Thomas Birley
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Thomas Birley has a total of 14 co-authors. Here are the top co-authors by number of shared publications:
- Cock van Oosterhout
School of Environmental Sciences University of East Anglia, Norwich Research Park Norwich UK
2 shared publications - Hernán E. Morales
GLOBE Institute University of Copenhagen Copenhagen Denmark
2 shared publications - Jim J. Groombridge
Durrell Institute of Conservation and Ecology, Division of Human and Social Sciences, University of Kent, Canterbury, England, UK
2 shared publications - Chiara Bortoluzzi
SIB Swiss Institute of Bioinformatics, Amphipôle, Quartier UNIL-Sorge, Lausanne, 1015, Switzerland
1 shared publication Cock van Oosterhout
School of Environmental Sciences, University of East Anglia, Norwich, Norfolk, United Kingdom
1 shared publication- Gernot Segelbacher
Chair of Wildlife Ecology and Management, Institute of Forest Sciences, Faculty of Environment & Natural Resources, University of Freiburg
1 shared publication
Latest Publications
- Open Access
- 7 October 2026
- Published by Wiley in Evolutionary Applications
- Vol. 19 (10) , e70331
Machine‐Learning Models Trained on Genomic Summary Statistics Assess Extinction Risk and Recovery Potential of Bottlenecked Populations
Forecasting extinction risk from genomic data is challenging because of interactions among environment, demography, and genomic erosion. Here, we use known outcomes from forward simulations as a proof‐of‐concept study to test whether machine‐learning (ML) models can discriminate relative extinction risk and recovery potential from genomic summary statistics, and identify the most informative predictors. Using individual‐based, forward‐in‐time simulations, we generated genomic and fitness data for 11,200 populations across 112 demographic parameter combinations (ancestral population size: 100–10,000, bottleneck size: 4–50, fecundity: 8–64). We evaluated populations at three time points relative to bottleneck onset: 100 years before, 10 years before and at onset. We then assessed their subsequent fate up to 200 years after the end of the bottleneck, classifying populations as extinct or recovered. From 10 sampled individuals per population, we derived 738 predictors spanning diversity, heterozygosity, genetic/realised/masked load, and ROH‐based inbreeding (FROH). Unsupervised analyses primarily captured historical demography and did not separate outcomes. In contrast, supervised models achieved high performance at bottleneck onset, correctly predicting population fate (i.e., future extinction or recovery) from genomics alone with ≈86%–87% accuracy. However, extinction was harder to predict correctly than recovery because simulated extinctions were rarer. Bottleneck size and fecundity provided strong baseline predictability, with genomics adding modest but complementary information. The upper‐quantile ROH/FROH summaries were consistently more informative than mean diversity or mean load. A complementary single‐genome analysis showed that reference‐genome‐style summaries retain useful signal when demographic information is sparse. Together, these results highlight the promise of ML‐informed extinction‐risk assessment to translate genomic indicators into more actionable, early‐warning signals for biodiversity conservation.
- 7 October 2026
- Published by Wiley in Evolutionary Applications
- Vol. 19 (10) , e70331
Open AccessMachine‐Learning Models Trained on Genomic Summary Statistics Assess Extinction Risk and Recovery Potential of Bottlenecked Populations
Forecasting extinction risk from genomic data is challenging because of interactions among environment, demography, and genomic erosion. Here, we use known outcomes from forward simulations as a proof‐of‐concept study to test whether machine‐learning (ML) models can discriminate relative extinction risk and recovery potential from genomic summary statistics, and identify the most informative predictors. Using individual‐based, forward‐in‐time simulations, we generated genomic and fitness data for 11,200 populations across 112 demographic parameter combinations (ancestral population size: 100–10,000, bottleneck size: 4–50, fecundity: 8–64). We evaluated populations at three time points relative to bottleneck onset: 100 years before, 10 years before and at onset. We then assessed their subsequent fate up to 200 years after the end of the bottleneck, classifying populations as extinct or recovered. From 10 sampled individuals per population, we derived 738 predictors spanning diversity, heterozygosity, genetic/realised/masked load, and ROH‐based inbreeding (FROH). Unsupervised analyses primarily captured historical demography and did not separate outcomes. In contrast, supervised models achieved high performance at bottleneck onset, correctly predicting population fate (i.e., future extinction or recovery) from genomics alone with ≈86%–87% accuracy. However, extinction was harder to predict correctly than recovery because simulated extinctions were rarer. Bottleneck size and fecundity provided strong baseline predictability, with genomics adding modest but complementary information. The upper‐quantile ROH/FROH summaries were consistently more informative than mean diversity or mean load. A complementary single‐genome analysis showed that reference‐genome‐style summaries retain useful signal when demographic information is sparse. Together, these results highlight the promise of ML‐informed extinction‐risk assessment to translate genomic indicators into more actionable, early‐warning signals for biodiversity conservation.
- Open Access
Selection mode governs the scaling of genetic load, diversity, and adaptation
The evolutionary consequences of selection depend on whether fitness is measured in absolute terms relative to a threshold (hard selection) or in relative terms among competing individuals (soft selection). Yet general predictions for how selection mode shapes the scaling of genome-wide diversity, genetic load, and adaptive change remain lacking. Using forward-time simulations, we show that hard and soft selection generate fundamentally different scaling relationships between population size, fecundity, nucleotide diversity (π), and genetic load. Under hard selection, increasing carrying capacity elevates both π and load, consistent with the accumulation of mildly deleterious variation at larger effective sizes. Under soft selection, load rapidly approaches an asymptote as intensified within-cohort competition enhances purifying efficiency, while π continues to increase, decoupling neutral diversity from mutational burden. High fecundity (r-strategy) strengthens this decoupling by promoting efficient purging but simultaneously generates extreme variance in reproductive success (“sweepstakes”), reducing effective population size (Ne) relative to census size (N) and limiting π despite large population numbers. Soft selection also enhances adaptive tracking under fluctuating phenotypic optima by avoiding the demographic costs associated with substitution load. Together, these results identify selection mode, alongside life history, as a key determinant of genome-wide load and diversity and provide a mechanistic explanation for why nucleotide diversity scales only weakly with census population size (Lewontin’s paradox).
Open AccessSelection mode governs the scaling of genetic load, diversity, and adaptation
The evolutionary consequences of selection depend on whether fitness is measured in absolute terms relative to a threshold (hard selection) or in relative terms among competing individuals (soft selection). Yet general predictions for how selection mode shapes the scaling of genome-wide diversity, genetic load, and adaptive change remain lacking. Using forward-time simulations, we show that hard and soft selection generate fundamentally different scaling relationships between population size, fecundity, nucleotide diversity (π), and genetic load. Under hard selection, increasing carrying capacity elevates both π and load, consistent with the accumulation of mildly deleterious variation at larger effective sizes. Under soft selection, load rapidly approaches an asymptote as intensified within-cohort competition enhances purifying efficiency, while π continues to increase, decoupling neutral diversity from mutational burden. High fecundity (r-strategy) strengthens this decoupling by promoting efficient purging but simultaneously generates extreme variance in reproductive success (“sweepstakes”), reducing effective population size (Ne) relative to census size (N) and limiting π despite large population numbers. Soft selection also enhances adaptive tracking under fluctuating phenotypic optima by avoiding the demographic costs associated with substitution load. Together, these results identify selection mode, alongside life history, as a key determinant of genome-wide load and diversity and provide a mechanistic explanation for why nucleotide diversity scales only weakly with census population size (Lewontin’s paradox).
- Top Cited PapersOpen Access
- 2 February 2026
- Published by Oxford University Press (OUP) in Journal of Heredity
- Vol. 117 (5) , 1008-1021
Genomic erosion in the assessment of species' extinction risk and recovery potential
Many species are undergoing rapid population declines and environmental deterioration, leading to genomic erosion. Here we define genomic erosion as the loss of genetic diversity, accumulation of deleterious mutations, maladaptation, and introgression, all of which can undermine individual fitness and long-term population viability. Critically, this process continues even after demographic recovery due to a time-lagged impact of genetic drift, which is known as drift debt. Current conservation assessments, such as the International Union for Conservation of Nature Red List, focus on short-term extinction risk and do not capture the long-term consequences of genomic erosion. Likewise, the longer-term assessments of the International Union for Conservation of Nature Green Status may overestimate population recovery by failing to account for the enduring effects of genomic erosion. As genome sequencing becomes increasingly accessible, there is a growing opportunity to quantify genomic erosion and integrate it into conservation planning. Here, we use genomic simulations to illustrate how different genomic metrics are sensitive to the drift debt. We test how ancestral effective population size (Ne) and bottleneck history influence the tempo and severity of genomic erosion. Furthermore, we demonstrate how these dynamics shape genetic load and additive genetic variation, which are key indicators of long-term evolutionary potential. Finally, we present a proof-of-concept for a Genomic Green Status framework that aligns genomic metrics with conservation impact assessments, laying the foundation for genomics-informed strategies to support species recovery.
- 2 February 2026
- Published by Oxford University Press (OUP) in Journal of Heredity
- Vol. 117 (5) , 1008-1021
Top Cited PapersOpen AccessGenomic erosion in the assessment of species' extinction risk and recovery potential
Many species are undergoing rapid population declines and environmental deterioration, leading to genomic erosion. Here we define genomic erosion as the loss of genetic diversity, accumulation of deleterious mutations, maladaptation, and introgression, all of which can undermine individual fitness and long-term population viability. Critically, this process continues even after demographic recovery due to a time-lagged impact of genetic drift, which is known as drift debt. Current conservation assessments, such as the International Union for Conservation of Nature Red List, focus on short-term extinction risk and do not capture the long-term consequences of genomic erosion. Likewise, the longer-term assessments of the International Union for Conservation of Nature Green Status may overestimate population recovery by failing to account for the enduring effects of genomic erosion. As genome sequencing becomes increasingly accessible, there is a growing opportunity to quantify genomic erosion and integrate it into conservation planning. Here, we use genomic simulations to illustrate how different genomic metrics are sensitive to the drift debt. We test how ancestral effective population size (Ne) and bottleneck history influence the tempo and severity of genomic erosion. Furthermore, we demonstrate how these dynamics shape genetic load and additive genetic variation, which are key indicators of long-term evolutionary potential. Finally, we present a proof-of-concept for a Genomic Green Status framework that aligns genomic metrics with conservation impact assessments, laying the foundation for genomics-informed strategies to support species recovery.