A new model for ancient DNA decay based on paleogenomic meta-analysis

Abstract

The persistence of DNA over archaeological and paleontological timescales in diverse environments has led to a revolutionary body of paleogenomic research, yet the dynamics of DNA degradation are still poorly understood. We analyzed 185 paleogenomic datasets and compared DNA survival with environmental variables and sample ages. We find cytosine deamination follows a conventional thermal age model, but we find no correlation between DNA fragmentation and sample age over the timespans analyzed, even when controlling for environmental variables. We propose a model for ancient DNA decay wherein fragmentation rapidly reaches a threshold, then subsequently slows. The observed loss of DNA over time may be due to a bulk diffusion process in many cases, highlighting the importance of tissues and environments creating effectively closed systems for DNA preservation. This model of DNA degradation is largely based on mammal bone samples due to published genomic dataset availability. Continued refinement to the model to reflect diverse biological systems and tissue types will further improve our understanding of ancient DNA breakdown dynamics.

Figure 1.

Locations of 185 samples (n = 94 unique sites) used in paleogenomic meta-analysis, global variation in mean temperature and temperature fluctuation, and timeline of sample ages. Note the absence of sites with annual mean temperature >20°C, reflecting known preservation bias toward cooler climates (25).

Figure 2.

Relationships between DNA degradation parameters and environmental variables. (A) DNA fragmentation is correlated with thermal fluctuation and precipitation. (B) DNA fragmentation is correlated with thermal fluctuation but is not influenced by sample age. (C) Deamination is a thermal age parameter, strongly associated with both age and temperature. Coloring in (A–C) is used to enhance the z-axis variation: red points are the nearest and blue are the most distant. (D) DNA fragmentation is highly predictive of base compositional biases, with fragmented datasets depleted of motifs with low base-stacking energy. (E) Histone periodicity in fragment length distribution is most pronounced in samples from cold environments. Blue circles represent samples where a histone periodicity estimate was possible (n = 112; see ‘Materials and Methods’ section for calibration against false positive results), red diamonds are samples where no periodicity was observed, visualized at −7.5 on the y-axis to reflect the observation of no detectable bias.

Figure 3.

Expectations of deamination over time for variable temperatures. Left: density-weighted linear regression of temperature and log deamination rate calculated using the formula rate = ln (1/(1−δ_s)) * (1/age). Right: using rate estimates from the weighted regression, we calculated the expected δ_s values over a 100 000 year time span using the formula δ_s = 1−(1/e^rate*age) re-arranged from the above, as well as δ_s values for rates estimated from 95% confidence intervals of the regression. We visualized the expected deamination levels for samples from −20, 0 and 20°C contexts (solid lines), along with upper (red) and lower (blue) confidence bounds. This predictive model is necessarily based primarily on mammalian bone tissue, and we expect refinements to these expectations based on sample type, for example, as more datasets become available.

Figure 4.

Fourteen samples included in the meta-analysis with saturated cytosine-to-uracil deamination in single-stranded overhangs.