This is a large question that has been very controversial, but as more and more animals are studied using an increasing number of different molecules, confidence in molecular taxonomy is growing fast also. An indication of the advantages and disadvantages of using molecules may still be useful.
1. Hie equivalence of data, since the nature and position of the unit is precisely defined.
2. Hie enormous size of the data set.
3. Statistical analysis of cladograms, avoiding the pitfalls of parsimony, is possible and only awaits agreement on the statistical methods to be used.
4. Where change in a molecule is rare, as in genes coding for ribosomal RNA, it becomes possible to trace relationships far back in time.
5. Non-heritable variation is avoided.
1. The underlying assumption for most methods is that change in a gene molecule will depend only on the mutation rate and the time elapsed, i.e. that an unvarying ‘molecular clock’ is ticking at a regular rate. However, the clock is known to be variable in certain conditions, and the whole idea of functionally neutral changes in genes is controversial. Some branches of the evolutionary tree are known to evolve very fast: should we compensate by a subjective decision to omit such species (or groups of species) from our calculations? This has frequently been necessary to obtain results from ribosomal genes.
2. There is no record of past changes in characters. This is a serious disadvantage, as there are only four possible nucleotides for any site in the DNA molecule. If there have been changes from one nucleotide to another and back again, such ‘multiple hits’ cannot be detected. This also is primarily a problem with ribosomal genes.
3. There is no recognizable intermediate condition between characters and, worse, no primitive condition for a given site can be recognized.
4. Functional correlates of character change can very seldom be traced.
5. It is very difficult indeed to root a tree derived from molecules; sequence similarity is the only guide, and the likelihood of convergence is usually impossible to assess.
For all these reasons, the need to use several (ideally many) different genes is apparent.