PCR-RFLP uses endonucleases to digest PCR products of regions with sequence polymorphisms. By using an endonuclease which recognizes and cleaves at the polymorphic sites, the digestion of a longer PCR fragment into smaller fragments will change the banding pattern. PCR-RFLP has been used for authentication of Panax species 14 35 36 , Fritillaria pallidiflora 37 , Atractylodes species 38 and differentiation of Codonopsis from their adulterants 39 .

Screening PCR products by using various restriction enzymes can be an alternative of sequencing to find out polymorphic regions among samples 34 40 . This method is more reproducible than random priming methods, but it is limited by the degree of polymorphism among individuals within a species 25 . The loss of restriction sites associated with degraded DNA, creation or deletion of restriction sites due to intra-specific variation 34 , and the presence of enzyme inhibitors may lead to incomplete digestion.

AFLP involves restriction of genomic DNA and ligation to adapters, selective amplification of restriction fragments using primers containing the adapter sequences and selective bases at the 3' terminals, and subsequent gel analysis of the amplified fragments 41 . The number of resulting fragments in this multi-locus approach is determined by the number and composition of selective nucleotides as well as the complexity of the genomes. Polymorphisms detected may be caused by a single nucleotide change at the restriction site or 3' end of the primer binding site, insertions and deletions as well as rearrangements. It has been used for authentication 42 and studying genetic diversity 43 44 45 46 of Chinese medicines.

AFLP combines the advantages of the reliability of RFLP and the power of PCR. The high reproducibility resulting from stringent reaction conditions enables the use of polymorphic bands in developing cultivar-specific probes, which can then be used for easy identification 47 . By using adapter sequences, fingerprints can be generated without prior sequence knowledge. The number of amplified fragments can be controlled by changing the restriction enzymes and the number of selective bases and thus this method is suitable for DNA of any origin and complexity. It is efficient in revealing polymorphisms even between closely related individuals 48 . The number of polymorphisms per reaction can be higher than RFLP or RAPD 47 49 . Three steps and four different primers are required for analysis of complex genomes 48 . Imperfect ligation or incomplete restriction of DNA will lead to artifactual polymorphisms because of partial fragments 48 . Moreover, if the sequence homology between two organisms is less than 90%, their fingerprints will share very few common fragments 50 .

RP-PCR involves amplification at low annealing temperatures using one or two random primers in each PCR reaction to generate unique fingerprints (Figure 1). At reduced stringency, the arbitrary primers bind to a number of sites randomly distributed in the genomic DNA template although the primer and the template sequences may not be perfectly matched. Each anonymous and reproducible fragment is derived from a region of genome that contains, on opposite DNA strands, two primer binding sites located within an amplifiable distance from each other. Polymorphisms are resulted from sequence differences which inhibit primer binding or interfere with amplification. The presence or absence of bands is scored and the results can be used for calculating genetic distances 51 52 53 and constructing phylogenetic trees 18 54 to study the relationships among samples.

Techniques based on this concept include arbitrarily primed PCR (AP-PCR) 55 , random amplified polymorphic DNA (RAPD) 56 and DNA amplification fingerprinting (DAF) 57 . AP-PCR employs primers of approximately 20 nucleotides long and only two relaxed PCR cycles while RAPD employs primers of 10 nucleotides long and all the PCR cycles are relaxed, and DAF employs primers of 5–8 nucleotides long and all the PCR cycles are relaxed with lower stringency than that of RAPD. AP-PCR results are the most reproducible among these three methods.

As any part of the genome, including non-coding regions, may be amplified, these methods can be used to discriminate between closely related individuals. RP-PCR have been used for marker-assisted selection in breeding 58 , identification at the individual, variety, strain and species levels 51 59 60 61 62 , study of genetic diversity 63 64 and differentiation of cultivated and wild samples 18 65 66 . It has been used to authenticate Chinese medicines 51 52 67 68 69 and identify their geographical origins 16 18 53 70 71 .

RP-PCR is a quick and easy method to screen a large number of loci for DNA polymorphisms in a single PCR. Polymorphic markers can be generated rapidly without sequence information. The marker sequences obtained from RP-PCR can also be used to design specific oligonucleotides to be used in SCAR assay 72 . Moreover, any single primer can be used, including those for specific PCR amplification or sequencing. However, there are some limitations in RP-PCR. Firstly, it is sensitive to the reaction conditions, including the amounts of templates 73 74 75 and magnesium ions 76 77 , the sequences of primers 78 79 80 81 , the presence of glycerol 82 and the quantity and quality of the polymerase 83 . Thermocyclers may also influence the banding patterns due to their different ramp time from the annealing step to the extension step 84 85 . Therefore, results generated with different reaction conditions cannot be compared directly. Secondly, the reproducibility of RP-PCR patterns can be influenced by the quality, quantity and purity of DNA templates 86 87 . Some researchers consider this to have a more significant effect to reproducibility than other factors such as enzyme quality or buffer conditions which may affect only the relative intensities of bands, but do not cause a band to appear or disappear 88 . The number of bands produced can be greatly increased by adding bovine serum albumin (BSA), which prevents various contaminants in impure DNA from binding to Taq polymerase 89 thereby stabilizing the polymerase 90 . This method is, therefore, unreliable for identification of medicinal materials with impure or degraded DNA 91 caused by processing or long storage time. Amplification should be performed using DNA templates at two concentrations with at least two-fold difference 92 to ensure reproducibility of the results. Thirdly, as arbitrary primers and low stringencies are used, DNA from any organisms including DNA from contaminants can be amplified, thereby contributing to the banding pattern. Fourthly, as different loci in the genome have different degree of homology among samples (e.g. coding regions are more conserved), the number of loci to be included should be large enough to reveal differences in the whole genome. Fifthly, fragments of the same size in the fingerprints are not necessarily the same sequence 93 . It should not be assumed that bands of a similar size are homologous sequences, especially when distant species are being examined. The extent of polymorphism is indicated by the presence and absence of bands of particular sizes and thus applying RP-PCR on high taxonomic levels leads to an increase in variance of genetic distance estimates.

DALP uses a selective forward primer containing a 5' core sequence (e.g. M13 universal sequencing primer) plus additional bases at the 3' end and a common reverse primer (e.g. M13 reverse primer) to generate multibanded patterns in denaturing polyacrylamide gel 48 . For identification of fragments having two different ends, two PCR reactions are performed for each sample, with each reaction containing either of the primers labeled. The common bands in both labeled reactions are the products having two different ends. Any of these bands can be excised from the gel and sequenced directly using forward or reverse primers. After sequencing the polymorphic bands among the samples, species- or strain-specific primers can be designed. These specific primers can then be used for mono-locus amplification (i.e. sequence-tagged site) (Figure 2). DALP is similar to AP-PCR except that it uses higher stringency (i.e. higher annealing temperature, lower concentration of magnesium ion and fewer cycles). At this stringency, a given arbitrary primer will anneal to fixed sites across experiments 48 . Moreover, the polymorphic bands can be sequenced directly to further design species- or strain-specific primers, whereas in AP-PCR the two ends of the products contain the same primer sequence. For this reason, AP-PCR products cannot be sequenced directly (Figure 1). DALP has been used to detect polymorphisms between species 48 94 and between strains 48 and to authenticate Panax ginseng (Renshen) and Panax quinquefolius (Xiyangshen) 95 . It has also been used in mapping for marker-assisted selection in breeding 96 .

Identification of polymorphic products can be done easily by searching in public sequence databases. However, each sample must be subject to two separate PCR reactions by alternative labeling of the two primers. This requires an additional step apart from identifying polymorphic bands among samples. Unlike RP-PCR, much time and effort are required in screening for suitable primer pairs and optimizing primer ratio 97 .

ISSR 98 employs a primer containing simple repeat sequences for PCR amplification to generate fingerprints (Figure 3). The primer can be 5' or 3' anchored by selective nucleotides to prevent internal priming of the primer and to amplify only a subset of the targeted inter-repeat regions, thereby reducing the number of bands produced by priming of dinucleotide inter-repeat region 98 99 100 . Therefore regions between inversely oriented closely spaced microsatellites are amplified 101 . This method relies on the existence of 'SSR hot spots' in genomes 98 101 . It has been used in the authentication of Dendrobium officinale (Tiepi Shihu) 102 and in the studies of genetic variations and relationships 103 104 105 106 .

As simple sequence repeats exist in any genome, this method allows fingerprints to be generated for any organism. However, the primers may anneal to sequences other than microsatellites of the desired repeats as in RAPD 99 . Moreover, the banding patterns can be affected by magnesium ion concentration, thermocycler and annealing temperature in use 102 . Compared to SSR markers, ISSR markers are not locus-specific and anonymous bands are produced in the fingerprints 101 .

Minisatellites are also known as variable number of tandem repeats (VNTR). They are similar to microsatellites except that the repeat unit sequence is longer than 10 bp (the distinction between microsatellites and minisatellites is often arbitrary with repeat units between 8 and 15 bp 107 ). DAMD 108 is a DNA fingerprinting method based on amplification of the regions rich in minisatellites at relatively high stringencies by using previously found VNTR core sequences as primers (Figure 1). It is employed in identification of species-specific sequences 109 . The method has been used for authentication of Panax ginseng (Renshen) and Panax quinquefolius (Xiyangshen) 42 and characterization of varieties 78 and cell lines 110 .

Similar to SSR, minisatellites are present in all organisms, making it possible to apply this method to any genome. Examples include plants such as mulberry 78 and oilseed 109 , mushrooms Agaricus bisporus and Pleurotus 111 , animals such as human, snake and mouse 110 , insects such as mosquito and moth. This method is more reproducible than RAPD due to the longer primers used 111 . However, similar to ISSR, it only reveals polymorphisms in the regions rich in repetitive sequences, while RP-PCR can represent the polymorphisms of the whole genome 78 . Moreover, polymorphic sequence must be obtained by cloning polymorphic bands prior to sequencing because of the same primer sequences at both ends of the PCR product.