardly affected the gene structure of the vertebrate lineage.

The last line of defense in our ambition for pre-eminence is the total size of our genomes. Indeed, the human haploid genome contains more than 3 billion nucleotides, or 3.5 picograms (pg) of DNA. According to the Animal Genome Size Database 9 this is several times more than the haploid genome size of the vast majority of invertebrates, including Drosophila melanogaster (0.18 pg), C. elegans (0.10 pg), and P. dumerilii (0.89 pg). There are, however, many exceptions to the correlation between an organism's morphological and functional complexity and the absolute size of its genome. Some invertebrates have a genome size comparable to that of humans 1011. The well-known example of the unicellular Amoeba dubia, with a genome 200 times larger than that of humans 9, demonstrates that mere DNA length does not determine an organism's complexity. The major fraction of an animal genome is represented by non-protein-coding sequences: for humans, the noncoding regions comprise 98.5% of the genome. For years, noncoding DNA (including introns) was largely ignored and was frequently referred to as 'junk' or 'selfish' DNA. But only a few years ago it became generally appreciated that three quarters of our genome is transcriptionally active and produces at least 16,000 non-protein-coding RNAs, many of which have important cellular functions 1213. Nevertheless, a considerable fraction of transcribed noncoding RNA from species with extra-large genomes probably represents random transcripts without any valuable role for the organism 14. Therefore, organismal complexity cannot be simply determined by the genome size, the number of protein-coding genes, the number of introns, or the total number of genomic transcripts. Rather, we should assess these parameters all together, and estimate how efficiently a particular species utilizes its genomic machinery, and the proportion of nonfunctional genomic ballast to functional elements.

Introns, for example, are ambivalent elements that create several problems for cells and perform various functions 1516. Jeffares et al. 17 have recently reviewed the process of intron gain and loss. They and others have demonstrated that intron evolution is not a simple stochastic process, and that a number of biological factors have important influences on change or preservation of exon-intron gene structures. Despite the fact that marine annelids and humans have the same number of introns, commonly in the same positions, the importance of these introns for the host cells could be different. Right now we only know for sure that human introns are longer than those of P. dumerilii and contain hundreds of noncoding RNAs as well as a number of valuable cellular signals 14. All in all, introns are not trivial elements, and the comparison of their positions in different species is only the first step in understanding their intricate evolution in animals and other eukaryotic taxa. Biologists have recently gained access to genomic information from dozens of eukaryotic and hundreds of prokaryotic species. This has only brought us to the embryonic stage of genome biology theory and numerous surprises are to be expected along the road ahead.