Neanderthals and humans interbred in Europe for 5,400 years

The Lamalunga cave opens in the limestone of the Murgia plateau at an elevation of 508 m a.s.l., near the town of Altamura (Puglia, Italy; Agostini, 2011). It constitutes the upper part of a larger karstic complex where stalactites, stalagmites, and flowstones occur together with “coralloid” formations, which mostly represent the last phase of calcite precipitation caused by spray/aerosol phenomena. This complex consists mainly of a sub-horizontal gallery that had developed at a shallow depth from the surface, intercepted by pits that had originally opened to the surface but which have subsequently been clogged by detritus. In this context, the discovery of a virtually complete fossilized hominin skeleton in an excellent state of preservation gives rise to interesting taphonomic considerations. Particularly, faunal remains found in some of the galleries are often isolated bony elements accumulated in depressed areas of the cave, suggesting that they were transported and dispersed by water. This was not the case with the human skeleton, given that it is largely represented and concentrated in a small area. Thus, we may hypothesize that, after death and decomposition of the body, the skeleton collapsed where it has been found. Thus far, no lithic tools have been found in the cave.


Figure 1. A) Position of Altamura within the Italian peninsula; B) hominin bones and calcite formations around the cranium (part of the mandible and right femur are visible); C) general topography of the northern part of the Lamalunga karstic system; note on the left the accumulation of detritus that represents the infilling of the probable main original access point from the external surface; and D) distribution of the main bones of the skeleton at the end of the so-called “ramo dell'uomo” (compare SOM Fig. 1). Drawing and data of Fig. 1D are from Vacca and Pesce Delfino (2004).

Even though the skeleton is partly incorporated into calcite concretions and is covered by coralloid formations, most of the bones are visible (see Fig. 1; see also Supporting Online Material [SOM] Fig. 1), including the cranium (upside down), the mandible, and several postcranial elements. From the photographs available and direct observations made in situ by one of us (GM), the skeleton appears to exhibit a mixture of archaic and derived features, which fit the range of variation typical of European hominins of the late Middle/early Late Pleistocene (Manzi et al., 2011). In fact, even though a number of Neanderthal traits can be seen—particularly in the face and in the occipital bone—there are features that distinguish this specimen from the more typical morphology of Homo neanderthalensis, such as the shape of the brow ridges, the relative dimension of the mastoids, and the general architecture of the cranial vault.

U/Th dating and petrography
A previous series of 25 U/Th dating was carried out by alpha spectrometry on a series of stalactites, flowstones, and coralloids (Fig. 2) by Branca and Voltaggio (2011). These revealed an ancient phase of speleothem formation, dating to between 189 ± 29 and 172 ± 15 ka, and a second phase, indicated by some of the flowstones, between 45.9 ± 1.7 and 34.4 ± 1.5 ka, while the ages of the coralloids (13 analyses in all) were distributed continuously between 43.3 ± 1.6 and 29.1 ± 1.0 ka and between 17.1 ± 0.7 ka and 13.4 ± 0.5 ka. Outside of these age groups, two additional single dates were recorded: 133 ± 9 ka for a stalactite and 98.7 ± 4.4 ka for a flowstone.


Figure 2. Synthesis of the Alpha spectrometry ages on speleothem from the Lamalunga cave and the new MC-ICP-MS analyses versus their U content; the error bars of the MC-ICP-MS analyses (2σ) are always smaller than the plotted symbols. (* Data from Branca and Voltaggio, 2011).

The microstratigraphy of the newly collected calcite samples ABS3 and ABS5 highlights four growth phases separated by three discontinuities (Fig. 3). The oldest growth phase, directly coating the hominin bone in ABS3 and the broken stalagmite in ABS5, is characterized by micrite laminae grading into microsparite and columnar calcite. This small similar pattern suggests that the two layers could be correlated and have a similar age. The other layers consist of elongated open and compact columnar calcite (cf. Frisia et al., 2000 and Frisia and Borsato, 2010). On the basis of the identical microstratigraphy in the two samples, a similar micro-sampling strategy was performed in order to double check the age of the calcite overgrowth on the hominin bone. Moreover, since the overgrowth on ABS5 is considerably thicker than that on ABS3, this approach permitted avoidance of possible contamination from adjacent layers. We performed six MC-ICP-MS analyses on the three calcite overgrowths over the hominin bones and four analyses on the corresponding growth phases identified on the overgrowth on stalagmite ABS5 (Fig. 3). The results (Table 1) revealed four growth episodes dated to 7.04 ± 0.72–7.6 ± 0.04 ka, 36.9 ± 0.17–38.1 ± 0.61 ka, 64.6 ± 0.33 ka, and 121.9 ± 2.22–130.1 ± 1.9 ka, respectively corresponding to the warm Marine Isotope Stages (MIS) 1.0, 3.1, 5.1, and 5.5 (cf. Martinson et al., 1987) and matching coeval speleothem growth phases in Mediterranean caves (Bar-Matthews et al., 2003 and Badertscher et al., 2011). The correlation between speleothem growth phases in Altamura cave and warm MIS gives rise to an important consideration: the warm Marine Isotope Stage 7.1, between ca. 185 and 200 ka (Martinson et al., 1987), was not recorded in the coralloid overgrowths over the hominin bones, although it was recorded in two other stalactites from the same cave chamber (growth phase between 189 ± 29 and 172 ± 15 ka; cf. Fig. 2), as well as in speleothems from other Mediterranean caves (Bar-Matthews et al., 2003 and Badertscher et al., 2011). Given the fact that all the other growth phases are represented in overgrowth ABS3, this suggests that the hominin bones could be more recent than 172 ± 15 ka.


Figure 3. Selected calcite crusts and coralloid overgrowths with the calculated U/Th ages: A) thin calcite crust coating the underside of a long bone (fibula; ABS2); B) polished slab of mm-thick coralloid overgrowth covering the termination of a short bone (ABS3); and C) thin section of coralloid mm-thick overgrowth covering a naturally broken stalagmite (ABS5). The dotted, dashed, and dotted-dashed red lines on ABS3 and ABS5 visualize similar discontinuities that define the different growth phases identified by the U/Th ages. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Morphometric analysis
Metrical variables show that the scapulo-humeral joint of the Altamura skeleton is similar to those of European samples from the Middle and Upper Pleistocene with respect to the diameter and depth of the SGF (as reported in the SOM Fig. 5). Geometric morphometrics may be used to better characterize the shape of the SGF, which has diagnostic significance (Di Vincenzo et al., 2012). When 2D landmark data are analyzed by Principal Component Analysis (PCA; Fig. 4B), the distribution of the samples reveals a signal of taxonomic and phylogenetic significance. Notably, the linear regression of the centroid size values on the 1st PC (r = −0.23; p = 0.057), as well as the multivariate regression of the centroid sizes on all the PC scores (p = 0.522) are not significant; thus, our results are not significantly influenced by allometry.


Figure 4. A) The articular portion of the right scapula from Altamura, viewed from the glenoid fossa (SGF), with landmarks (darker/green points in the online version of the paper) and semilandmarks (lighter/yellow points); B) Variance along PC1 and PC2 of the whole sample; OTUs are bounded by lines, while labels of the specimens are as in SOM Table 1; deformation grids represent extreme shape variations. C) Neighbour Joining of the phenetic relationships between the averaged OTUs and the specimen from Altamura. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Paleogenetic analysis
Despite several PCR attempts, it was only possible to obtain positive results from one of the shortest amplicons (82 bp; see SOM Fig. 6). All sequence results, obtained in two different laboratories, were concordant and presented a typical Neanderthal mtDNA haplotype (Condemi et al., 2013; A16230G, G16244A, C16256A, A16258G; Fig. 5). The consensus sequence was deposited in GenBank under the accesion number KJ888153. Despite its short length, the mtDNA fragment from Altamura falls within the known range of Neanderthal diversity, being clearly distinct from both modern humans and the Denisova sequences. The sample from Atapuerca Sima de los Huesos (Meyer et al., 2013) was excluded from the subsequent analysis because for the fragment considered here it is poorly covered and misses informative sites. The sequence from Altamura clusters together with other Neanderthals from Western Europe in the tree reported in Figure 6. This picture seems to confirm the existence of some degree of population structuring among Neanderthals, already present around 150 ka. Unfortunately, the genetic data used in the analysis are too scant to make robust inferences regarding patterns of temporal and/or geographic variation in Neanderthal genetic diversity.


Figure 6. Neighbor joining tree showing the phylogenetic relationships among Neanderthals on the basis of the present analysis. The name of the samples are reported next to their accession number (if present); Denisova is used as an outgroup.

Conclusions
Overall, the results of our morphometric and the paleogenetic analyses concur in indicating that the skeleton from Altamura belongs to a Neanderthal. In addition, using U/Th dating we were able to provide the first range of dates for the specimen, between 130 ± 2 ka and 172 ± 15 ka. Nevertheless, some features exhibited by the skeleton and observed in situ (on the cranium, in particular, as summarized in the Introduction) differ from the morphology known among the typical representatives of Homo neanderthalensis, while they appear consistent with the pre-Würmian age we obtained. Metrical variables show that the scapula-humeral joint is closer to the morphotype usually referred to the so-called “early Neanderthals,” including specimens such as those from Saccopastore (e.g., Bruner and Manzi, 2006), Krapina (e.g., Monge et al., 2008), or Apidima ( Harvati et al., 2011). In addition, geometric morphometric analysis of the SGF from Altamura suggests some peculiarities of this small piece of bone, while (consistent with the mtDNA data) the same analysis strengthens the notion that the Neanderthal morphology was essentially present in the late Middle Pleistocene.

Finally, it is of great interest that mtDNA was sufficiently preserved to permit paleogenetic analysis. The results of the explorative approach used here have shown that the sample contained endogenous DNA (although highly fragmented) with a typical Neanderthal haplotype; moreover, there was no evidence of modern human contamination in the bone fragment, at least not at the mtDNA level. For these reasons, the Altamura skeleton should be considered a good candidate for more innovative genomic analyses, like capture approaches or ultra-deep shotgun sequencing, especially when we consider that Altamura represents the most ancient Neanderthal from which endogenous DNA has been retrieved so far.

doi:10.1016/j.jhevol.2015.02.007