Climate-induced range shifts drive adaptive response via spatio-temporal sieving of alleles

Present-day species have endured fluctuating Quaternary climates, marked by range shifts, adaptation, and local extinctions. Past climate changes significantly influenced the distribution of genetic variation, impacting species' adaptive potential. While the biotic impacts of climate change are widely studied, few studies integrate past and present, neutral and adaptive processes. This integrative approach is crucial for understanding and predicting species' evolutionary responses to climate change. Previous research focused on range shifts and expansions, reconstructing past distributions using fossil and contemporary records and inferring past demography from neutral genetic variation. Davis and Shaw (2001) argued that adaptation during range shifts, through selective 'sieving' of genotypes during colonization and establishment, is crucial. Local conditions act as filters sorting standing genetic variation. Adaptation and range shifts acting together lead to different outcomes than either process alone. Empirical evidence supporting this interplay is scarce due to methodological challenges in jointly reconstructing these processes. Recent studies show adaptive responses to climate change can be captured by modelling genotype frequencies against environmental gradients, assuming contemporary gene-environment associations reflect those across time. However, these models typically don't integrate demographic processes like migration and drift. This study uses an integrative approach to study the interplay of adaptation and range shifts in *Dianthus sylvestris*, a species inhabiting environmentally diverse landscapes affected by Quaternary glacial cycles.

The literature review highlights the existing knowledge gap in understanding the combined effects of range shifts and adaptation in response to past climate change. While previous studies have explored range shifts and adaptation separately, the synergistic effects of both processes have received limited attention. The concept of 'selective sieving' of genotypes during range expansions has been proposed, but empirical evidence is lacking due to methodological challenges in jointly reconstructing these processes. Existing studies focusing on gene-environment associations have successfully captured adaptive responses but largely neglected the role of demographic processes. The need for an integrative approach combining neutral and adaptive processes with past and present data is emphasized in this context.

This study employed whole-genome re-sequencing of 1261 individuals from 115 populations of *Dianthus sylvestris* across its contemporary range. The average sequencing depth was approximately 2x. Principal component analysis (PCA) and pairwise genetic distances revealed distinct geographic clusters (Alpine, Apennine, and Balkan). Admixture analyses supported these clusters, although further analysis using chromosome painting suggested recent bottlenecks rather than between-lineage admixture in certain populations. Analysis of the Alpine lineage showed evidence of an eastward origin for the expansion, with a clinal genetic structure mirroring geography. Distribution models, including lineage-specific models, were constructed using contemporary occurrence records and present-day climate and projected to the Last Glacial Maximum (LGM). These models identified three distinct refugia (Alps, Apennines, and Balkans). Gradient forest (GF) models were used to associate SNP allele frequencies with contemporary environments, correcting for structure, focusing on the Alpine lineage. GF models were projected to both present-day and LGM conditions to compare climate adaptation across time. The 'glacial genomic offset' metric was developed to quantify the difference in genomic composition between time points, accounting for expansion and isolation-by-distance (IBD). This metric was validated by correlating it with population genetic statistics capturing site frequency spectrum (SFS) biases, showing a significant excess of high-frequency derived alleles with increasing glacial genomic offset. Nucleotide diversity (π) was compared between low- and high-elevation populations, revealing lower diversity in low-elevation populations, potentially due to founder effects and/or polygenic selection.

The study identified three distinct evolutionary lineages (*D. sylvestris*) separated by geography (Alpine, Apennine, and Balkan), diverging during the Penultimate Glacial-Interglacial Period. Minimal between-lineage migration occurred in the last 115,000 years. The Alpine lineage exhibited a spatial expansion from an eastern origin, with genetic structure mirroring geography. Distribution models identified three glacial refugia, largely aligning with the geographic lineages. Gradient Forest models revealed distinct adaptive genomic compositions for present-day and LGM populations, with low-elevation populations possessing a subset of the adaptive variation present in high-elevation populations. The 'glacial genomic offset' metric positively correlated with population genetic statistics indicative of selective sweeps, confirming the adaptive shifts. Low-elevation populations showed significantly lower nucleotide diversity compared to high-elevation populations, particularly at the expansion front, suggesting limited adaptive potential for future climate change. The interplay of migration, adaptation, and expansion was central to the species' response to past climate shifts.

The findings directly demonstrate the interplay between adaptation and range shifts as a central response to past climate changes, challenging the previously held assumption that range shifts were the primary response. The study's integrative approach, combining neutral and adaptive genetic variation with past and present data, allows for a comprehensive understanding of *D. sylvestris*' response to climate change. The observed decay of adaptive diversity in low-elevation populations highlights the vulnerability of these populations to future selection pressures. While they might be pre-adapted to tolerate further warming, the loss of diversity may be detrimental to responses to other environmental changes and biotic interactions. The integration of historical perspective complements future-oriented approaches by providing evolutionary context.

This research provides novel empirical evidence for the combined influence of migration, adaptation, and expansion in shaping the genetic diversity of *D. sylvestris* across its range. The study highlights the importance of integrating historical and ecological data to understand species responses to climate change and underscores the vulnerability of populations at expansion margins in heterogeneous landscapes. Future research could explore the specific genes underlying adaptive variation, examine the interaction between adaptation and other evolutionary forces, and extend this integrated approach to other species and ecosystems.

The study's reliance on distribution models, assuming niche conservatism, is a potential limitation, although lineage-specific models and population genetic inferences help mitigate this. The relatively low sequencing depth (2x) may have impacted the accuracy of certain analyses, although genotype likelihoods were used to account for uncertainty. The focus on the Alpine lineage limits the generalizability of the findings to other lineages, although similar patterns may exist in other regions. The assumption that contemporary gene-environment associations reflect past associations is an inherent constraint in this kind of retrospective analysis.