Lombards on the Move – An Integrative Study of the Migration Period Cemetery at Szólád, Hungary

In 2005 to 2007 45 skeletons of adults and subadults were excavated at the Lombard period cemetery at Szólád (6th century A.D.), Hungary. Embedded into the well-recorded historical context, the article presents the results obtained by an integrative investigation including anthropological, molecular genetic and isotopic (δ15N, δ13C, 87Sr/86Sr) analyses. Skeletal stress markers as well as traces of interpersonal violence were found to occur frequently. The mitochondrial DNA profiles revealed a heterogeneous spectrum of lineages that belong to the haplogroups H, U, J, HV, T2, I, and K, which are common in present-day Europe and in the Near East, while N1a and N1b are today quite rare. Evidence of possible direct maternal kinship was identified in only three pairs of individuals. According to enamel strontium isotope ratios, at least 31% of the individuals died at a location other than their birthplace and/or had moved during childhood. Based on the peculiar 87Sr/86Sr ratio distribution between females, males, and subadults in comparison to local vegetation and soil samples, we propose a three-phase model of group movement. An initial patrilocal group with narrower male but wider female Sr isotope distribution settled at Szólád, whilst the majority of subadults represented in the cemetery yielded a distinct Sr isotope signature. Owing to the virtual absence of Szólád-born adults in the cemetery, we may conclude that the settlement was abandoned after approx. one generation. Population heterogeneity is furthermore supported by the carbon and nitrogen isotope data. They indicate that a group of high-ranking men had access to larger shares of animal-derived food whilst a few individuals consumed remarkable amounts of millet. The inferred dynamics of the burial community are in agreement with hypotheses of a highly mobile lifestyle during the Migration Period and a short-term occupation of Pannonia by Lombard settlers as conveyed by written sources.


Osteology
The ages at death, sexes, body heights, diseases, traumata and stress markers were determined using standardised osteological methods [1]. The tooth development and the closure of the epiphyseal sutures provided the values underlying the age estimations of children and juveniles [2], while the aging criteria for adults were the closure of the ectocranial sutures [3], the structure of the facies auricularis [4], the relief of the symphysis of the os pubis [5], tooth abrasion [6], as well as the combined method proposed by Nemeskéri et al. [7]. The sex determination of juveniles and adults was based on morphometric features of the pelvis and the skull [8][9][10][11]. The sexes of the children were not determined. The body heights were calculated from long bone measurements using formulas proposed by Pearson [12]. The Pearson formulas were used because they were established based on populations that had not yet been affected by the increase of average body heights in the 20 th and 21 st centuries.
In order to ensure a detailed data acquisition the skeletons were broken down into 140 sections, all of which were assessed. This allowed us to calculate the ratio of individuals or bone regions affected by disease or trauma in relation to the total number of observable sections [23]. Osteoarthritis was scored in four grades including slight (1), moderate (2) and severe (3) alterations as well as age-related or arthritis-induced ankylosis (4). Caries appearance was scored into four stages: caries superficialis (enamel affected), media (dentin affected), profunda (2/3 of dentin affected, pulpa aperta), and tooth destroyed [24]. (haplogroup U). Sequence data were replicated by at least three independent amplifications from two different samples. In addition, selected PCR products were cloned and 5-8 clones per PCR were sequenced to monitor possible background contamination and DNA damage. The sample preparation, DNA extraction, amplification, cloning, and sequencing followed standard protocols as previously described [25][26][27]. Sequence polymorphisms (S2, Tab. 12) were reported relative to the revised Cambridge Reference Sequence (rCRS) [27] and haplogroup assignment was carried out using the software HaploGrep (http://haplogrep.uibk.ac.at/) [29] based on the mitochondrial haplogroup phylogeny of phylotree (http://www.phylotree.org/, built 14, accessed 5 Apr 2012) [30]. Shared haplotypes of the Szólád cemetery were compared to an in-house database of 21,724 published HVS-I and II sequences, with the aim of assessing kinship probabilities. Sequences have been deposited in GenBank (http://www.ncbi.nlm.nih.gov/genbank/; AccNo. KM114982-KM115015).

Carbon and nitrogen isotope analysis
Collagen extraction followed [31] and [32] with some modifications. Samples were cut, and the surfaces and adhering cancellous bone abraded with dental cutting and milling equipment. Spectrometer (GV Instruments). Isotope compositions were reported in δ notation in per mil relative to V-PDB for carbon and AIR for nitrogen. The raw data were normalised using two-point calibrations based on USGS 40 and IAEA N2 for nitrogen and CH6 and CH7 for carbon [33]. Reproducibility of internal and external standards (sulfanilic acid, USGS 40) was better than ± 0.2 ‰ for nitrogen and ± 0.1 ‰ for carbon.

Strontium isotope analysis
The teeth were sliced in half using a water-cooled, diamond-impregnated rotating blade of 0.3 mm thickness. Enamel samples were prepared from one half of each tooth using a rotating disc attached to a handheld drill; the remaining dentine was removed and finally all surfaces were cleaned using a dental burr. No dentine remained attached to the enamel, which was thoroughly checked under a binocular microscope. Enamel sample sizes prior to leaching ranged between 6.3 and 84 mg.
Overall, analytical protocols followed [34] with certain modifications: Prior to dissolution, enamel fragments were cleaned ultrasonically in reagent-grade acetone, methanol and repeatedly in deionized H 2 O, and leached overnight with 1 ml of ~0.25 M acetic acid to remove adsorbed or least strongly bound and most easily diagenetically affected Sr [35]; leachates were not analysed. Subsequent dissolution (wet-ashing) was performed overnight in thoroughly acid-cleaned, closed 7 ml PFA vials leaching removes a non-reproducible amount from each sample and no dry weights were recorded following the leaching procedure. Procedure blanks for tooth analysis were ~60 pg Sr and no blank corrections were necessary. Constants used were those from [36].
Eight grams of each soil sample were leached in ~0.25 M acetic acid for one week and shaken regularly. Each leachate was centrifuged, dried down and re-dissolved with ~15 M HNO 3 to remove organics, after which Sr was separated using the same Sr-Spec procedure described above.
In order to minimise modern anthropogenic influence to the highest possible extent, the vegetation samples were collected in forests. Fresh leaves of ground vegetation, mostly, were cleaned using demineralised water, dried overnight at 50°C and then ashed in acid-cleaned silica crucibles (550ºC/12h).
The plant ashes were digested in ultrapure concentrated HNO 3 in teflon beakers on a hot plate at 160°C. Strontium matrix separation was carried out with the Sr-Spec resin on mini-columns (250 µl Resin) using 3 N HNO 3 for loading and washing and H 2 O to elute the Sr from the resin. The strontium isotope compositions were measured with TIMS (Sector 54). The TIMS raw data were corrected for a potential rubidium contribution on mass 87 and mass fractionation was correct using an 86 Sr/ 88 Sr isotope ratio of 0.1194. Finally the 87 Sr/ 86 Sr isotope ratios were normalised to the widely accepted isotope ratio of 0.71025 for NIST SRM 987. The accuracy of the method used is demonstrated by mean 87 Sr/ 86 Sr isotope ratios of multiple aliquots of fully processed NIST SRM 987 and the seawater standard IAPSO of 0.710260 (N = 6) and 0.709146 (N = 4) respectively. The expanded analytical uncertainty of the 87 Sr/ 86 Sr isotope ratios is less than 0.012 % (exemplarily calculated using the IAPSO seawater standard).
The water samples were collected in 30 ml acid-cleaned Teflon tubes and acidified with 100 µl of HNO 3 . The strontium isotope ratios of the water samples were obtained using an MC-ICP-MS (VG Axiom) at the Curt-Engelhorn-Center for Archaeometry Mannheim following previously described methods [37,38]. Raw data were corrected according to the exponential mass fractionation law to 88 Sr/ 86 Sr = 8.375209. The Eimer & Amend standard that was analysed along with the samples yielded a mean of 0.70801 ± 0.00004 (2 SD, n = 45). The inter-laboratory mean is 0.708027 ± 0.000035 (1 SD) [39]. Two aliquots of the same plant sample measured with the TIMS and the MC-ICP-MS provided consistent 87 Sr/ 86 Sr (difference < 0.00009) values.