**4. PCR performance**

To study the effects of the Taq polymerase inhibitors, we added decrease quantities of the DNA extracted by the phenol-chloroform-isoamyl alcohol technique (shown in Figure 1) to the amplification reaction of the hypervariable segment I (15975-16420) using contemporary DNA. Figure 4, lane 1 displays the 100 bp molecular weight marker, lane 2 the negative control, lane 3 the PCR product of the contemporary DNA with the aDNA without dilution, lanes 4, 5, 6 and 7 show aDNA diluted 1:1, 1:2, 1:4 and 1:8, respectively, added to the PCR reaction mix and lane 8 contemporary DNA without any addition (positive control). Contaminated DNA allowed positive amplifications when aDNA was diluted at least 1:4 (Figure 4, lane 6). Therefore, an additional method to obtain the PCR product from the aDNA is by sample dilution.

Because DNA concentration is only possible with limited precision and concentrations of standard dilution series change over time in storage, we evaluated the relative performance of the DNA during PCR amplification using serial dilutions of the extracted DNA starting from 5 μl. Using this method, we were able to dilute the inhibitors of the Taq DNA

Fig. 3. Extraction of DNA of pre-Hispanic samples by the silica technique. Lanes 1 and 10, molecular weight markers of 23 kbp and 100 bp, respectively. Lanes 2 to 9, DNA extracted

Silica gel was also used to purify aDNA, results are shown in Figure 3. Each lane of this figure (2-9) displays aDNA extracted from 0.25 g of different pre-Hispanic bone samples. Lanes 1 and 10 show molecular weight markers. The quantity of Taq polymerase inhibitors is not evident, although we know that all ancient samples contain some of these inhibitors in

To study the effects of the Taq polymerase inhibitors, we added decrease quantities of the DNA extracted by the phenol-chloroform-isoamyl alcohol technique (shown in Figure 1) to the amplification reaction of the hypervariable segment I (15975-16420) using contemporary DNA. Figure 4, lane 1 displays the 100 bp molecular weight marker, lane 2 the negative control, lane 3 the PCR product of the contemporary DNA with the aDNA without dilution, lanes 4, 5, 6 and 7 show aDNA diluted 1:1, 1:2, 1:4 and 1:8, respectively, added to the PCR reaction mix and lane 8 contemporary DNA without any addition (positive control). Contaminated DNA allowed positive amplifications when aDNA was diluted at least 1:4 (Figure 4, lane 6). Therefore, an additional method to obtain the PCR product from the

from different samples of Mexican pre-Hispanic populations.

different concentrations.

**4. PCR performance** 

aDNA is by sample dilution.

polymerase.

Fig. 4. Inhibition of the mitochondrial DNA hypervariable segment I amplification via inhibition of Taq polymerase by aDNA contaminants.

Fig. 5. Positive effect of bovine serum albumin (BSA) on the PCR performance of aDNA extracted by the phenol-chloroform-isoamyl alcohol procedure.

Sometimes aDNA dilution is not enough to obtain the PCR products, so we tested the effect of BSA addition by increasing the concentration of BSA in the reaction from 0.1 to 0.25 mg/ml. Figure 5 shows the positive effect of BSA on the PCR of aDNA extracted by the phenol-chloroform-isoamyl alcohol procedure. Increased amplification was observed in the PCR experiments when BSA was added in increasing concentrations (Figure 5, lanes 2-7, BSA at a concentration of 0.05, 0.1, 0.15, 0.2, 0.25 and 0.3 mg/ml, lane 8, negative control and

Extraction and Electrophoresis of DNA from the Remains of Mexican Ancient Populations 489

The HVI mtDNA segment of 445 bp was amplified in Figure 6, panels A and B with the primers L15975-15996 and H16401-16420. When the primers L15975-15996 and H16228- 16248 were used, the PCR product is shorter (273 bp), and the presence of PVP makes evident the PCR fragment at a dilution of 1:16 (Figure 6, panel C). In addition, PVP at a concentration of 0.4% in the PCR experiment did not inhibit amplification, as was previously published (Young et al., 1993). The molecular weight marker is in lane 1; positive control with contemporary DNA alone is in lane 11; and the negative control with no DNA is in lane 12. Nevertheless, the positive effect was not evident in all aDNA bone samples,

Contemporary DNA is very easy to amplify. However, when working with aDNA, the PCR reaction efficiency is greatly reduced. For example, in Figure 7, we show a PCR-amplified fragment of mDNA using the human-specific primers L15975-15996 and H16236-16255 where 4 of the 7 bone samples from the pre-Hispanic populations displayed the PCR product (Figure 7, lanes 2 to 8). Positive and negative controls are shown in lanes 9 and 10,

Fig. 7. Agarose gel showing the amounts of PCR-amplified product obtained after DNA

Using the procedure indicated in the methods section, we purified and amplified the DNA from 14 bone samples of pre-Hispanic Native Americans to type them for haplogroup A described for Amerindians. The PCR products obtained were digested by the restriction enzyme *HaeIII*. Haplogroup A was detected in the 10 samples typed (Figure 8, lanes 2-11). Partial restriction digestion was observed in all of the ancient PCR products. This finding suggests the presence of the Amerindian polymorphism; however, we must sequence these amplification fragments or use real-time PCR to confirm the presence of the specific

extraction with the phenol-chloroform-isoamyl alcohol procedure.

polymorphism because the partial restriction observed.

likely because the amount of Taq polymerase inhibitors is different in each sample.

respectively, and the molecular weight marker is shown in lane 1.

lane 9, positive control). Based on these results, we added 0.25 mg/ml BSA to all PCR experiments.

Fig. 6. Inhibition of the mitochondrial DNA hypervariable segment I amplification by inhibition of the Taq polymerase with aDNA contaminants and the effect of PVP.

Soils with high organic contents have humic acids with phenolic groups that denature biological molecules by bonding to N-substituted amides or oxidise to form a quinone that bonds to DNA or proteins (Young et al., 1994). Because aDNA contains these Taq polymerase inhibitors from soil, we tested the effect of Polyvinylpyrrolidone (PVP) during DNA extraction as has been suggested previously (Young et al., 1994; Rohland and Hofreiter, 2007). In addition, to make sure that PVP did not inhibit the PCR experiment, the reagent was added directly to the amplification mix. Figure 6 shows that 2% PVP added during the DNA extraction had a positive effect on DNA amplification. PCR amplification of contemporary DNA containing different dilutions of aDNA (1:1, 1:2, 1:3, 1:4, 1:8, 1:16, 1:32, 1:64, 1:128; Figure 6, lanes 3 to 10, respectively) and 0.25 mg/ml BSA is shown in Figure 6, panel A, and amplifications using the same conditions in the presence of 2% PVP during aDNA extraction is shown in Figure 6, panel B. The use of 2% PVP during aDNA extraction resulted in amplification at an aDNA dilution of 1:64, in contrast to the aDNA sample without PVP in which amplification is only observed at 1:128 dilution or beyond.

lane 9, positive control). Based on these results, we added 0.25 mg/ml BSA to all PCR

Fig. 6. Inhibition of the mitochondrial DNA hypervariable segment I amplification by inhibition of the Taq polymerase with aDNA contaminants and the effect of PVP.

Soils with high organic contents have humic acids with phenolic groups that denature biological molecules by bonding to N-substituted amides or oxidise to form a quinone that bonds to DNA or proteins (Young et al., 1994). Because aDNA contains these Taq polymerase inhibitors from soil, we tested the effect of Polyvinylpyrrolidone (PVP) during DNA extraction as has been suggested previously (Young et al., 1994; Rohland and Hofreiter, 2007). In addition, to make sure that PVP did not inhibit the PCR experiment, the reagent was added directly to the amplification mix. Figure 6 shows that 2% PVP added during the DNA extraction had a positive effect on DNA amplification. PCR amplification of contemporary DNA containing different dilutions of aDNA (1:1, 1:2, 1:3, 1:4, 1:8, 1:16, 1:32, 1:64, 1:128; Figure 6, lanes 3 to 10, respectively) and 0.25 mg/ml BSA is shown in Figure 6, panel A, and amplifications using the same conditions in the presence of 2% PVP during aDNA extraction is shown in Figure 6, panel B. The use of 2% PVP during aDNA extraction resulted in amplification at an aDNA dilution of 1:64, in contrast to the aDNA sample without PVP in which amplification is only observed at 1:128 dilution or beyond.

experiments.

The HVI mtDNA segment of 445 bp was amplified in Figure 6, panels A and B with the primers L15975-15996 and H16401-16420. When the primers L15975-15996 and H16228- 16248 were used, the PCR product is shorter (273 bp), and the presence of PVP makes evident the PCR fragment at a dilution of 1:16 (Figure 6, panel C). In addition, PVP at a concentration of 0.4% in the PCR experiment did not inhibit amplification, as was previously published (Young et al., 1993). The molecular weight marker is in lane 1; positive control with contemporary DNA alone is in lane 11; and the negative control with no DNA is in lane 12. Nevertheless, the positive effect was not evident in all aDNA bone samples, likely because the amount of Taq polymerase inhibitors is different in each sample.

Contemporary DNA is very easy to amplify. However, when working with aDNA, the PCR reaction efficiency is greatly reduced. For example, in Figure 7, we show a PCR-amplified fragment of mDNA using the human-specific primers L15975-15996 and H16236-16255 where 4 of the 7 bone samples from the pre-Hispanic populations displayed the PCR product (Figure 7, lanes 2 to 8). Positive and negative controls are shown in lanes 9 and 10, respectively, and the molecular weight marker is shown in lane 1.

Fig. 7. Agarose gel showing the amounts of PCR-amplified product obtained after DNA extraction with the phenol-chloroform-isoamyl alcohol procedure.

Using the procedure indicated in the methods section, we purified and amplified the DNA from 14 bone samples of pre-Hispanic Native Americans to type them for haplogroup A described for Amerindians. The PCR products obtained were digested by the restriction enzyme *HaeIII*. Haplogroup A was detected in the 10 samples typed (Figure 8, lanes 2-11). Partial restriction digestion was observed in all of the ancient PCR products. This finding suggests the presence of the Amerindian polymorphism; however, we must sequence these amplification fragments or use real-time PCR to confirm the presence of the specific polymorphism because the partial restriction observed.

Extraction and Electrophoresis of DNA from the Remains of Mexican Ancient Populations 491

Ancient DNA was added to the PCR mix without any dilution (lane 2), diluted 1:10 (lane 3); aDNA phenol-chloroform-isoamyl alcohol extracted (lane 4); same procedure but diluted 1:10 (lane 5); washing buffer of the Centrifugal filter units (Centricon®) that were used to purify and concentrate aDNA in the phenol-chloroform-isoamyl alcohol purification procedure (lane 6) and diluted 1:10 (lane 7); negative control without DNA (lane 8); positive

Fig. 9. Amplification of mtDNA HVI segment of 410 bp with specific primers (L15975-15996

chloroform-isoamyl alcohol procedure followed by treatment with Chelex. Lanes 1 and 10, molecular weight markers φ. Lanes 2-4 pre-ceramic 7 diluted 1:10; 1:20 and 1:30; lanes 1-7, pre-ceramic 9 same dilutions as in lanes 2-4; lanes 8-10, pre-Ceramic 10, same dilutions as

When extracting DNA from small, degraded forensic samples or degraded ancient samples, the final concentration of DNA is usually too low for subsequent amplification. Consequently, we concentrated the aDNA samples extracted by the phenol-chloroformisoamyl alcohol with filter units (lanes 2-3 and 4-5). Figure 10 shows clearly how aDNA was amplified using the DNA that was concentrated with the filter units. Furthermore, the washing buffer did not show any amplification, confirming in part that we do not have

Next, we wanted to test all these methods with tissue from different mummies and determine the differences in using internal tissue and skin. From our results, we observe that when the mummy tissue is compact and from an internal organ the quantity of aDNA is very high compared with that obtained from bone samples. In addition the aDNA from the internal tissue was better as far as content is concerned. We had the opportunity to obtain DNA from the internal tissue of the mummy called Pepita that was intact and had no contamination by contemporary DNA. We were able to amplify the HVI segment using the specific primers for a mtDNA fragment of 445 bp (L15975-15996 and

and H16380-16398) using aDNA from pre-ceramic samples extracted with phenol-

**410 bp**

control with contemporary DNA (lane 9); and no sample (lane 10).

**1353**

**bp**

**603 310**

lanes 2-4.

DNA contaminating our assays.

Fig. 8. Agarose gel showing the amounts of PCR-amplified product obtained after DNA extraction with the phenol-chloroform-isoamyl alcohol procedure and digested with the restriction enzyme *HaeIII*. Primers for amplification were specific to type haplogroup A.

Amplification of DNA extracted by Chelex was tested in the samples from pre-ceramic bones (Figure 9). Although experiments that compared the phenol-chloroform and Chelex method concluded that the Chelex method was simple and fast, inhibitory substances had not been eliminated in most of the cases (Kalmár et al., 2002). In our experience, DNA extracted by the phenol-chloroform method followed by Chelex treatment may improve DNA purification. Nevertheless, the silica method was better overall in our experience. The amplification products are observed at DNA dilutions of 1:30 in all samples, as shown in Figure 9.

Figure 10 displays the PCR amplification fragments using the specific primers L15975-15996 and H16236-16255 producing a fragment of 281 bp in panel A and L16140-16159 and H16380-16398 producing a fragment of 259 bp in panel B. We compared the amplification of aDNA extracted by the silica procedure and phenol-chloroform-isoamyl alcohol. Our results showed that aDNA extraction with the silica procedure was better than the phenolchloroform-isoamyl alcohol method in this specific sample from pre-Hispanic populations because the amplification was observed exclusively in samples in which the DNA was extracted by the silica method. However, this may not be the case for all types of samples, and it is important to consider that when one method does not give good results, other methods may be useful.

Figure 10 shows aDNA from pre-Hispanic samples extracted by the silica gel method (lanes 2 and 3) compared with the phenol-chloroform-isoamyl alcohol procedure followed by concentration of aDNA with filter units (Centricon®) (lanes 4-5).

Fig. 8. Agarose gel showing the amounts of PCR-amplified product obtained after DNA extraction with the phenol-chloroform-isoamyl alcohol procedure and digested with the restriction enzyme *HaeIII*. Primers for amplification were specific to type haplogroup A.

Figure 9.

methods may be useful.

Amplification of DNA extracted by Chelex was tested in the samples from pre-ceramic bones (Figure 9). Although experiments that compared the phenol-chloroform and Chelex method concluded that the Chelex method was simple and fast, inhibitory substances had not been eliminated in most of the cases (Kalmár et al., 2002). In our experience, DNA extracted by the phenol-chloroform method followed by Chelex treatment may improve DNA purification. Nevertheless, the silica method was better overall in our experience. The amplification products are observed at DNA dilutions of 1:30 in all samples, as shown in

Figure 10 displays the PCR amplification fragments using the specific primers L15975-15996 and H16236-16255 producing a fragment of 281 bp in panel A and L16140-16159 and H16380-16398 producing a fragment of 259 bp in panel B. We compared the amplification of aDNA extracted by the silica procedure and phenol-chloroform-isoamyl alcohol. Our results showed that aDNA extraction with the silica procedure was better than the phenolchloroform-isoamyl alcohol method in this specific sample from pre-Hispanic populations because the amplification was observed exclusively in samples in which the DNA was extracted by the silica method. However, this may not be the case for all types of samples, and it is important to consider that when one method does not give good results, other

Figure 10 shows aDNA from pre-Hispanic samples extracted by the silica gel method (lanes 2 and 3) compared with the phenol-chloroform-isoamyl alcohol procedure followed by

concentration of aDNA with filter units (Centricon®) (lanes 4-5).

Ancient DNA was added to the PCR mix without any dilution (lane 2), diluted 1:10 (lane 3); aDNA phenol-chloroform-isoamyl alcohol extracted (lane 4); same procedure but diluted 1:10 (lane 5); washing buffer of the Centrifugal filter units (Centricon®) that were used to purify and concentrate aDNA in the phenol-chloroform-isoamyl alcohol purification procedure (lane 6) and diluted 1:10 (lane 7); negative control without DNA (lane 8); positive control with contemporary DNA (lane 9); and no sample (lane 10).

Fig. 9. Amplification of mtDNA HVI segment of 410 bp with specific primers (L15975-15996 and H16380-16398) using aDNA from pre-ceramic samples extracted with phenolchloroform-isoamyl alcohol procedure followed by treatment with Chelex. Lanes 1 and 10, molecular weight markers φ. Lanes 2-4 pre-ceramic 7 diluted 1:10; 1:20 and 1:30; lanes 1-7, pre-ceramic 9 same dilutions as in lanes 2-4; lanes 8-10, pre-Ceramic 10, same dilutions as lanes 2-4.

When extracting DNA from small, degraded forensic samples or degraded ancient samples, the final concentration of DNA is usually too low for subsequent amplification. Consequently, we concentrated the aDNA samples extracted by the phenol-chloroformisoamyl alcohol with filter units (lanes 2-3 and 4-5). Figure 10 shows clearly how aDNA was amplified using the DNA that was concentrated with the filter units. Furthermore, the washing buffer did not show any amplification, confirming in part that we do not have DNA contaminating our assays.

Next, we wanted to test all these methods with tissue from different mummies and determine the differences in using internal tissue and skin. From our results, we observe that when the mummy tissue is compact and from an internal organ the quantity of aDNA is very high compared with that obtained from bone samples. In addition the aDNA from the internal tissue was better as far as content is concerned. We had the opportunity to obtain DNA from the internal tissue of the mummy called Pepita that was intact and had no contamination by contemporary DNA. We were able to amplify the HVI segment using the specific primers for a mtDNA fragment of 445 bp (L15975-15996 and

Extraction and Electrophoresis of DNA from the Remains of Mexican Ancient Populations 493

analyses. Sequences from Monte Alban and Teotihuacán from this study as well as those

from the Oneota population were clustered in the haplogroup D linage.

Fig. 11. Amplification of aDNA extracted from a Mexican mummy.

the time.

Haplotype network analyses were carried out on 290 bp of the mtDNA HVI from nucleotides 16104 to 16394. These networks were constructed using the Network package, v4.5.1.0 (Fluxus Engineering). These analyses included sequences from our own work and from other authors. The accession numbers of the sequences included in this network analyses were mentioned in the data analysis section. The pre-Hispanic DNA sequences included two ancient sequences from the prehistoric Oneota population (Stone and Stoneking, 1998), two sequences from Monte Alban, Oaxaca, Mexico and one from Teotihuacán, Mexico. Interestingly, the haplotype from the Oneota sequence may be derived from the Teotihuacán haplotype. The sequences from Monte Alban were grouped in the same haplotype as the more frequent haplotype from Native American populations. These results showed the potential to know the relationship among all Mexican pre-Hispanic populations or other populations as well as some haplotypes that were lost through

It is important to mention that we never observed contaminant fragments with the specific HVR-1 mutations carried by the excavators or the geneticists. Therefore we are confident that following the procedures recommended by previous authors and our laboratory generates authentic sequences. Problems arise when the samples come

H16401-16420, Figure 11, lane 8) and a fragment of 281 bp (L15975-15996 and H16236- 16255, Figure 11, lane 1) or to amplify the specific second segment of HVI (L16140- 16159 and H16401-16420, lane 5). The aDNA from this mummy was very well conserved. We have previously published DNA extraction from Mexican mummies with different origin and age (López-Armenta et al., 2008; Bustos-Ríos et al., 2008; Herrera-Salazar et al., 2008).

Fig. 10. Amplification of mtDNA HVI segments with the specific primers L15975-15996 and H16236-16255 producing a fragment of 281 bp (A); and L16140-16159 and H16380-16398 producing a fragment of 259 bp (B).

To examine the relationships between mtDNA lineages found in ancient and contemporary Native Americans, phylogenetic trees were constructed with the Jukes-Cantor method, and the distances were obtained from a neighbour joining algorithm and optimised for maximum likelihood, using Hy-Phy software (Kosakovsky-Pond et al., 2005). A total of 290 bp (nucleotides 16104–16394) of the HVI common to all sequences were used for these

H16401-16420, Figure 11, lane 8) and a fragment of 281 bp (L15975-15996 and H16236- 16255, Figure 11, lane 1) or to amplify the specific second segment of HVI (L16140- 16159 and H16401-16420, lane 5). The aDNA from this mummy was very well conserved. We have previously published DNA extraction from Mexican mummies with different origin and age (López-Armenta et al., 2008; Bustos-Ríos et al., 2008; Herrera-Salazar et al.,

**1 2 3 4 5 6 7 8 9 10**

**1 2 3 4 5 6 7 8 9 10**

Fig. 10. Amplification of mtDNA HVI segments with the specific primers L15975-15996 and H16236-16255 producing a fragment of 281 bp (A); and L16140-16159 and H16380-16398

To examine the relationships between mtDNA lineages found in ancient and contemporary Native Americans, phylogenetic trees were constructed with the Jukes-Cantor method, and the distances were obtained from a neighbour joining algorithm and optimised for maximum likelihood, using Hy-Phy software (Kosakovsky-Pond et al., 2005). A total of 290 bp (nucleotides 16104–16394) of the HVI common to all sequences were used for these

**281 bp** 

**259 bp** 

2008).

A

**500**

**100**

producing a fragment of 259 bp (B).

**1000** 

**bp**

**500**

**100**

B

**1000** 

**bp**

analyses. Sequences from Monte Alban and Teotihuacán from this study as well as those from the Oneota population were clustered in the haplogroup D linage.

Fig. 11. Amplification of aDNA extracted from a Mexican mummy.

Haplotype network analyses were carried out on 290 bp of the mtDNA HVI from nucleotides 16104 to 16394. These networks were constructed using the Network package, v4.5.1.0 (Fluxus Engineering). These analyses included sequences from our own work and from other authors. The accession numbers of the sequences included in this network analyses were mentioned in the data analysis section. The pre-Hispanic DNA sequences included two ancient sequences from the prehistoric Oneota population (Stone and Stoneking, 1998), two sequences from Monte Alban, Oaxaca, Mexico and one from Teotihuacán, Mexico. Interestingly, the haplotype from the Oneota sequence may be derived from the Teotihuacán haplotype. The sequences from Monte Alban were grouped in the same haplotype as the more frequent haplotype from Native American populations. These results showed the potential to know the relationship among all Mexican pre-Hispanic populations or other populations as well as some haplotypes that were lost through the time.

It is important to mention that we never observed contaminant fragments with the specific HVR-1 mutations carried by the excavators or the geneticists. Therefore we are confident that following the procedures recommended by previous authors and our laboratory generates authentic sequences. Problems arise when the samples come

Extraction and Electrophoresis of DNA from the Remains of Mexican Ancient Populations 495

Fig. 13. Haplotype network of American native populations including five sequences from samples of pre-Hispanic populations. Each haplotype is represented by a circle in which the square radius (surface) is proportional to its population frequency. Circle colours show the site location as represented by the indicated colours. Dark circles without number in the

unrepresented in the sample). Numbers between haplotypes represent mutational steps.

This review offers a direct overview of the different methods of aDNA extraction, including all special conditions needed in the laboratory to avoid contamination by contemporary DNA. It reveals the complexity involved in demonstrating the authenticity of human aDNA because the risk of contamination is very high. However, exogenous DNA contamination can be avoided if the necessary care is taken. In our experience and the experience of other laboratories, obtaining the ancient sample with coat, gloves and mask, and maintaining it in sterile conditions without human contact reduces the chances of sample contamination. It is also very important to test all reagents to verify that they are free of contemporary DNA. In addition, we also recommend performing negative control PCR experiments with at least 45 cycles to convincingly demonstrate the absence of contemporary DNA contamination. In our experience the best method to purified aDNA is phenol-chloroform-isoamyl alcohol with concentration using Amicon® Ultra-0.5 30 kDa columns (Millipore, Billerica, USA) or the Silica gel method using the QIAquick (Qiagen) columns. We also prefer to include the EDTA in the extraction buffer to optimise the aDNA extraction. This is supported by recent

network indicate mutational steps between haplotypes (theoretically extinct or

**5. Conclusion** 

from museums or collections where the researcher does not know how they were managed. In these conditions, additional controls are recommended for all of the procedures.

Fig. 12. Phylogenetic analyses of American native populations including five sequences from samples of pre-Hispanic populations. Tree of Native American and ancient pre-Hispanic Amerindian, constructed with the Jukes-Cantor method, and the distances were obtained from a neighbour joining algorithm and optimised for maximum likelihood using Hy-Phy software (Kosakovsky-Pond et al., 2005). The lanes in different colours indicate the haplogroup designation of lineages. Sequences of this study are marked with an asterisk.

Fig. 13. Haplotype network of American native populations including five sequences from samples of pre-Hispanic populations. Each haplotype is represented by a circle in which the square radius (surface) is proportional to its population frequency. Circle colours show the site location as represented by the indicated colours. Dark circles without number in the network indicate mutational steps between haplotypes (theoretically extinct or unrepresented in the sample). Numbers between haplotypes represent mutational steps.
