While the mtDNA mutation rate, despite the environmental and genetic variables to which it is subject 33, approximates to a consensual average value 21,22, we have verified here that the rate of evolution of mtDNA varies over time and can be less than or greater than the mutation rate. Thus, we have reached to the same situation as for the whole genome for which it was found that the evolution rate doubles the mutation rate 11 questioning the chronology of human evolution 34,35. However, we should expect this to be the case if we admit that the rate of evolution is dependent on the population size 20 and that there have been enormous differences in the human population size throughout its demographic history with explosive population growth in recent times 36. This apparent contradiction with the neutral theory of molecular evolution deserves some clarification. The first is to differentiate between the rate of transmission and the fixation rate.
When a mutation appears in a population of size N0 it has a probability of being transmitted to the next generation (N1) of 1/N0, and a probability of fixation, given that it is transmitted, of 1/N1. If the population size remains constant along generations (N0 = N1) the probability of transmission and the probability of fixation is the same just as the neutral theory predicts 37. However, when the population size fluctuates across generations the probability of transmission behaves just the opposite as the probability of fixation. Suppose the case that the population size doubles in the next generation (N1 = 2N0), then the probability of transmission is 2/N0 (it has two chances of being transmitted). However, the probability of fixation of the mutation, given that it is transmitted, is 1/2N0. On the contrary, if the population decreases by half in the next generation (N1 = N0/2) the probability of transmission is 1/2N0 but the probability of fixation, given that it is transmitted, is 2/N0. Again, this is according to the neutral theory that predicts that in a large population the probability of fixation of a mutation is lower than in a small population. However, it is the rate of transmission which drives the evolutive rate.
When we estimate the rate of evolution between populations or species, we are counting mutation differences between sequences regardless of whether these variants are fixed for different alleles or segregate as transient polymorphisms in each population/species. So, what really matters is the amount of variation accumulated independently in each lineage. Indeed, mutations appear according to the mutation rate, but it is known that in a growing population there are more lineages, more mutations accumulate per lineage and they are maintained for longer time than in decreasing population 38,39. This explains why in a growing population the evolutionary rate accelerates and can be higher than the mutation rate, since the mutations that appear are more likely to be transmitted and those that already segregate in the population remain polymorphic for a longer time. Just the opposite occurs in a decreasing population.
The time dependency effect, confirmed here by the comparative analysis of ancient and modern mitogenomes, empirically corroborates the arguments exposed above about the dependency of the evolution rate on the population size dynamics. It is indubitable that selection also plays an important role in the rate of evolution, as repeatedly demonstrated by the differences in the evolution rate of synonymous versus non synonymous substitutions 40–42. However, its effect may be explained into the nearly neutral theory of molecular evolution that gives to genetic drift the main role on molecular evolution 43.
Finally, although we proposed elsewhere a new algorithm to counterbalance the effects of the exponential growth observed in the human population 20, it is based on the topology of the tree grouping the sampled lineages. However, we have constated here that trees ignore the persistence of ancestral lineages along time with the result of giving coalescent ages younger than the real ones. Although we have proposed to take the lineages with the major number of mutations within each haplogroup instead of the average, as a practical approach to estimate the real coalescence times, more sophisticated models should be implemented to cope with the problems that fluctuating population sizes and tree constrains impose to the evolution rate.
Nevertheless, it deserves mention that the two approaches proposed by us have given coalescence times for the origin of modern humans around 300,000 years ago (Table 3) which is more into line with recent archaeological findings 15,16. With a similar delay, the out-of-Africa of modern humans has been estimated as early as 150,000 years ago a time that could match the archaeological records of Skull and Hafez or even those of the Misliya cave in Israel.18,19. Also the coalescence of the Eurasian macro-haplogroups M, N and R would delay the human occupation of Eurasia as early as around 100,000 years ago which coincides with the estimated age for human dental remains excavated in China 44. The split of the Australasian haplogroup P seems to indicate that modern humans reached Sahul at the same time that east Asia (Table 3). As part of this coeval expansion, Eurasian populations might have also returned to Africa coinciding with the African L3 split (table 3) 45. Finally, although we proposed a unique migration for the colonization of the Americas around 40,000 years ago 20 which is directly or indirectly supported by archaeological dates 46,47, it seems possible that this first migration, signaled by the ages of haplogroups A2 and B2 was followed afterwards by a second wave, also before the Last Glacial Maximum, marked by haplogroups C1, D1, D4h3a and X2a around 27,500 years ago (Table 3).
The rate of evolution in humans persists as an open question. However, we think there is enough evidence to question the ages proposed by geneticists for the main events in human evolution.