Evolution

High investment in current reproduction can limit future reproductive opportunities, but how selection shapes these hormone-mediated traits remains poorly understood. Androgens can mediate male reproductive investment, and in three-spined stickleback (Gasterosteus aculeatus), exert antagonistic effects on breeding effort versus spermatogenesis. To understand how shifts in reproductive strategy shape this tradeoff, we compared testes transcriptomes and androgen production between two recently diverged stickleback ecotypes that differ in reproductive strategy: the ancestral "common" ecotype, which provides paternal care, and the non-parental "white" ecotype, which has lost paternal care and prioritizes mating effort. During typical breeding, testes gene expression differed little between ecotypes. However, under prolonged summer-like conditions, testes gene expression diverged substantially. Common-biased genes were enriched for meiotic functions and spermatogenic cell type markers, suggesting commons had initiated spermatogenesis while whites had not. Instead, whites expressed higher levels of steroidogenic candidate genes and released significantly more 11-ketotestosterone than commons, indicating sustained investment in current reproduction. F1 hybrids released 11-ketosterone at intermediate rates, suggesting a genetic basis for this divergence. Sustained androgen production in whites may possibly delay the transition into spermatogenesis, limiting investment in future reproduction. These results illustrate how selection on hormonally-integrated traits can drive rapid divergence in life history strategy.

Eusocial Hymenoptera present the highest known recombination rates among metazoans, which evolved several times independently among bees, ants and wasps. Several hypotheses have been proposed to explain this observation, including stronger selection for recombination caused by coevolving parasites and pathogens, and strong sexual selection among haploid males due to male-biased sex ratios among reproductive individuals. In this article, we explore the effects of haplodiploidy and differential selection between sexes on the evolution of recombination, by analyzing a three-locus model in which selection for recombination stems from negative epistasis between selected loci. Our analytical predictions are compared with the results of individual-based simulations in which deleterious mutations occur along a linear chromosome. Our results show that, at mutation-selection balance for deleterious alleles, increasing the strength of selection against deleterious alleles (due to the effect of male haploidy and/or sexual selection) tends to reduce selection for recombination. However, an increase in the overall magnitude of negative epistasis (which may also be due to male haploidy and/or sexual selection) combined with the fact that recombination only occurs in females may increase selection for recombination substantially. Our model also shows that, in conditions favoring recombination, increasing recombination in meioses leading to parthenogenetic ovules (and male offspring) may yield stronger benefits than in meioses leading to fertilized ovules (and female offspring).

Does the distribution of fitness effects of new mutations vary across the genome? Under the classical Fisher Geometric Model (FGM) we might not expect it to. In FGM, phenotypic traits are envisioned as dimensions of a landscape, with fitness determined by position in the landscape, i.e., the particular combination of traits of an individual. New mutations are represented by vectors that move from an ancestral to a new phenotype. In classical FGM these vectors affect all trait dimensions simultaneously (universal pleiotropy). However, introducing partial and modular pleiotropy into an FGM framework leads to an expectation that parameters of the DFE will vary with mutational pleiotropy-the number of traits affected by individual mutations. Here we address this prediction by investigating whether traits related to mutational pleiotropy, expression level and network connectivity, affect the parameters of the DFE using whole genome data from A. thaliana and C. grandiflora, two closely related Brassica species that vary significantly in their demography and mating system, and therefore, in effective population size and the effects of linked selection. Results were similar across both species. We found that expression level and network connectivity were predictive of the parameters of the deleterious DFE, even once co-correlations among genome biology traits were accounted for. Our results suggest that, across the genome, molecular evolutio(high mutational pleiotropy). nary patterns agree with the predictions of FGM, albeit relaxing the assumption of universal pleiotropy, and that variation in mutational pleiotropy among genes is sufficient to have detectible effects on the DFE. Significance statementHow do the effects of new mutations vary across the genome? If mutations in some genes affect many traits (high mutational pleiotropy), we hypothesise they will be more strongly deleterious, with lower variance in their selective effects. We test this by investigating the distribution of effects of new mutations across genes that vary in features that are related to mutational pleiotropy: expression level, gene network connectivity, and number of associated GO terms. The mean strength and coefficient of variation of selection of new mutations varied across genes with different features in the manner expected by our hypothesis. This demonstrates that important parameters of molecular evolution can vary across the genome with genome architecture.

The number of chromosome sets per nucleus is a fundamental trait, but why this number is nearly always two for multicellular eukaryotes remains unclear. Chromosomes are made of nucleic acids, which possess abundant phosphorus (P). Therefore, producing new chromosomes, as well as generating new cells and organismal growth, demands substantial phosphorus. Yet, because P is often limiting in nature, P availability could influence the prevalence of diploidy versus polyploidy. Here, we compare growth rates of diploid and triploid Potamopyrgus antipodarum, a freshwater snail, relative to P availability. Because diploid P. antipodarum are obligately sexual while obligately asexual individuals are polyploid, costs associated with sensitivity to P limitation in polyploids could also help explain the maintenance of sexual P. antipodarum. We raised juvenile diploid and triploid snails on either P-adequate or P-deficient diets and found that independent of P availability, juvenile triploid asexual snails grew faster and harbored higher P content as adults than sexual diploid conspecifics. Together, these results suggest life-history advantages of polyploidy or asexual reproduction that exacerbate rather than ameliorate the cost of sex. These outcomes suggest that P availability is unlikely to be a main driver of ploidy polymorphism or the maintenance of sex in P. antipodarum.

In many species with female sperm storage, ejaculates from multiple males overlap in the female reproductive tract, making sperm competitive ability a key component of male reproductive fitness and a target of rapid evolutionary change in the underlying genes. Here, we used controlled laboratory assays of Drosophila melanogaster sperm competition, with doubly-mated females and paternity assignment of offspring, to ask whether a Bradley-Terry framework can effectively summarize and predict competitive outcomes. The Bradley-Terry model is a probabilistic approach that estimates a latent "ability" score for each contestant based on outcomes of pairwise contests, and thus is naturally suited to data from sperm competition, which are intrinsically pairwise. We selected five distinct male genotypes: four carried strongly expressed RFP or GFP markers that allowed us to distinguish their heterozygous offspring under UV illumination, and the fifth was Canton-S, a standard wild-type genotype that served as our reference. Using Canton-S females, we assayed all 20 ordered pairwise combinations of first and second male, recorded successful double matings, and quantified the offspring sired by each male. We then extended the Bradley-Terry model to estimate genotype-specific competitive success separately for first-male "defense" (fertilization success following initial mating, also called "P1") and second-male "offense" (fertilization success following a remating, also called "P2"). This framework provides a flexible and efficient way to integrate results across large arrays of pairwise mating tests and to derive predictive scores for sperm competitive performance.

Sexual conflict has been hypothesized as a driver of speciation, though its effects are likely heterogeneous across phylogenies and between sexes. The semi-aquatic bug, which inhabits water surfaces across diverse aquatic environments, has long served as a model for studying sexual conflict. While previous studies have focused on rapid antagonistic coevolution and the genetic basis of sexually antagonistic traits, the macroevolutionary consequences of asymmetrical sexual conflict--particularly male-dominated grasping traits versus female resistance--remain largely unexplored. Within the subgenus Pseudovelia, males exhibit pronounced phenotypic diversification in grasping structures, whereas females show modest, clade-specific resistance traits, suggesting male-biased asymmetric conflict. This system presents a valuable opportunity to examine how sexual conflict influences diversification and asymmetrical trait evolution across lineages. Using 204 individuals, representing over half of the subgenus's species diversity, we reconstructed a time-calibrated phylogeny, quantified diversification rates, assessed sexual conflict intensity across clades, and analyzed correlations between sexual trait evolution and diversification. Our results reveal extensive phylogenetic conflict, particularly within the East Asian clade, driven by introgression and incomplete lineage sorting (ILS). Furthermore, we observe significant phylogenetic heterogeneity in both phenotypic evolution and diversification rates. Notably, a male "trait package" enhancing grasping ability likely drives rapid diversification in the recently radiated "South China" lineage. In contrast, grasping traits involving abdominal segment VIII are associated with lower conflict intensity, facilitating greater evolutionary flexibility in female resistance and resulting in lineage-specific counter-adaptations. These findings highlight the heterogeneous dynamics of asymmetrical sexual conflict in shaping diversification and speciation.

Climate change can threaten population viability by disrupting sex ratios in species whose sex is influenced by temperature. While species with sex chromosomes were historically considered immune, in some species, temperatures can override genetic sex determination via sex reversal, leaving them vulnerable to climate-driven sex ratio shifts. The Tasmanian spotted snow skink (Carinacincus ocellatus), a viviparous reptile with an XX/XY system, provides a compelling case study. While laboratory studies demonstrated that extreme thermal conditions induce female-to-male sex reversal (XX males), its occurrence in the wild remains unexplored, limiting our understanding of actual climate impacts. Integrating 23 years of phenotypic and genetic sexing data across two climatically distinct populations, we provide the first evidence of sex reversal in a wild viviparous reptile. XX reversal occurred in both populations, affecting up to 23.5% of XX births in the warmer population, and was associated with colder minimum daily temperatures. Despite high birth rates in some years, sex-reversed adults were rare. We also identified putative XY females, suggesting bidirectional sex reversal and reinforcing the extreme plasticity of reptilian sex determination. Ultimately, sex reversal could act as an evolutionary trap, potentially compromising population viability as climate instability increases.

Animals fight for resources to obtain fitness benefits; most contests are intrasexual, and males tend to fight more than females. Although the genetic basis of male aggression is well studied, we know little about the genetic variation of female aggression. Female aggression varies with reproductive status and is potentially influenced not only by her genotype, but also by the genotype of her mate. Here we measured both male and female aggression in a set of Drosophila melanogaster inbred lines by competing each line against a standard competitor. Aggression varied among lines for both sexes, but male and female aggression were not correlated. Female aggression for many lines increased with mating, as expected, but not all lines changed aggression. However, when females were mated to males of different lines, male genotype did not affect the post-mating change in aggression, suggesting that ejaculate-mediated effects do not vary across these lines. The aggression level of the standard opponent was positively correlated with that of focal individuals indicating that individuals modulate their behavior according to the genotype of their opponent.

Hosts defend themselves against parasites through resistance (reducing parasite burden) and tolerance (reducing the fitness cost of infection without affecting parasites). This distinction has important evolutionary implications: resistance is predicted to maintain polymorphism while tolerance tends to fix, and only resistance is expected to provoke parasite counter-adaptation. The reaction-norm framework, which infers tolerance from the slope of host fitness regressed on parasite burden, assumes that a shallow slope reflects parasite-independent host protection. We test this assumption using a within-host model in two variants: microparasites (Model 1, with within-host replication) and macroparasites (Model 2, without). Sublethal immunity impairs the host-exploitation rate of the parasite, reducing both growth and per-parasite virulence without killing them. We show that this generates systematic slope differences among host genotypes that the framework interprets as variation in tolerance. Furthermore, the ranking of slopes across genotypes reverses between linear and sigmoidal damage functions: under linear damage, the strongest immune responder appears most tolerant; under sigmoidal damage, the weakest responder does. Decomposition of the damage reduction shows that virulence reduction accounts for the majority of the effect across both model variants. Thus, the reaction-norm slope cannot determine whether host fitness is maintained by parasite-independent tissue protection or by sublethal impairment of parasites.

Reproductive success requires allocating effort across lifespan in a manner that balances the risk of early mortality against the benefit of higher fecundity or parental expertise that increase with body size or age. Here we report a cross-taxonomic analysis of reproductive schedules in plants, animals, and humans, showing that peak reproductive effort consistently occurs at approximately 1/e (~37%) of species-specific maximum lifespan. The pattern is robust across major phylogenetic groups and independent of absolute lifespan. This convergence is both logically and numerically consistent with the optimal stopping fraction (1/e), which maximizes the probability of selecting a superior option under uncertainty by delaying commitment until 1/e of the available options have been examined. By integrating population dynamics and empirical data with a formal decision-theoretic model, our results suggest a striking previously unrecognized quantitative regularity linking lifespan and reproductive timing. These findings provide a unifying perspective on life-history evolution and suggest that complex biological scheduling strategies are governed by probabilistic principles.

Brain size and structure are hypothesized to be adaptively designed to satisfy the behavioral requirements of securing food and living socially. The importance of these socioecological and sociobiological selective forces in brain evolution is constantly debated. Socioecological divergence is striking in the Neotropical ant genus Neoponera: N. apicalis is a generalist solitary predator forming small colonies of ~100 whereas N. commutata colonies are approximately 10 times as large and workers pheromonally organize cooperatively raids only on Syntermes termite colonies. We interspecifically compared the size and structure of the compound eyes, size and number of antennal glomeruli, mosaic brain scaling and synaptic processing (microglomeruli-MG). Our results indicate that N. apicalis workers have a larger number of ommatidia, antennal lobe glomeruli, and allometrically larger antennal and optic lobes than N. commutata. These sensory traits were associated with differences in higher-order processing architectures in the mushroom body (MB) microglomeruli (MG). N. commutata workers had an allometrically larger MB, perhaps due to their socially complex chemical foraging communication, although MG density in N. apicalis was higher in both the MB lip and collar, regions associated with processing olfactory and visual information, respectively. The increase in MG density in N. apicalis may be associated with higher demands for navigation, learning, and memory, as well as a higher density of antennal lobe glomeruli to support prey odor discrimination. In contrast, N. commutata workers had larger ommatidia and antennal lobe glomeruli. Larger ommatidia correlate with their diurnal/nocturnal habits and a larger MB Our findings indicate that differences in behavioral performance demands associated with socioecological differentiation are reflected in variation in visual and olfactory system structure, brain size, mosaicism, and synaptic organization. Our results support both social and ecological brain hypothesis as drivers of mosaic brain evolution.

Reproductive isolation is a central driver of speciation and can be reinforced by both innate mate preferences and mate preference learning. Pigmentation genes are strong candidates for pleiotropic effects on these processes because they shape visual traits used in mate choice and may also influence neural function, yet their role in learning is poorly understood. Here, we test whether the pigmentation gene yellow affects innate mate preference and aversive mate preference learning in female Bicyclus anynana, and whether these effects are associated with changes in brain dopamine levels. Using CRISPR-Cas9, we knocked out the yellow gene, developed a mutant line, and tested the mate preferences and mate preference learning ability of mutant females compared to wild-type (WT) females. We find that loss of yellow does not alter assortative mating based on pigmentation or disrupt innate visual or olfactory preferences. In contrast, loss of yellow does influence learning ability, as mutant females failed to modify mate preference following aversive premating experience, indicating an impairment in aversive learning. Despite yellows role in the melanin biosynthesis pathway, brain dopamine levels remain unchanged in mutant females relative to WT females. These findings identify yellow as a pleiotropic gene influencing both pigmentation and aversive mate preference learning, providing evidence that pigmentation genes can shape behavioral processes important for reproductive isolation and speciation. Significance statementPigmentation genes are best known for controlling color patterns, but they may also influence behavior through shared neural pathways. Here, we show that the pigmentation gene yellow is required for aversive mate preference learning in the butterfly Bicyclus anynana. Females lacking yellow retain normal innate visual and olfactory mate preferences but fail to learn to avoid previously unattractive mates. Surprisingly, this learning deficit is not associated with altered dopamine levels, suggesting that downstream neural signaling pathways are involved instead. These findings identify a gene that influences both pigmentation and learned mate preference, providing evidence that pigmentation genes can shape behavioral processes important for reproductive isolation and speciation.

Allele frequency (AF) timeseries allow us to directly observe the dynamics of evolution at a genetic level. However, extracting useful inferences from AF timeseries has proved difficult due to the model uncertainties and noisiness inherent in AF change at fine temporal scales. Here we present three new permutation tests --- which do not assume a model of evolutionary change or a parametric statistical model --- to detect AF timeseries features of evolutionary interest. The features identified by these approaches are: 1) any evolutionary change (as opposed to apparent change due to measurement error); 2) directional selection; 3) fluctuating selection with a propensity to change sign (negative autocorrelation). We are not aware of existing tests for features 1 and 3. Feature 2 is commonly tested using standard evolutionary models such as the Wright-Fisher; we show that the permutation approach has comparable statistical power. We apply our new approaches to AF timeseries data from D. melanogaster and D. pulex.

While asexual species can often outcompete their sexual counterparts over ecological timescales, their long-term evolutionary success is hindered by a diminished ability to purge deleterious mutations and to adapt to changing environments. However, some asexual species persist for millions of years, and a major question in evolutionary biology is how they do so. One solution is to occasionally reproduce sexually, as has been shown in a handful of primarily asexual species. Here, we investigate the possibility of rare sex in the clonal raider ant, Ooceraea biroi. We report the whole-genome sequence of a previously uncharacterized clonal line and, using population genetic and phylogenetic analyses, show that it originated through sexual reproduction between two extensively studied clonal lines. The mitochondrial genome of this clonal line differs from that of the maternal clonal line at only a single nucleotide, suggesting that the sexual reproduction event occurred within the past few hundred years. These results demonstrate that sex occurs sporadically in the clonal raider ant, allowing it to generate new genetic combinations and potentially to overcome some of the costs of asexuality.

Shifts in flowering phenology are one of the most well studied plant responses to global climate change. Many studies have documented these shifts and their drivers, including some that describe altered patterns of co-flowering among taxonomically broad species within communities. In comparison, few analyses have examined systematic changes in co-flowering between closely related, interfertile species, where co-flowering can have unique evolutionary consequences. To address such shifts in co-flowering among close relatives, we investigate phenological responses to climate change and its effect on patterns of co-flowering over the past 124 years in 52 species of North America violets (Viola). This genus has many co-occurring species that reproductively interact via shared pollinators and hybridization. We use ~14,000 herbarium records along with environmental and species trait data to model the magnitude of recent flowering phenology shifts, environmental variables and/or species traits associated with these shifts, and resulting changes in co-flowering among species. While both the magnitude and direction of phenological shifts varied among Viola species, nearly half (25/52) show significant changes in flowering day. Regardless of whether flowering was advanced or delayed, flowering date was most consistently associated with local mean temperature. Of six species-level traits, geographical region also significantly predicted flowering shifts, consistent with environment and geography together explaining broad phenological responses across this group. These shifts have produced significant changes to pairwise patterns of co-flowering among species -- ranging from a 59 day increase in co-flowering to complete loss of co-flowering overlap. Sympatric pairs specifically have experienced both increases and decreases in co-flowering, with a geographic pattern of increased co-flowering occurring mainly in eastern US and decreased co-flowering common in western US. Because Viola species are generalist pollinated and already known to hybridize, these new co-flowering patterns could further undermine reproductive barriers among species in this genus.

The general random walk model (GRW) of Hunt (2006) is used to infer directional evolution in mean trait values from sparse fossil data by modeling phenotypic change as the accumulated result of small steps with mean step sizes and step variances. Using simulations and real data cases, Ergon (2026) showed that the step variances can be estimated reasonably well only when the mean trait values have small measurement errors, while for fossil data with realistic measurement errors they appear to be extremely difficult to find, and they are often found to be negative. In the simulations Ergon (2026) assumed that the true phenotypic mean values were known. Here, I essentially repeat these simulations under the assumption that only mean trait values with large measurement errors are known, and based on weighted mean squared error (WMSE) comparisons the conclusion is that weighted least squares (WLS) is a better method than GRW. A second conclusion is that WLS is a better method also in the possibly rare cases with large measurement errors where the GRW parameters are estimated well. The GRW method is simply not flexible enough to handle such cases. A third conclusion is that Akaike Information Criterion (AIC) results for GRW models with large measurement errors relative to the step variance may be overly optimistic.

Empirical studies have demonstrated that mutation rates may change with individual condition, such as in the case of stress-induced mutagenesis. This has led to the hypothesis that condition-dependent (or "plastic") mutation rates could be selectively favoured, as the increased production of new mutants in times of maladaptation enhances evolvability, the ability to undergo adaptive evolution. However, while empirical evidence for condition-dependent mutation rates is accumulating, theoretical models studying their evolution are lacking. Here, we employ an individual-based simulation approach to examine the evolution of condition-dependent mutation rates in a changing environment. We find that condition-dependent mutation rates consistently evolve when the environment changes at an intermediate pace. Furthermore, populations with condition-dependent mutation rates are substantially better adapted to their (changing) environment. Finally, the evolutionary dynamics of condition-dependent mutation rates are both accelerated and destabilised when the mutation rate is self-referential (i.e., when mutator loci affect their own mutation rate). We conclude that condition-dependent mutation rates (and thus evolvability) can readily evolve in changing environments. Significance statementMutation provides the raw material for evolution. Mutation rates thus tune evolvability, the ability to undergo adaptive evolution: if mutation rates are too low, evolution is impeded; if mutation rates are too high, adaptive traits cannot be maintained. Using a theoretical model, we explore the evolution of plastic mutation rates that systematically depend on the condition of the organism and its environment. An example is stress-induced mutagenesis in bacteria, which is implicated in the evolution of antibiotic resistance. We show that plastic mutation rates readily evolve, providing "well-timed" variation specifically when organisms are poorly adapted. Such plastic mutation rates thus facilitate better adaptation to changing environments, and their evolution provides an example of evolvability itself evolving.

Introgression following hybridization is increasingly recognized as a major driver of evolution. However, its importance depends on its frequency in nature, which remains to be quantified. To address this, we provide a snapshot of ongoing introgression in a whole species assemblage (4126 ant colonies). 23% of all 82 local species show signs of introgression, which is more than twice previous estimates. Introgression is typically subtle, yet we find that it contributes measurably to genetic diversity. Species divergence, rather than classical prezygotic reproductive barriers (mating phenology, ecological niche, fine-scale habitat use) constrains introgression, suggesting that the main reproductive barriers are postzygotic at this stage of divergence. Our results indicate that introgression may be a common but often overlooked feature of natural communities.

Many proposed mechanisms for the evolution of cooperation among unrelated individuals rely on relatively demanding cognitive abilities that are not widespread across taxa. In contrast, individual heterogeneity is a pervasive feature of animal groups, encompassing differences in personality as well as physical and cognitive traits. Such heterogeneity can promote the evolution of cooperation, yet its role has received comparatively little attention, particularly as a source of variation giving rise to social organization such as leadership. A specific form of leadership can emerge under unstable environmental conditions, when some individuals become better suited than others to initiate action and influence the behavior of their peers. Unlike fixed dominance hierarchies, emergent leadership can rapidly adjust to changing environmental conditions, thereby reshaping group organization. Because it does not require the maintenance of stable hierarchies, this form of leadership can arise even in species that do not have the cognitive capabilities to sustain complex social structures. In this work, we investigate the combined effects of individual heterogeneity and emergent leadership on the evolution of cooperation using an evolutionary game-theoretic model in which individuals may assume the roles of leaders or followers according to their strength, representing individual differences in suitability to prevailing environmental conditions. We examine different levels of population heterogeneity together with increasingly complex strategy sets requiring progressively greater informational requirements, allowing individuals to condition cooperation on their own strength, leadership role, or both. Our results show that the interplay between leadership and heterogeneity promotes the evolution of cooperation, particularly when only a small fraction of individuals act as leaders. Under these circumstances, cooperation evolves even when individuals employ the simplest possible strategies. Under harsher ecological conditions, cooperation can be sustained by more sophisticated strategies, specifically by conditional strategies that prescribe cooperation when individuals are strong or leading and defect when acting independently. Author summaryIn this study, we propose that emergent leadership mediated by individual diversity can boost the evolution of cooperation in animal groups. Building on growing evidence on the heterogeneity of animal capabilities and personalities, we focus on the fleeting leadership that emerges in animal groups when facing rapidly changing environmental conditions. We suggest that this type of leadership that emerges from individual differences in strength--a generic quality encompassing those characteristics that make an individual more fit to lead in a given situation--does not require complex cognitive capabilities from the animals and represents a valid alternative to more demanding strategies proposed in the past to explain the evolution of cooperation. Using an evolutionary game theory model, we show that if a population includes a few strong players, these can become influential leaders and guide the actions of their peers to achieve cooperation. Although the naive strategy of always cooperating is sufficient for cooperation to evolve, the introduction of more complex strategies leads players to cooperate only when they are more likely to be recognized as influential leaders. These strategies are more effective in promoting cooperation under unfavorable ecological conditions and are also more robust against exploitation by defectors.

The potential for environmental change to compound the detrimental effects of inbreeding depression in small and isolated populations is a significant concern in conservation biology. Previous evidence syntheses suggested that environmental stress exacerbates inbreeding depression, but were based on limited data. Here, we comprehensively test the relationship between inbreeding depression and environmental stress in natural populations using Bayesian mixed-effects meta-analysis on a large, high-quality data set of 2127 inbreeding depression effect sizes from animals and plants. Our results show that inbreeding depression is significantly higher in benign than in stressful environments. Analyses of both inbreeding depression and stress-induced changes in genetic load supported a unimodal (humped) relationship between the costs of inbreeding and stress intensity, with a peak at intermediate stress. At the highest levels of stress there was, on average, a significantly greater inbreeding load in benign than in stressful environments. We suggest that the lower cost of inbreeding associated with extreme stress results from constraints on the expression of inbreeding depression as fitness and phenotypes decline towards zero. Our findings help to resolve long-standing uncertainty around how inbreeding and environmental change interact, revealing that inbreeding responses vary non-linearly with environmental stress intensity, but showing that stress does not generally amplify inbreeding depression. As such, they will inform both the management of populations of conservation concern and predictions of species responses to global environmental change.