Yoav Ram

Humans routinely keep souvenirs objects retained for the memories they evoke, yet the evolutionary and cognitive consequences of mnemonic collecting remain unclear. Existing discussions of curated objects concentrate on later prehistory and modern humans and view such objects at the social level, as carriers of shared cultural meaning or as markers of status. Here, we address an earlier time period and locate the phenomenon at the personal rather than the social level. By integrating archaeology with theories of extended memory and niche construction, we synthesize evidence consistent with systematic curation of items beyond purely utilitarian needs from the terminal Lower Palaeolithic (~500 kya) onward, including transported distinctive stones, fossils, and the prolonged retention of patinated “old” stone tools whose earlier surfaces were deliberately preserved. We propose that curated keepsakes functioned as portable external memory devices. Beyond cueing recall, they may have helped authenticate self‐relevant past events by providing durable material traces that support source monitoring—distinguishing experienced from imagined events. As such practices became routine, they could have created a “souvenir niche” in which individuals who better integrated internal recollection with external tokens gained fitness benefits. The token's reference is personal, but its consequences can be social, supporting tasks like tracking alliances, reciprocity, and place‐based obligations. These benefits may have strengthened selection on neural systems facilitating the development of autobiographical memory and sense of self via gene–culture coevolution. We outline archaeological, developmental, and experimental predictions that distinguish this mnemonic‐authentication account from aesthetic or economic explanations and suggest approaches for testing them. Viewing curated objects as memory technology links early material culture to the evolutionary shaping of human autobiographical memory.

High-throughput experimental platforms now routinely generate data from dozens or hundreds of independent observations. Simulation-based inference (SBI) offers a powerful framework for estimating model parameters from such complex datasets, but standard methods struggle to scale to the noisy multiple-replicates regime without incurring prohibitive computational costs or careful hyperparameter tuning. Here, we introduce a new method for fast and robust collective posterior inference from multiple independent replicates using a robust product-of-experts aggregation scheme that automatically mitigates the influence of outliers. Evaluating it on synthetic and empirical evolutionary datasets, we find it achieves state-of-the-art estimation accuracy and computational efficiency, including inference from noisy observations. Our method is compatible with any SBI framework, providing a scalable, plug-and-play solution for inference from noisy multiple-replicate datasets.

Copy number variants (CNVs) are DNA duplications and deletions that cause genetic variation, underlying rapid adaptive evolution. CNVs often confer selective advantages but can also incur fitness costs. Evolution of Saccharomyces cerevisiae in nutrient-limited chemostats recurrently selects for amplifications of nutrient transporter genes. However, their fate upon return to a non-selective environment remains unknown. To investigate CNV fitness and stability upon removing the original selection pressure, we studied 15 CNV lineages (11 segmental and 4 whole-chromosomal amplifications) selected in nitrogen-limited chemostats. CNV stability was monitored using fluorescent reporters during propagation in nutrient-rich batch cultures for 110 to 220 generations. All aneuploid lineages showed rapid CNV loss and reversion to a single-copy genotype, whereas segmental amplifications were remarkably stable; one of the 11 strains reverted. Pairwise fitness competitions in rich media revealed strong fitness defects associated solely with CNVs that reverted; reversion led to increased fitness. Using simulation-based inference to estimate reversion rates and fitness effects, we determined negative selection as the primary driver of CNV loss. Whole-genome sequencing revealed that reversion of aneuploids and a segmental amplification left no evidence of prior CNV existence, rendering revertant genomes indistinguishable from the single-copy ancestor. Detailed characterization of a partial revertant identified chromosomal translocation, suggesting that extant CNVs can undergo structural diversification. Our findings provide novel evidence that most segmental CNVs adapted to nitrogen limitation are stable upon removal of selection, but costly gene amplifications are readily reversible. Together, these highlight the importance of CNVs in both long-term genome evolution and rapid, reversible adaptation to transient selection.

Human handedness results from the interplay of genetic and cultural influences. A gene-culture co-evolutionary model for handedness was introduced by Laland et al. (1995), and this study generalizes that model and the related analysis. We address ambiguities in the original methodology, particularly regarding maximum-likelihood estimation, and incorporate sex differences in cultural transmission. By fitting this extended framework to existing familial and twin datasets, we demonstrate that accounting for criterion shifts significantly improves model fit and parameter estimation accuracy. We find stronger maternal than paternal effects on handedness, with daughters exhibiting greater sensitivity to these effects than sons. We provide an open-source Python implementation of the model, which is a robust platform for comparing gene-culture models and applying them to diverse datasets.

High-throughput experimental platforms now routinely generate data from dozens or hundreds of independent observations. Simulation-based inference (SBI) offers a powerful framework for estimating model parameters from such complex datasets, but standard methods struggle to scale to the noisy multiple-replicates regime without incurring prohibitive computational costs or careful hyperparameter tuning. Here, we introduce a new method for fast and robust collective posterior inference from multiple independent replicates using a robust product-of-experts aggregation scheme that automatically mitigates the influence of outliers. Evaluating it on synthetic and empirical evolutionary datasets, we find it achieves state-of-the-art estimation accuracy and computational efficiency, including inference from noisy observations. Our method is compatible with any SBI framework, providing a scalable, plug-and-play solution for inference from noisy multiple-replicate datasets.

Accurate individual identification is essential for monitoring rare amphibians, yet invasive marking is often unsuitable for critically endangered species. We evaluate state-of-the-art computer-vision methods for photographic re-identification of the Hula painted frog (Latonia nigriventer) using 1,233 ventral images from 191 individuals collected during 2013-2020 capture-recapture surveys. We compare deep local-feature matching in a zero-shot setting with deep global-feature embedding models. The local-feature pipeline achieves 98% top-1 closed-set identification accuracy, outperforming all global-feature models; fine-tuning improves the best global-feature model to 60% top-1 (91% top-10) but remains below local matching. To combine scalability with accuracy, we implement a two-stage workflow in which a fine-tuned global-feature model retrieves a short candidate list that is re-ranked by local-feature matching, reducing end-to-end runtime from 6.5-7.8 hours to ~38 minutes while maintaining ~96% top-1 closed-set accuracy on the labeled dataset. Separation of match scores between same- and different-individual pairs supports thresholding for open-set identification, enabling practical handling of novel individuals. We deploy this pipeline as a web application for routine field use, providing rapid, standardized, non-invasive identification to support conservation monitoring and capture-recapture analyses. Overall, in this species, zero-shot deep local-feature matching outperformed global-feature embedding and provides a strong default for photo-identification.

We replicate and critically evaluate McManus’ (1985) single-locus genetic model of handedness, which remains influential in laterality research. Using the original familial and twin datasets, we reproduce McManus’ parameter estimates while correcting reporting errors and miscalculations. Our reproduction confirms that the model is reproducible but reveals sensitivity to dataset inconsistencies and outliers. We extend the framework by generalizing correction matrices, implementing simulations, and applying bootstrap methods to estimate confidence intervals. We further analyze a modern dataset from Flores and Adonara, Indonesia, providing a test of the model in a different cultural setting and comparing triplet versus multi-offspring representations. Our findings emphasize both the historical value and the current limitations of the Dextral–Chance framework, offering a modern foundation for testing genetic and cultural theories of handedness with contemporary datasets.

Copy number variants (CNVs) are DNA duplications and deletions that cause genetic variation, underlying rapid adaptive evolution. CNVs often confer selective advantages, but can also incur fitness costs. Evolution of Saccharomyces cerevisiae in nutrient-limited chemostats recurrently selects for amplifications of nutrient transporter genes. However, their fate upon return to a non-selective environment remains unknown. To investigate CNV fitness and stability upon removing the original selection pressure, we studied 15 CNV lineages (11 segmental, 4 whole-chromosomal amplifications) selected in nitrogen-limited chemostats. CNV stability was monitored using fluorescent reporters during propagation in nutrient-rich batch cultures for 110–220 generations. All aneuploid lineages showed rapid CNV loss and reversion to a single-copy genotype, whereas segmental amplifications were remarkably stable– one of 11 strains reverted. Pairwise fitness competitions in rich media revealed strong fitness defects associated solely with CNVs that reverted; reversion led to increased fitness. Using simulation-based inference to estimate reversion rates and fitness effects, we determined negative selection as the primary driver of CNV loss. Whole-genome sequencing revealed that reversion of aneuploids and a segmental amplification left no evidence of prior CNV existence, rendering revertant genomes indistinguishable from the single-copy ancestor. Detailed characterization of a partial revertant identified chromosomal translocation, suggesting that extant CNVs can undergo structural diversification. Our findings provide novel evidence that most segmental CNVs adapted to nitrogen limitation are stable upon removal of selection, but costly gene amplifications are readily reversible. Together, these highlight the importance of CNVs in both long-term genome evolution and rapid, reversible adaptation to transient selection.

Evolutionary rescue occurs when a population, facing a sudden environmental change that would otherwise lead to extinction, adapts through beneficial mutations, allowing it to recover and persist. A prime example of evolutionary rescue is the ability of cancer to survive exposure to treatment. One evolutionary mechanism by which a population of cancer cells can adapt to chemotherapy is aneuploidy. Aneuploid cancer cells can be more fit in an environment altered by anticancer drugs, in part because aneuploidy may disrupt the pathways targeted by the drugs. Indeed, aneuploidy is highly prevalent in tumors, and some anticancer drugs fight cancer by increasing chromosomal instability. Here, we model the impact of aneuploidy on the fate of a population of cancer cells. We use multitype branching processes to approximate the probability that a tumor survives drug treatment as a function of the initial tumor size, the rates at which aneuploidy and other beneficial mutations occur, and the growth rates of the drug-sensitive and drug-resistant cells. We also investigate the effect of the preexistent aneuploid cells on the probability of evolutionary rescue. Finally, we estimate the tumor’s mean recurrence time to revert to its initial size following treatment and evolutionary rescue. We propose that aneuploidy can play an essential role in the relapse of smaller secondary tumors.

Copy number variants (CNVs) are an important source of genetic variation underlying rapid adaptation and genome evolution. Whereas point mutation rates vary with genomic location and local DNA features, the role of genome architecture in the formation and evolutionary dynamics of CNVs is poorly understood. Previously, we found the GAP1 gene in Saccharomyces cerevisiae undergoes frequent amplification and selection in glutamine-limitation. The gene is flanked by two long terminal repeats (LTRs) and proximate to an origin of DNA replication (autonomously replicating sequence, ARS), which likely promote rapid GAP1 CNV formation. To test the role of these genomic elements on CNV-mediated adaptive evolution, we evolved engineered strains lacking either the adjacent LTRs, ARS, or all elements in glutamine-limited chemostats. Using a CNV reporter system and neural network simulation-based inference (nnSBI) we quantified the formation rate and fitness effect of CNVs for each strain. Removal of local DNA elements significantly impacts the fitness effect of GAP1 CNVs and the rate of adaptation. In 177 CNV lineages, across all four strains, between 26% and 80% of all GAP1 CNVs are mediated by Origin Dependent Inverted Repeat Amplification (ODIRA) which results from template switching between the leading and lagging strand during DNA synthesis. In the absence of the local ARS, distal ones mediate CNV formation via ODIRA. In the absence of local LTRs, homologous recombination can mediate gene amplification following de novo retrotransposon events. Our study reveals that template switching during DNA replication is a prevalent source of adaptive CNVs.