Deriving organismal phylogenies from morphological or behavioural data is a process which is open to error because of our lack of understanding of the mode and tempo of the evolution of these characters. Molecular data is often used as it allows the derivation of phylogenies (or measures of relatedness or difference) which are based on quantifiable data, and which derive from the genetic core which underlies organismal fitness. The molecular methods are not without their problems, however. The primary problem is assumption of homology: this is not unique to molecular data, but becomes critical when genes exist as a set of paralogues in the genomes under study. Three additional sources of problems occur:

1 robustness of the method. Several molecular methods rely on stochastic events which can result in non-linear accumulation of differences, or failure on repeated sampling. Methods which are hard to reproduce between laboratories include RAPD.

2 the existence of credible models for the accumulation of change over time. For some methods, for example sequence comparison, sophisticated models exist for the accumulation of changes over time. For others, such as RFLP, the differing rates for acquisition of restriction sites versus their loss are difficult to model

3 the rate of accumulation of differences (or loss of similarity). Some methods (eg SSCP, DDGE, VNTR, microsatellites) are only applicable to the study of individuals or populations which are separated for a short time (geologically/evolutionarily speaking). The accumulation of change in these sorts of markers is so rapid as to reach saturation within the lifetime of populations and sister species. For sequence data, regions of the genome can be chosen which have evolutionary rates commensurate with the evolutionary problem: highly conserved proteins for comparison of phyla, noncoding segmants for the comparison of populations.

For each method cost is an important consideration. Cost has two factors: the cost of material acquisition and storage (methods which require large amounts of tissues are less favoured than thos which work on a single blood spot), and the cost of the assay (in reagents and investigator time). Sequencing is uniformly the most costly per sample, but with the advent of PCR and automated sequencing is becoming more and more accessible.

Li and Graur Molecular Evolution

Hillis et al (eds) Molecular Systematics especially the "methods" chapters on PCR and DNA sequencing

or most any molecular biology methods book. I have appended brief descriptions below (follow the links)

molecular marker	robustness	cost	gene evolution	parentage	diversity	phylogeny to 5 MY	phylogeny to 50 My	phylogeny >500 MY
cytogenetics	yes	low	no	no	*	*	+/-	no
DNA hybridisation	yes	low	+/-	no	no	+/-	****	no
isozymes	yes	medium	+/-	+/-	+/-	***	+/-	no
RFLP	yes	medium	+/-	+/-	***	***	**	no
SSCP/DDGE	OK	medium	+/-	**	**	*	no	no
RAPD	so-so	low-medium	+/-	**	***	+/-	no	no
VNTR	yes	low-medium	+/-	***	***	+/-	no	no
microsatellite PCR	yes	low-medium	+/-	****	****	+/-	no	no
sequencing	yes	high	*****	*****	*****	*****	*****	*****

This technique involves using microscopy (often fluorescence microscopy) to examine the macroscopic architecture of chromosomes. The chromosomes can be identified by their number, size, form (position of centromere, heterochromatic regions) and by the banding pattern (either under normal light microscopy - eg Drosophilid polytene chromosomes, or after banding with fluorescent dyes).

DNA hybridisation

Single stranded DNA will reanneal (hybridise) with its sister strand. If two homologous DNA samples are mixed, annealing between complementary strands from the two samples is possible. The extent to which the DNAs of two species will hybridise to each other is a measure of the amount of overall sequence difference between the two genomes. The extent of hybridisation can be measuresd optically (double stranded DNA absorbs UV differently from single stranded) or by measurement of remaining single stranded DNA.

Isozymes

Isozymes are genetic variants of enzymes which have differing electrophoretic mobilities. The change in mobility is usually due to changes in overall charge following aminoacid substitution, though changes in size are also used. The measurement of differences is usually made using enzyme-specific colourimetric stains. The reliance on measurable changes in overall charge means that much variation can go unmeasured, but the method is cheap and robust, and easily applicable to any species.

RFLP Restriction Fragment Length Polymorphism

This method relies on the detection of sequence changes in DNA through the use of sequence specific restriction enzymes. Restriction enzymes cleave DNA at a target site, and changes in this site will abrogate cleavage. Sites can also be acquired through mutation. The sites are examined by either Southern blotting (a hybrisidation detection method) or by restriction digestion of PCR amplified gene fragments. The method requires knowledge of the variation in the genomes to be examined and can be costly in terms of material and reagents.

SSCP/DGGE Single strand conformation polymorphism/Denaturing gradient gel electrophoresis

These methods are used to detect single base changes in DNA fragments, and can be used when the changes do not happen to affect a restriction enzyme site. A fragment of known sequence is annealed with a test fragment, and the resulting hybrids analysed by gel electrophoresis systems which exaggerate the differences in mobility and stability of heteroduplexes compared to homoduplexes.

RAPD Randomly amplified polymorphic DNA

By using short (usually 10 base pair) primers in a PCR reaction with genomic DNA as target, it is possible to sample many sites in the genome at once. Stochastically, two sites for the single primer used will exist in close proximity on the genome and allow amplification of an intervening fragment. For each primer a variable number (0-very many) such sites will exist and a variable number of different sized fragments will be produced. These can be analysed by electrophoresis. The multiplicity of sites sampled means that different individuals can differ in the fragments amplified. The problems with this technique are (1) reliability (small changes in primer concentration, salt concentration, annealing temperature and target integrity can result in differing patterns) and (2) the possibility of fragments from distinct amplicons comigrating, or of different amplicons covering the same fragment not comigrating makes analysis of banding patterns for phylogenetics (rather than just taxonomy) problematic.

VNTR Variable number tandem repeats

This PCR-based method uses primers in nonrepetitive flanking DNA to amplify short (5-10 base pair) repeats from a known locus. These repeats can vary in number, and the different numbers of repeats in the two alleles at each locus of an individual genome can be measured by gel electrophoresis.

Microsatellite PCR

Micro sattelites are short (2-4 base pair) repeats which vary in number (due to slippage during DNA synthesis). They are often highly variable and are useful as molecular "barcodes" to follow alleles through populations, as markers for genetic mapping and as tags to analyse parentage. The amplified markers are analysed by gel electrophoresis: detection of the alleles (which differ by as little as 2 base pairs) is through chemical staining, radioactive tracer marking or fluorescent tracer marking.

Sequencing

Both DNA and RNA molecules can be sequenced. The method most commonly used is chain termination sequencing of DNA, as this can be performed using radioactive or nonradioactive methods. DNA fragments to be sequenced can be cloned, or an uncloned amplified fragment, or (if technique is excellent) the locus in situ in the genome of interest. To follow evolutionary change on different timescales, segments of the genome with different rates of acceptance of mutation can be sampled.