We analyzed eight different cpDNA regions in extant and fossil material from Norway spruce and six of them showed polymorphism (Table ). The alignment of the extant sequences revealed 13 point mutations and five length polymorphisms that yielded nine haplotypes (Table and Additional File 1
). Two haplotypes were the most common (Ht2 and Ht8) found in 15 and 22 out of 58 individuals, respectively. Out of 462 fossil pollen grains analyzed we obtained totally 28 positive amplifications (three amplifications for fragment B, four for fragment D, ten for fragment TL, seven for fragment LF, two for the non-variable fragment CK, and finally two for the non-variable fragment MD). The higher amplification success (16.5%) obtained with pollen compared to Parducci et al.
] was likely due to the choice of using more (usually five) pollen grains instead of one during amplification trials, or it may simply indicates favorable condition for DNA survival in the lake sediment at the Retezat. After sequencing, four of the 28 amplicons (two obtained with primers B and two with primers D) showed to be of Pinus
origin, while seven amplicons obtained with primers LF were of Abies
origin. Pollen grains of these two taxa were present in the examined sediment layer [30
] and they were likely selected by mistake during the analyses. The rest of the amplicons were assigned to P. abies
. Another possible explanation for the findings of Pinus
sequences is DNA contamination from molecules of the surrounding sediment that may permeate or adhere to macrofossils in the sediments [36
]. Nevertheless, these sequences still provide authentic early Holocene DNA information from Pinus
Base substitutions detected in extant and ancient cpDNA sequences of Norway spruce (Picea abies)
Out of the investigated six Norway spruce seeds, DNA was successfully amplified from five seeds. Altogether, we obtained 15 positive amplifications (one for fragment B, three for fragment D, three for fragment TL, two for region Li, one for region LF, two for fragment K2i, one for the non-variable CK fragment, and two for the non-variable fragment M) (Table ). When sequences were concatenated, the total length of the haplotypes varied between 124 and 1016 bp in the five seeds, and between 136 and 251 bp in pollen. Concatenation of the four regions amplified from the 4 cone scales gave a haplotype with a total length of 901 bp. Because cpDNA is paternally inherited in Norway spruce, sequence difference between cone and seed haplotypes (mother-tree versus offspring) were not unexpected. We found three transversion substitutions at position 77 in fragment D (C > A), position 197 in fragment TL (C > A) and position 157 in fragment Li (A > C). Finally, replication of the analyses done at Uppsala University using four new seeds retrieved at 502, 513, 525 and 545 cm depths in the same sediment, gave two positive amplifications (from sample 502 and 545 cm) using primers B. Both sequences were of Norway spruce origin and showed the 12A + 5G block, the most common variant present in the extant population.
Comparison of ancient and extant sequences
After manually editing for base call errors, we performed an alignment between homologous regions obtained in extant and ancient individuals. The alignment revealed mismatches at several positions. Transition substitutions (G > A and C > T) were interpreted as incorrect nucleotide PCR incorporations due to post-mortem lesions in the ancient DNA molecules, a common problem when amplifications are performed from degraded DNA molecules [5
]. We therefore neglected two G > A substitutions detected in sample P13 (fragment TL, positions 168 & 206 in Table ). The rest of the substitutions were considered authentic, i.e. due to true polymorphism. Two variants were unique to the ancient population: a 10A + 7G repetition found in sample S9 (fragment B), and a GA dinucleotide substitution present in sample P13 (fragment TL). As expected, variants that were common to the extant population were also found in the ancient material (e.g. the common A, A, and C nucleotides at positions 77, 78 and 82 in fragment D was found in seeds S3 and S10). However, also rare variants found in extant individuals were detected among the ancient samples. For example, the 10A + 6G repetition in fragment B, and the C > A substitution at position 157 in fragment Li, both with a frequency 0.017 in the extant population, were found in sample P1 and S4, respectively. Similarly, a C > A substitution at position 197 in fragment TL was detected in samples P12 and S10.
Population genetic analyses
Estimates of nucleotide diversity, haplotype diversity and average number of nucleotide differences calculated at the six polymorphic cpDNA loci are presented in Table . Observed values for fragment D, TL and Li suggest that the ancient spruce population harbored higher levels of genetic variation compared to the extant population. Fst values at these loci were also statistically significant, an unusual result considering the low level of differentiation usually observed among populations of conifer species. Nucleotide-based genetic differentiation (Ks and Kst,) was also statistically significant at these three loci. On the whole, our results suggest differentiation between extant and ancient spruce populations that inhabited the areas surrounding Tăul dintre Brazi at 10,700-11,000 cal yr BP.
Genetic diversity and population differentiation estimates in ancient and extant Norway spruce (Picea abies) populations based on six cpDNA loci
Phylogenetic relationships of ancient and extant specimens
Results of the phylogenetic analyses conducted on separate loci based on all sequences available (data set c in the Methods) for fragments D, TL, Li and LF are presented in Figure . A small group of sequences is formed by nine extant and four ancient sequences in the case of fragment D, while the other two ancient sequences are grouped with the rest of the extant material. In the case of LF fragment, the ancient sequences obtained from the cone and seed S8 are clustering together with the majority of the extant samples, while a smaller group of extant sequences that included samples collected on the northern slopes of the Retezat, form a well-supported cluster. The phylogenetic analysis conducted on fragment Li grouped together the majority of extant samples with one ancient sequence obtained from seed S10, while the other two ancient sequences form a well-defined clade with one extant sequence collected on the shores of Tăul dintre Brazi. For fragment TL we obtained the largest number of sequences from the ancient material and the majority of them grouped together with the extant specimens. Only three ancient and one herbarium specimen form a well-separated group.
Phylogenetic tree based on B/MCMC analysis of D, LF, Li and TL chloroplast fragments with ancient sequences indicated in boxes. Posterior probabilities are shown along the branches.
Finally, the a median-joining network created using microsatellite B sequences (Figure ) grouped the two previously identified 100-year-old spruce pollen from Sweden [7
] together with the common haplotype detected in this study in extant and herbarium specimens. Our ancient sequences were also similar to this common extant haplotype with a single mutation step away from it, thus suggesting genetic continuity between Norway spruce populations.
Figure 4 Median-joining haplotype network of the chloroplast fragment B. Numbers denote haplotypes found for fragment B only. Haplotype 1 corresponds to Ht9 (Table 1); haplotype 2 includes Ht2, Ht3; haplotype 3 includes Ht1, Ht6, Ht8; haplotype 4 includes Ht 4, (more ...)
PAR-based population size estimates
Figure shows pollen accumulation rates (PAR) of Norway spruce calculated at Tăul dintre Brazi for the last ca
. 15,700 cal yr BP. PAR values are very low until 11,100 cal yr BP (0-270 grains cm-2
) with an average of 120 grains cm-2
that indicate very small population size in the Retezat Mts (ca
. 1.2 - 6.4 trees per hectare according to the equation of Seppä et al.
] for the Finnish woodlands). The Lateglacial interstadial, between 14,450 and 12,900 cal yr BP (zone 2 in Figure ) show moderately higher PAR values (av. 155 grains cm-2
). These values, together with the concurrent findings of spruce stomata (Figure ), suggest the presence of a small lakeshore population during the Lateglacial interstadial at the study site. It is likely that spruce was absent from the lakeshore prior to this period and for a short period during the successive Younger Dryas climatic reversal (zone 3), between ca
. 12,600-12,400 cal yr BP. From ca
. 11,100 cal yr BP (period 4 and 5) a massive increase in spruce PAR, stomata concentration and pollen frequencies commenced with PAR values increasing from 200 to nearly 7000 grains cm-2
in about 1000 years. Such values are equivalent to an approximate population size increase from 9 to 101 trees per hectare [35
]. It should be noticed however, that due to approximation in the PAR-based population size extrapolations only the magnitude of this change is certain. Subsequently, the highest overall Holocene PAR values and hence inferred population size was detected between ca
. 10,100 and 9180 cal yr BP (av. 4886 grains cm-2
) followed by a modest decline, with still high PAR values until 7200 cal yr BP (zone 6). At this time, a significant decrease suggests also a decrease in population size (from ca
. 252 to 120 trees per hectare on the basis of Seppä et al.
]). These values were maintained until ca
. 3900 cal yr BP and were followed by a second significant increase between 3900 and 2200 cal yr BP (zone 8). Finally, the PAR curve suggests a third decrease in population size after 2200 cal yr BP. This period coincides with the late Iron Age increase in livestock and the development of high altitude seasonal pasturing in the Romanian Carpathians [37
] suggesting intensive timber exploitation by humans in the region. Overall, the PAR record suggests that following an early Holocene massive population increase on the northern slopes of the Retezat, the spruce population size did not remain constant, but showed marked fluctuation with three clear population bottlenecks dated to ca
. 9180, 7200 and 2200 cal yr BP.
Lateglacial and Holocene pollen accumulation rates (PAR) of Norway spruce (Picea abies) from Tăul dintre Brazi in the Retezat Mountains. The dashed line on top of the PAR curve displays a three-term running mean.
Figure 6 Pollen and stomata record of Tăul dintre Brazi plotted against depth and calibrated BP timescales. Norway spruce (Picea abies) fossils are highlighted in red. LPAZ: local pollen assemblage zones as in Table 3; ➊ first occurrence of spruce (more ...)