Demographic Inferences from (Paleo-) Biological Data: The Frame of Reference of Extant Mammals

Paleodemographic Estimates

Two death distributions by age, pre-Neanderthal (Homo heidelbergensis; Bermúdez de Castro and Nicolás 1997; Bocquet-Appel and Arsuaga 1999; minimum number of individuals = 32, now 29; Bermúdez de Castro et al. 2004) and Neanderthal (Trinkaus 1995; N = 206 registration numbers), were obtained, along with life span estimates of the australopithecines, until modern humans (Caspari and Lee 2004). The pre-Neanderthal distribution was obtained by a technique of age estimation by self-reference² based on dental attrition developed by Miles (1963, 2001; Bocquet-Appel and Arsuaga 1999). The Neanderthal age distribution was obtained using the usual statistical techniques exploiting morphological age indicators for adults (Trinkaus 1995), which are known to be biased toward a younger age (Bocquet and Masset 1982). These two sampled paleontological distributions produce very high proportions of young and mature adults (in the Neanderthal sample from Trinkaus 1995, 80% of adults aged 20 yr and over die before the age of 40, not counting the question of children under 5 yr of age, who are still significantly underrepresented in paleodemographic data). These proportions have no equivalent in the many controlled distributions of attritional death in other primates (Rawlins and Kessler 1987; Richard 1985), apes (Courtenay and Santow 1989; Hill et al. 2001; Teleki, Hunt, and Pfifferling 1976), preindustrial humans (Henry and Blayo 1975; Jannetta and Preston 1991), or ethnographic humans (Hill and Hurtado 1996; Howell 1979).³

Caspari and Lee (2004, 2005a, 2005b, 2006) have detected “increased longevity, expressed as the number of individuals surviving to adulthood” (Caspari and Lee 2004:10895) from the ratios of old (30+ yr) to young (15–29 yr) adults (noted OY, which is written in demographic notation as _ωd₃₀/₁₅d₁₅ and read as _ωd₃₀, the number of deaths at age 30 plus a number of years coinciding with the end of life, noted ω, over ₁₅d₁₅, the number of deaths at age 15, plus 15 yr (i.e., between 15 and 29.9 yr). With samples of australopithecines, early Homo, Neanderthals, and Early Upper Paleolithic (modern human), Caspari and Lee obtained OY values of 0.12, 0.25, 0.39, and 2.08, respectively, with an abrupt change between Neanderthals and anatomically modern humans. If the information provided by the data is true, then these ratios, with an increasing number of older individuals relative to younger ones, express a trend toward an increasing average life span in the metapopulation of extinct hominins (Caspari and Lee 2004, 2005a, 2005b, 2006). The robustness of Caspari and Lee’s demonstration resides especially in the relatively large sizes of the paleontological samples collected and in the use of the age estimation technique of Miles, which is homogeneous between groups and allows human groups to be set into a common comparative frame without which this demographic signal, based on the OY indicator alone, would not have been detected.

Derived OY values express another interesting point of information that combined with the estimates of potential longevity L given above (111.7 and 111.2) produces the outlines of living population pyramids. Assuming that populations are stable on average, the pyramids for these populations are regular and can be represented here by a triangular polygon where the height represents the ages and the width the age distribution of the living population. The technical details for the construction of the population pyramid from Caspari and Lee’s data and an observed statistical relationship between OY values for dead and living in sample life tables of modern human populations (table 1; fig. 1) are given in supplement A. What is obtained is a highly schematic representation of a population pyramid at 15+ yr. The pyramids for the two groups are represented in figure 2. They show the flattening effect produced by the departure of the pyramid at the point of the statistical longevity ω rather than at the maximum longevity L, which gives the population pyramids a more acute angle.

Group	Death at 15 − 29 yr = Y = 15d15	Death at 15+ yr = O = ωd15	Y/(Y + O) dead = 15d15/ωd15	Estimated OY livinga = 15L30/15L15	Estimated maximum longevity L
Australopithecinesb	316	353	.895	.505	51.5c
Early Homo	166	208	.798	.539	73.5d
Neanderthal	96	133	.721	.569	111a,b
Early Upper Paleolithic	24	74	.324	.805	111a,b

^a From , , , preindustrial life tables with low life expectancy at birth, excluding Yanomamo, stationary population (see Bocquet and Masset 1977; Bocquet-Appel 2002), , , .

^b Number of Australopithecines to Early Upper Paleolithic, aged by Miles’s technique (Caspari and Lee 2004).

^c Australopithecine and early Homo (Cutler 1975, table 2, excluding Homo habilis; see Caspari and Lee 2004).

^d Homo erectus, average.

View Table Image

The first pyramid for the australopithecines, compared with the others, resembles those known for the great apes. Particularly notable is the similarity of the Neanderthal and Early Upper Paleolithic pyramids, which both have a base width at 15 yr similar to that for early Homo, estimated from paleontological data, and a height identical to the Early Upper Paleolithic, estimated from zoological data. If L and ω are correctly estimated, then in the Neanderthal paleontological data, the proportion of deaths corresponding to 45 − ω of the population pyramid is either missing or not recognized among the skeletal remains. The similarity in the representation of population pyramids for Neanderthals and the Early Upper Paleolithic highlights the contradiction between the biodemographic pattern of longevity of extant mammals, including primates, and the paleodemographic signal detected by Caspari and Lee that does not fit in the frame. We have no solution to this contradiction, but it cannot be ignored (see also Trinkaus 2011). It raises two issues: one is technical and concerns the recurrent problems of paleodemographic age estimation, while the other concerns population biology and the assumption of faster maturation in Neanderthals than modern humans, which we will now discuss.

Miles’s technique is based on a self-reference in the analyzed sample. It therefore seems to escape from the a priori probabilities that are encysted in the anthropological ages/indicators reference sample used in other techniques, which predetermine the age estimates and for which there is now a range of solutions currently under evaluation (Bocquet-Appel and Bacro 2008; Caussinus and Courgeau 2010; Hoppa and Vaupel 2002; Lucy, Ackroyd, and Pollard 2002; Lucy et al. 1996). Nevertheless, it may seem surprising that Miles’s technique, which has been used many times in paleoanthropology since its publication in 1963 (for an overview, see Miles 2001), has never, to the best of our knowledge, been tested with a sample of skeletons of known ages, with its share of immature individuals to calibrate the standards and its share of adults to apply them to. An old idea is that tooth wear is not linear with chronological age, as implied in the technique, but decreases asymptotically in older individuals. As long as this validation test has not been done, in particular to estimate the confidence intervals (CIs), this technique should be considered conservatively, as indicated by Miles himself, as “an art, not a precise science” (Miles 2001:980).

The hypothesis of significantly faster biological maturation in Neanderthals than in modern humans has been put forward in recent years as estimated from the growth of tooth enamel (Guatelli-Steinberg et al. 2005; Ramirez-Rozzi and Bermudez de Castro 2004; Smith et al. 2007). As there is a correlation between tooth development and other life-history traits, then there are grounds for asking whether the life span of Neanderthals, despite their large brains, could also have been shorter, as revealed by the growth of tooth enamel. If this is indeed the case, it would suggest a death distribution at a younger age, on average, than in anatomically modern humans. But there is no consensus over this hypothesis of faster maturation for reasons to do with the dental techniques used, the sizes of samples counted in units, and the failure to consider interindividual variations in Neanderthals and modern human in comparisons (for a summary, see Guatelli-Steinberg 2009). Moreover, the hypothesis of rapid maturation does not imply, in paleodemographic distributions, an absence of individuals with recognized aging markers, as seems to be the case in the data. These faster-maturing individuals would also have these markers of aging, but they would simply be chronologically younger than their modern human counterparts. Finally, ethnographic assumptions (Trinkaus 1995, 2011), or assumptions by analogy with ethnohistorical environmental crises (Bocquet-Appel and Arsuaga 1999), have been put forward to explain the apparent deficit of adults over 40 yr old and even over 25 yr old in pre-Neanderthal and Neanderthal samples.

To summarize, the mortality pattern of Neanderthals, as revealed by paleontological data, does not correspond to the promise of longevity displayed by the manifest continuum of extant mammals, including primates. Could one of the two be wrong? This question, which is beyond the scope of this paper, cannot be evaded.

Demographic Inferences from Paleoclimatic and Archaeological Data

We know there is a connection between climatic variations in isotopic stages and the frequency of Mousterian archaeological sites in Europe (van Andel, Davies, and Weninger 2003). This link is determined—via the amount of biome distribution of primary biomass—by the biomass of ungulate populations available for hunter populations. From these relatively long-cycle climatic variations in the isotopic stages of celestial mechanics, it gradually becomes clear that medium-cycle cold fluctuations (Heinrich events 10–8 kyr [H]) are superimposed over relatively short-cycle and short-duration temperate-cold fluctuations (Oeshger-Dansgaar events 1.5 kyr [DO]). These DO changes, which occur rapidly over time, significantly amplify the predominant climate in the isotopic stages. From oxygen isotope stage (OIS) 5 of the Eemian onward, 25 DO events occurred in Europe (Sanchez-Goñi et al. 2008). Others would be expected to appear in earlier periods when Neanderthals lived but are so far unknown for lack of data. A correlation between the frequency of Upper Paleolithic archaeological sites and the variation in DO events has been interpreted in terms of demographic and cultural changes in Europe, which roughly correspond via changes in the ecological zones of primary and secondary ungulate biomass (d’Errico, Sanchez-Goñi, and Vanhaeren 2006). How did local populations react to these significant variations across their expansion area? Rather than a model of the ebb and flow of small local populations on the geographical margins of the metapopulation to and from attested refuge zones over evolutionary time—which were, in fact, already occupied by the cores of the metapopulation—a model of regional extinction in situ is proposed (Hublin and Roebroeks 2009).

Other consequences of large-amplitude and long-duration climatic fluctuations—in their low points of glacial temperature, moreover amplified by H and DO events—were likely to create demographic bottlenecks for the Neanderthals, who avoided permafrost zones (Aiello and Wheller 2003), that determine stochastic genetic drifts (Bruner and Manzi 2006). But it is not certain, on the other hand, that the peak interglacial temperatures (isotopic stage 5e and 7e), which were similar to current temperatures and corresponded to a plant recolonization on the latitudes, also corresponded to periods of population growth. Recolonization can occur with a very low population density. This is because for the interglacial periods, account must be taken of the result, in terms of ungulate biomass, of the combined effects of (i) the expansion of habitable areas due to the withdrawal of ice masses and despite the rise in sea level (3–7 m above the current level), and (ii) the decrease in the area of the steppe tundra as it shifted toward the high latitudes up to the pole. The conditions that brought this expansion of habitable areas were favorable to forests (van Andel and Tedzakis 1996; Kukla et al. 2002), where ungulate biomass was very low. The northward shift of steppe tundra toward the high latitudes resulted not only in a smaller area, as mentioned above, but also in lower productivity as available sunlight diminished. Until all net effects relative to the distribution of the major biomes during the Eemian have been simulated, taking into account the conflicting effects described above, it will not be possible to equate the interglacial with a demographic expansion of the Neanderthal metapopulation (see also Frenzel 1985; Gamble 1987).

Bottleneck situations and purifying selection are both chronologically identifiable. These correspond to the low (Bruner and Manzi 2006) and likely midpoints of climate series. Nevertheless, and underlying the anatomical traits that appeared in some glacial episodes because of drift (Hublin 1998), eventually producing the classic Neanderthal morphotype described as “hyperarctic” (Aiello and Wheeler 2003; Holliday 1997), the metapopulation had to reach a significant size in these cold and harsh environments for the filter of purifying selection to act on the phenotypes carrying selected genotypes, in particular via Bergman and Allen’s rule. These periods, when populations were relatively more numerous, may have matched phases of OIS 8 and 6, in which long, cold, but not extreme periods emerge (except H and DO events, which are not yet listed).

Finally, from the differential distribution of well-documented archaeological remains in the Perigordian region—between the Châtelperronian 45,000–40,000 yr ago and the Aurignacian 40,000–35,000 yr ago, attributed respectively to Neanderthals and to anatomically modern humans—Mellars and French (2011) have estimated vestige quantity as 10 times larger for the modern human population. The authors equate this relative difference in the amount of remains in this area to the demographic sizes of both populations. This area is a refuge zone on the evolutionary timescale. It should be remembered that the size of the Aurignacian population in this same area has been estimated at 795–12,980 individuals with a 95% CI (Bocquet-Appel et al. 2005:1665, fig. 5). By applying the magnitude of the relative difference estimated by Mellars and French (2011) to this Aurignacian population size, we obtain a local Neanderthal population of 80–1,300 individuals in the Perigordian refuge area before the time of contact.

Hypothesis of a Boserupian Neanderthal Population Trap

The homogeneity of the lithic cultural remains of Neanderthals during their last 150 kyr is striking, except, apparently, at the point of their extinction (Bocquet-Appel and Tuffreau 2009), suggesting very low technical elasticity despite the significant pressure of the environmental hazards summarized above, which should have favored innovations because of the overall change in the ungulate biomass. Questions about the cognitive efficiency of Neanderthals are thereby raised (Neubauer and Hublin 2012). But the hypothesis of technical limitations in the production system of Neanderthal hunters due to a demographic trap (Bocquet-Appel and Tuffreau 2009) must also be put forward again but more properly defined.

In any population, the production of innovations depends not only on its cognitive biological capacities but also on its demographic size. With the same cognitive capacity, under the simple assumption that innovations are produced at a low frequency in any population, then the most demographically numerous population, in absolute terms, will produce the greatest number of innovations (Kremer 1993; Kuznets 1973; Simon 1977; see also Powell, Shennan, and Thomas 2009; Shennan 2001). If the size of the Neanderthal metapopulation remained very low in terms of carrying capacity—that is to say, the maximum number of mouths that it was possible to feed per square kilometer of ungulate biomass given its production system (the technical and social relationships in an environment)—then the metapopulation could have maintained itself in a state of demographic equilibrium at a “critical level of density” (Boserup 1965:33), but its potential technical creativity would have been strongly limited in what is known as a “Boserupian trap” in macrodemography theory.

The technical and social characteristics of the Neanderthal production system might, in central and northern Europe, have operated in an open environment: high residential mobility but not to great distances (Conarda, Bolusc, and Münzeld 2012) and targeted to large gregarious herbivores and resources (horses, bison, reindeer, and ibex) with a smaller amount of many other larger game species typically dominating the assemblages (Conarda, Bolusc, and Münzeld 2012; Gamble 1999; Patou-Mathis 2000); consumption of shellfish, birds, and turtles in the peripheral southern latitudinal zones of the expansion area (Finlayson et al. 2001); direct and dangerous contact with prey animals by killing with lances rather than killing at a distance using projectiles (spears; Gamble 1999) with the aid of beaters and with no division of labor by gender (or by age? Kuhn and Stiner 2006) between hunting and gathering, as observed ethnographically, that is, with both males and females working as hunters and beaters. The carrying capacity of this hunter-gatherer production system along with high incidental mortality (Trinkaus 1995) should be lower than in other systems in which clearly more efficient hunter-gatherer hunting techniques (spears, bows) were used. In addition, the energy balance of this putative production system, combining high energy expenditure due to mobility (following herds) and energy gain from a low calorie diet (essentially game), determines long birth intervals in modern human females and, therefore, low fertility (see Bocquet-Appel 2008). By analogy, it may be thought that the energy balance/fertility reaction norm was similar in Neanderthals and that their female fertility (total fertility rate) therefore tended, on average, toward low values (!Kung: 5 children and less) rather than high values. The relatively low fertility and high mortality of the Neanderthal hunter-gatherer production system would have been a contributing factor in locking them into the Boserupian trap.

Although ethnographic demographic control data are lacking on this point, there are two conceivable circumstances in which a population of hunters may have escaped stagnation in a Boserupian trap: (i) the outcome of the rate of innovation, which would sporadically increased the carrying capacity, and (ii) favorable environmental change (Wood 1998:112). Unlike the realization of the rate of innovation, which is a simple statistical function of the rate and the metapopulation size and cannot be located precisely in time, favorable environmental changes were a certainty in western Eurasia, especially over the relatively long durations of OIS 5 and 7. These favorable environmental changes resulted in the extension of steppe savannah and grassland areas together with the corresponding ungulate biomass and its carnivorous predators, which included the increasing Neanderthal metapopulation. Along with this demographic growth, the population expansion area was gradually homogenized, particularly through a defragmentation of the geographical structure inherited from the previous cold period, the increase in interpopulation migration, and the redensification of local populations, all of which favored the spread of innovations.

With no change in hunting technique (“in the herd”: Gamble 1999) but simply by virtue of the increase in the overall biomass of ungulates during OIS 5 and 7, the logistic mobility of the Neanderthal hunter system could have diminished. The effect of reduced logistic mobility (with the corresponding decrease in energy expenditure) is to improve the energy balance, which results in a rise in female fertility through a reduction in the birth interval (Bocquet-Appel 2008). An increase in the metapopulation or local populations could then occur through an opportunistic adjustment of female fertility to the new carrying capacity rather than a “technological” reduction in mortality due to hunting accidents (via killing at distance). But 150 kyr of apparent Neanderthal technological stability do not make the case for an escape from the Boserupian trap during the climate windows of OIS 5 and 7.

Demographic Inference from Ancient Neanderthal DNA

While there has been general agreement since the 1970s over the effective population size (N_e) of 10,000 individuals for modern humans (Hammer and Zegura 1996; Harding et al. 1997; Nei and Graur 1984; Rogers and Jorde 1995; Takahata 1993; Takahata, Satta, and Klein 1995; Wilson et al. 1985), there is no such consensus over the Neanderthal N_e. There is now enough data from Neanderthal DNA analysis to provide an outline. Analysis of Neanderthal DNA sequences began in 1997 (Krings et al. 1997) from the original specimen discovered in 1856, but information has since become available on mitochondrial sequences from 14 additional Neanderthal specimens (see Degioanni, Fabre, and Condemi 2011 for a review), including complete mitochondrial genome sequences from nine Neanderthal individuals (Briggs et al. 2009; Green et al. 2008, 2010). Neanderthal genomic DNA has also been sequenced since 2006 (Green et al. 2006; Noonan et al. 2006: 65,000 and 1,000,000 base pairs, respectively). Finally, in 2010, the genes of three Neanderthal individuals to 1.3-fold genomic coverage were sequenced by the Neanderthal Genome Project (Green et al. 2010). This project mainly aims to determine the complete sequence of Neanderthal DNA, to answer the question of recent interbreeding between Neanderthals and modern humans, and to provide a catalog of differences between the human and Neanderthal genomes. The Neanderthal Genome Project has also revealed the sequence of several Neanderthal protein-coding genes that have recently evolved adaptively in modern humans (Green et al. 2010). Several publications on Neanderthal DNA sequences focus on the time of the most recent common ancestor of Neanderthals and modern humans, and several publications report estimations for Neanderthal N_e.

The following is a brief digest of the demographic findings of these studies. A first study comparing three short mtDNA sequences (Krings et al. 2000) suggested that Neanderthals had expanded from a small population to explain the lower diversity of Neanderthal mtDNA than in the great apes and the same order of magnitude in modern humans. This result was confirmed by analyzing five Neanderthal mtDNA genomes (Briggs et al. 2009): the N_e was small and probably included fewer than 3,500 females (mean N_e = 1,476; 268 to 3,510, 95% HPD), and the authors proposed that the low mtDNA diversity might reflect a low Neanderthal N_e over a large part of their history. Estimating the θ = N_eμ parameter (θ is the nucleotide diversity among sequences, and μ is the substitution rate per generation), assuming constant population size and 20 yr per generation over a short sequence of 9 HVI mtDNA, Lalueza and colleagues (2005) propose an N_e ranging from 5,000 to 9,000 individuals and confirm that this population would have been constant over time. The complete Vindija 33.16 mitochondrial genome sequence (Green et al. 2008) showed a significantly higher ratio of nonsynonymous to synonymous evolutionary (d_N/d_S) rates in Neanderthals versus modern humans. This result could be explained by a smaller Neanderthal N_e and contrasts with HVI 1 sequence results from Teshik Tash and Okladnikov individuals (Krause et al. 2007)—based on mean pairwise differences, which suggested that Neanderthals had an N_e similar to that of modern Europeans or Asians but lower than that of modern Africans—and also contrasts with the results of Ovchinnikov and Kholina (2010), where the d_N/d_S ratio indicates that the N_e for their common ancestor (the human-Neanderthal branch) tends to be larger than in either Neanderthals or modern humans.

On mtDNA data again, but using a modeling approach, Fabre, Condemi, and Degioanni (2009) found that the demographic scenario that best explains the variability of Neanderthal is a population with an N_e ranging from 3,000 to 25,000 individuals that will grow in size up to 50,000 yr BP and then decline slowly until extinction. A larger distribution value for Neanderthal N_e, whose median corresponds to 32,263 individuals, was proposed instead by the best scenario analyzing a possible admixture process between Neanderthals and early European modern humans (Ghirotto et al. 2011). N_e values from nuclear DNA are very rare but confirm that Neanderthals derived from a very small ancestral population with an N_e of about 3,000 ranging up to 12,000 (Green et al. 2006). This estimated N_e turns out to be similar to that proposed for a population of modern humans: this “small population” character seems to be a feature of both Neanderthal and modern human evolution.

As we have seen, the literature offers different N_e values, but all publications agree that the Neanderthal N_e would be very low. But what is the N_e? N_e is defined as the size of an ideal population, a Wright-Fisher population (Fisher 1930; Wright 1931) that has the same rate of change of allele frequencies or heterozygosity as the population under study. It is not easy to relate the N_e to the census size of a population (N_c). The N_e is almost always smaller than the actual size N_c. Some authors (Belle et al. 2006; Ray 2003; Wood 1987; for a review, see Hawk 2008) propose that the N_e is approximately one-half of the census size corresponding to the reproductive individuals of the population. Moreover, many parameters can affect the N_e value; this raises the question of whether N_e is in fact a reliable predictor of N_c.

In the field of animal species conservation, where it is essential not only to maintain a sufficient population size for the survival of the species but also some degree of genetic variability, researchers were already focusing in the 1970s on the relationship between N_e and N_c, for which Hill (1972) proposed an initial formula. This formula, modified by Nunney (1993) and widely used in animal studies, is written

Combining these values produced nine different scenarios. For each scenario, we apply the formula to estimate the N_a value of the Neanderthal population given the extreme values of N_e estimated from genomic DNA (N_e range from 3,000 to 12,000).

We used N_e values estimated from the analysis of genomic DNA for two reasons. First, because the formulas available to calculate N_a are more suited to this type of data, and second, because we believe that the information derived from the entire genome is less biased than the information contained in mitochondrial DNA alone (females only, very wide chronological dispersion of sequences).

We also tested three different proportions (juvenile population): 40%, 50%, or 60% of the total population. The data used for the nine scenarios are shown in table 2, and the results are shown in figure 3. The figure clearly shows that the N_e/N_a ratio reaches its maximum when the generation time (T) is the longest and life expectancy (A_i) the shortest: in this case the N (N_a and N_c) values are the lowest, even with 60% of young individuals in the population, ranging from 5,000 to 50,000. On the other hand, when the ratio N_e/N_a is the lowest, that is to say, when T is short and A_i is long, N_a and N_c are the highest, with a maximum of 70,000 individuals.

An estimated population size of 5,000 to 70,000 individuals should not be considered as an exact value but rather as an order of magnitude. It should be remembered, furthermore, that the formula proposed by Nunney applies to a population with no subdivisions and no generation overlap. If this is not the case, then generation overlap can reduce N_e to 25%–75% of N_a (Felsenstein 1971). Reproductive variance (variation in the contribution to the next generation) between males and females (A_i) implies that the N_e of portions of the genome with different inheritance patterns can be different (the higher the reproductive variance, the lower the N_e). This means that N_e estimated for the same population but with different markers (Y chromosome, mtDNA, and autosomal markers) can be very different and therefore difficult to compare. Nonrandom mating, in particular assortative mating (mate chosen on the basis of phenotypic similarities) decreases the N_e and increases the genetic drift. In a subdivided population, nonrandom mating can have a greater effect on members of the same subpopulation: the N_e of a subdivided population can be different (lower) compared with a randomly mating population of the same size.

Finally, but crucially, it is important to keep in mind that the average N_e over the long term is not the “classical” arithmetic mean but rather the harmonic mean over several generations (Crow and Kimura 1970; Wright 1938). This means that the N_e is strongly affected by the smaller N_e values and will be close to the smallest N_e over several generations; that is, bottlenecks can mask previously high N_e values.

The question we must ask is to what period the value of N_e calculated from N_e corresponds: beginning, average value, or end? The current proliferation of studies geared to the conservation and management of endangered or exploited species (Gomez-Uchida et al. 2013; Serbezov et al. 2012; Whiteley et al. 2012) has produced new methods for estimating long-term and short-term N_e. This research field will therefore provide answers to our question within a short time.

Concluding Remarks

Paleodemographic data are eclectic by nature, but the effort must be made to integrate their interpretation without hiding the difficulties and to resolve contradictions, especially regarding paleontological metapopulations such as the Neanderthals, where the mists of time become increasingly impenetrable with chronological depth. In Neanderthal paleodemographic death distributions by age, very few adults are older than 40, while the promise of potential maximum longevity implied by the quasi-biological continuum of mammals points to much more. One could even venture to assume that the Neanderthal and modern human death distributions should be similar.

Increasingly detailed reconstitutions of climate, layering multiple sequences of variations that range from very long to short periods (Sanchez Goñi et al. 2008), and their connections to the Neanderthal population through geographically distributed primary and secondary biomass raise new questions. It is as if the last 10,000 yr of the Holocene, during which the modern human metapopulation will reach 9 billion people, were a temperate niche of stability, as the latest similar niches date back to OIS 5 (Eemian: 114–130 kyr) and OIS 11 (core: 400–420 kyr; see also Richerson, Boyd, and Bettinger 2009). Except in these three temperate niches outside the Mediterranean zone, the vegetation was mainly cold steppe tundra and was regularly devastated by what would now be akin to catastrophic DO and H climate events. It can be hypothesized that the demography of the Neanderthal metapopulation, living under conditions where extreme environmental instability with short periods was the norm, was primarily stagnant, with frequent bottlenecks and episodes of decline.

Finally, the Neanderthals may have suffered the additional handicap of a system of specialized hunters, described as top-level carnivores (Richards and Trinkaus 2009), even though plants were also consumed (Henry, Brooks, and Piperno 2011). Relative to a modern human, the estimated individual metabolic cost of an adult Neanderthal is very high (3,500–5,000 kcal/d vs. 2,150–2,400 kcal/d for a male modern human; Churchill 2007; Sorensen and Leonard 2001; Steegman, Cerny, and Holliday 2002). Assuming a similar distribution of nutrients between Neanderthals and modern humans, food intake per capita for Neanderthals would have required almost twice the mass of ungulate meat as that consumed by modern humans. Mechanically, given similar natural hunting conditions, this level of ungulate consumption implies that the Neanderthal population density was half that of modern humans. The demographic instability of this metapopulation of specialized hunters, which was small on average, and the variation of its geographical area of expansion and fragmentation, should help to understand why it stagnated technologically (Bocquet-Appel and Tuffreau 2009) and probably also socially (see the array of Hayden 2012), as it spent most of its evolutionary time moldering in the depths of a Boserupian demographic trap.

Measuring the acquisition or lack of acquisition by Neanderthals of a modern form of human behavior—that is, the ability, usually shown in modern humans, to express a very wide range of cognitive responses to contrasting socionatural situations—by simply comparing lists of archaeological cultural items between the two groups is not an appropriate approach (see Shea 2011). If one of the metapopulations (the Neanderthals) remained largely within the confines of the experience of the Boserupian traps mentioned above while the other (modern human), numbering several million, was able to experience the African and Eurasian population expansion, then comparing their lists of cultural items at a point of fortuitous spatiotemporal contact (say in the Périgord at 40–35 kyr) does not inform us about their different biological cognitive potential. The differences between these lists will mirror those of the amplitudes of the socionatural experiences of these contemporaneous metapopulations whose demographic numbers varied by a factor of 100 and perhaps even more. The cause of the cultural limitation of the Neanderthal metapopulation compared with that of modern humans may well have resided in its small numbers alone.