Female bone physiology resilience in a past Polynesian Outlier community

Remodelling is a fundamental biological process involved in the maintenance of bone physiology and function. We know that a range of health and lifestyle factors can impact this process in living and past societies, but there is a notable gap in bone remodelling data for populations from the Pacific Islands. We conducted the first examination of femoral cortical histology in 69 individuals from ca. 440–150 BP Taumako in Solomon Islands, a remote ‘Polynesian Outlier’ island in Melanesia. We tested whether bone remodelling indicators differed between age groups, and biological sex validated using ancient DNA. Bone vascular canal and osteon size, vascular porosity, and localised osteon densities, corrected by femoral robusticity indices were examined. Females had statistically significantly higher vascular porosities when compared to males, but osteon densities and ratios of canal-osteon (~ 8%) did not differ between the sexes. Our results indicate that, compared to males, localised femoral bone tissue of the Taumako females did not drastically decline with age, contrary to what is often observed in modern populations. However, our results match findings in other archaeological samples—a testament to past female bone physiology resilience, also now observed in the Pacific region.


SI
. Descriptive summary of data differences between the left and right femora. SD: standard deviation, SE mean: standard error mean.

EXTENDED DNA METHODS
The DNA processing was conducted in dedicated aDNA facilities at the Max Planck Institute for the Science of Human History, Jena, Germany. For each individual, the dense part within the petrous portion of the temporal bone was drilled for DNA sampling [1]. DNA was extracted from around 50 mg of the sampled powder following published protocols [2]. To prepare the extract for next-generation sequencing a 25-ul aliquot was processed to produce a double-stranded and double-indexed Illumina DNA library following [3,4]. To prevent that post-mortem deamination damages would be mistaken as authentic sequences in downstream analysis, damage caused by cytosine deamination was partially removed using uracil-DNA glycosylase and endonuclease VIII as described in [5]. Damage was retained in the two terminal positions to be later used for estimating the fraction of deaminated reads [5]. The DNA libraries were subsequently amplified using Herculase II Fusion DNA polymerase according to the manufacturer's protocol. All libraries were directly shotgun single-end sequenced on an Illumina HiSeq 4000 platform (1 × 75 + 8 + 8 cycles). To control for potential laboratory contamination, blank extractions and library preparations were included for each sample batch.
The sequenced reads were binned (demultiplexed) allowing for one mismatch per index. The multiplexed libraries were than processed using the EAGER (v 1.92.54) pipeline [6]. As part of the pipeline, the Illumina adapter sequences were clipped off and the reads were filtered, retaining only reads longer than 30 base pairs using AdapterRemoval (v2.2.0) [8]. The clipped and filtered reads were mapped against the human genome reference hg19 using the BWA aln/samse alignment software (v0.7.12), with a stringency parameter of 0.01, seeding off (-l 16,500), and only retaining reads with Phred-scaled mapping quality scores higher than 30 [8]. Duplicate reads were removed using DeDup v0.12.2 [6]. To authenticate the ancient DNA library, levels of DNA deamination post-mortem damage were measured using mapDamage (v2.0) [9] and compared to the expected values in similar libraries prepared from ancient skeletal elements. Two terminal positions of each fragment were then masked to exclude DNA damage from following analyses [10].
Due to the low number of sequences yielded for each library, the genetic sex was inferred using two independent approaches. Both approaches aim to determine the copy number of each sex chromosome by calculating the number of reads mapping to sex-and autosomal chromosomes. Since genetic females have two copies of the X-chromosome and two copies of each autosomal chromosome, their X-chromosome coverage is expected to be comparable to the autosomal one. However, males have only one copy of the X-chromosome and one of the Y-chromosome and therefore the coverage of each of their sex chromosomes is expected to be half of the autosomal one.
The first method uses the mapping counts across a total of around 1.24 million genome-wide SNPs that were ascertained since they are informative for population history studies [11][12][13]. However, they can also be useful to estimate genetic sex [14]. For this purpose, the reads mapping to each ascertained position were counted using SAMtools and averaged for each chromosome using an inhouse script [15]. The Y-and the X-chromosome average coverages were each normalized using the average autosomal coverage. Then the normalized Y-and the X-chromosome average coverages were compared and used for the sex assignment.
The second approach was specifically designed for low-covered shotgun genomes and has been shown to confidently estimate genetic sex for libraries with as little as 1,000 mapping reads. In contrast to the first method, here the average coverage is estimated across the entire X-and the entire autosomal-chromosome sequences of the human reference hg19 (and not on specific positions). The ratio between the X and the autosomal average coverages is calculated and used for the sex assignment as described in [16].

SI Table 3 (continues p. 6).
Matching of sex results based on gross anatomical methods and those supported by aDNA. There was total n = 69, total n of mismatches = 6, total n of matches = 42, which results in 88% success rate of sex estimation using both methods. aDNA was not available (n/a) for n = 21 individuals.

ID Estimated sex
Genetic sex approach 1 (all positions; X/auto ratio) Female n/a n/a n/a B71 Female n/a n/a n/a B103 Female  Female  Female  B163  Female  Female  Female  B159  Female  Female_low_certainty  Female_low_certainty  B6 Female n/a n/a n/a B65 Female Female Female B79 Female n/a n/a n/a B23 Female n/a n/a n/a B15 Female Male Male x B25 Female n/a n/a n/a B38 Female n/a n/a n/a B109 Female  Female_low_certainty  Female  B59  Female  Female  Female  B152  Female  Female  Female  B37 Female n/a n/a n/a B110 Female Female Female B160 Female n/a n/a n/a 105-1 Female n/a n/a n/a 105-2 Female n/a n/a n/a B180 Female  Female  Female  B30  Female  Female  Female  B45 Female n/a n/a n/a B95 Female Female Female B69 Female Female Female