Is there plasticity in developmental instability? The effect of daily thermal fluctuations in an ectotherm

Diversified bet‐hedging (DBH) by production of within‐genotype phenotypic variance may evolve to maximize fitness in stochastic environments. Bet‐hedging is generally associated with parental effects, but phenotypic variation may also develop throughout life via developmental instability (DI). This opens for the possibility of a within‐generation mechanism creating DBH during the lifetime of individuals. If so, DI could in fact be a plastic trait itself; if a fluctuating environment indicates uncertainty about future conditions, sensing such fluctuations could trigger DI as a DBH response. However, this possibility has received little empirical attention. Here, we test whether fluctuating environments may elicit such a response in the clonally reproducing crustacean Daphnia magna. Specifically, we exposed genetically identical individuals to two environments of different thermal stability (stable vs. pronounced daily realistic temperature fluctuations) and tested for effects on DI in body mass and metabolic rate shortly before maturation. Furthermore, we also estimated the genetic variation in DI. Interestingly, fluctuating temperatures did not affect body mass, but metabolic rate decreased. We found no evidence for plasticity in DI in response to environmental fluctuations. The lack of plasticity was common to all genotypes, and for both traits studied. However, we found considerable evolvability for DI, which implies a general evolutionary potential for DBH under selection for increased phenotypic variance.     

Evolution of a single niche specialist in variable environments

The pattern (space versus time) and scale (relative to the lifetime of individuals) of environmental variation is thought to play a central role in governing the evolution of the ecological niche and the maintenance of genetic variance in fitness. To evaluate this idea, we serially propagated an initially genetically uniform population of the bacterium Pseudomonas fluorescens for a few hundred generations in environments that differed in both the pattern and scale at which two highly contrasted carbon substrates were experienced. We found that, contrary to expectations, populations often evolved into a single niche specialist adapted to the less-productive substrate in variable environments and that the genetic variance in fitness across different components of the environment was not generally higher in variable environments when compared with constant environments. We provide evidence to suggest that our results reflect a novel constraint on niche evolution imposed by the supply of beneficial mutations available to selection in variable environments.     

Effects of predation environment and food availability on somatic growth in the Livebearing Fish Brachyrhaphis rhabdophora (Pisces: Poeciliidae)

Variation in somatic growth rates is of great interest to biologists because of the relationship between growth and other fitness‐determining traits, and it results from both genetic and environmentally induced variation (i.e. plasticity). Theoretical predictions suggest that mean somatic growth rates and the shape of the reaction norm for growth can be influenced by variation in predator‐induced mortality rates. Few studies have focused on variation in reaction norms for growth in response to resource availability between high‐predation and low‐predation environments. We used juvenile Brachyrhaphis rhabdophora from high‐predation and low‐predation environments to test for variation in mean growth rates and for variation in reaction norms for growth at two levels of food availability in a common‐environment experiment. To test for variation in growth rates in the field, we compared somatic growth rates in juveniles in high‐predation and low‐predation environments. In the common‐environment experiment, mean growth rates did not differ between fish from differing predation environments, but the interaction between predation environment and food level took the form of a crossing reaction norm for both growth in length and mass. Fish from low‐predation environments exhibited no significant difference in growth rate between high and low food treatments. In contrast, fish from high‐predation environments exhibited variation in growth rates between high and low food treatments, with higher food availability resulting in higher growth rates. In the field, individuals in the high‐predation environment grow at a faster rate than those in low‐predation environments at the smallest sizes (comparable to sizes in the common‐environment experiment). These data provide no evidence for evolved differences in mean growth rates between predation environments. However, fish from high‐predation environments exhibited greater plasticity in growth rates in response to resource availability suggesting that predation environments may exhibit increased variation in food availability for prey fish and consequent selection for plasticity.     

Plasticity and evolution in correlated suites of traits

When organisms are faced with new or changing environments, a central challenge is the coordination of adaptive shifts in many different phenotypic traits. Relationships among traits may facilitate or constrain evolutionary responses to selection, depending on whether the direction of selection is aligned or opposed to the pattern of trait correlations. Attempts to predict evolutionary potential in correlated traits generally assume that correlations are stable across time and space; however, increasing evidence suggests that this may not be the case, and flexibility in trait correlations could bias evolutionary trajectories. We examined genetic and environmental influences on variation and covariation in a suite of behavioural traits to understand if and how flexibility in trait correlations influences adaptation to novel environments. We tested the role of genetic and environmental influences on behavioural trait correlations by comparing Trinidadian guppies (Poecilia reticulata) historically adapted to high‐ and low‐predation environments that were reared under native and non‐native environmental conditions. Both high‐ and low‐predation fish exhibited increased behavioural variance when reared under non‐native vs. native environmental conditions, and rearing in the non‐native environment shifted the major axis of variation among behaviours. Our findings emphasize that trait correlations observed in one population or environment may not predict correlations in another and that environmentally induced plasticity in correlations may bias evolutionary divergence in novel environments.     

Effects of assay conditions in life history experiments with Drosophila melanogaster

Selection experiments with Drosophila have revealed constraints on the simultaneous evolution of life history traits. However, the responses to selection reported by different research groups have not been consistent. Two possible reasons for these inconsistencies are (i) that different groups used different environments for their experiments and (ii) that the selection environments were not identical to the assay environments in which the life history traits were measured. We tested for the effect of the assay environment in life history experiments by measuring a set of Drosophila selection lines in laboratories working on life history evolution with Drosophila in Basel, Groningen, Irvine and London. The lines measured came from selection experiments from each of these laboratories. In each assay environment, we measured fecundity, longevity, development time and body size. The results show that fecundity measurements were particularly sensitive to the assay environment. Differences between assay and selection environment in the same laboratory or differences between assay environments between laboratories could have contributed to the differences in the published results. The other traits measured were less sensitive to the assay environment. However, for all traits there were cases where the measurements in one laboratory suggested that selection had an effect on the trait, whereas in other laboratories no such conclusion would have been drawn. Moreover, we provide good evidence for local adaptation in early fecundity for lines from two laboratories.     

Variation in DNA methylation transmissibility,genetic heterogeneity and fecundity‐related traits in natural populations of the perennial herb Helleborus foetidus

Inferences about the role of epigenetics in plant ecology and evolution are mostly based on studies of cultivated or model plants conducted in artificial environments. Insights from natural populations, however, are essential to evaluate the possible consequences of epigenetic processes in biologically realistic scenarios with genetically and phenotypically heterogeneous populations. Here, we explore associations across individuals between DNA methylation transmissibility (proportion of methylation‐sensitive loci whose methylation status persists unchanged after male gametogenesis), genetic characteristics (assessed with AFLP markers), seed size variability (within‐plant seed mass variance), and realized maternal fecundity (number of recently recruited seedlings), in three populations of the perennial herb Helleborus foetidus along a natural ecological gradient in southeastern Spain. Plants (sporophytes) differed in the fidelity with which DNA methylation was transmitted to descendant pollen (gametophytes). This variation in methylation transmissibility was associated with genetic differences. Four AFLP loci were significantly associated with transmissibility and accounted collectively for ~40% of its sample‐wide variance. Within‐plant variance in seed mass was inversely related to individual transmissibility. The number of seedlings recruited by individual plants was significantly associated with transmissibility. The sign of the relationship varied between populations, which points to environment‐specific, divergent phenotypic selection on epigenetic transmissibility. Results support the view that epigenetic transmissibility is itself a phenotypic trait whose evolution may be driven by natural selection, and suggest that in natural populations epigenetic and genetic variation are two intertwined, rather than independent, evolutionary factors.     

Seedling response to water stress in valley oak (Quercus lobata) is shaped by different gene networks across populations

Drought is a major stress for plants, creating a strong selection pressure for traits that enable plant growth and survival in dry environments. Many drought responses are conserved species‐wide responses, while others vary among populations distributed across heterogeneous environments. We tested how six populations of the widely distributed California valley oak (Quercus lobata) sampled from contrasting climates would differ in their response to soil drying relative to well‐watered controls in a common environment by measuring ecophysiological traits in 93 individuals and gene expression (RNA‐seq) in 42 individuals. Populations did not differ in their adjustment of turgor loss point during soil drying, suggesting a generalized species‐wide response. Differential expression analysis identified 689 genes with a common response to treatment across populations and 470 genes with population‐specific responses. Weighted gene co‐expression network analysis (WGCNA) identified groups of genes with similar expression patterns that may be regulated together (gene modules). Several gene modules responded differently to water stress among populations, suggesting regional differences in gene network regulation. Populations from sites with a high mean annual temperature responded to the imposed water stress with significantly greater changes in gene module expression, indicating that these populations may be locally adapted to respond to drought. We propose that this variation among valley oak populations provides a mechanism for differential tolerance to the increasingly frequent and severe droughts in California.     

Genetic basis of between‐individual and within‐individual variance of docility

Between‐individual variation in phenotypes within a population is the basis of evolution. However, evolutionary and behavioural ecologists have mainly focused on estimating between‐individual variance in mean trait and neglected variation in within‐individual variance, or predictability of a trait. In fact, an important assumption of mixed‐effects models used to estimate between‐individual variance in mean traits is that within‐individual residual variance (predictability) is identical across individuals. Individual heterogeneity in the predictability of behaviours is a potentially important effect but rarely estimated and accounted for. We used 11 389 measures of docility behaviour from 1576 yellow‐bellied marmots (Marmota flaviventris) to estimate between‐individual variation in both mean docility and its predictability. We then implemented a double hierarchical animal model to decompose the variances of both mean trait and predictability into their environmental and genetic components. We found that individuals differed both in their docility and in their predictability of docility with a negative phenotypic covariance. We also found significant genetic variance for both mean docility and its predictability but no genetic covariance between the two. This analysis is one of the first to estimate the genetic basis of both mean trait and within‐individual variance in a wild population. Our results indicate that equal within‐individual variance should not be assumed. We demonstrate the evolutionary importance of the variation in the predictability of docility and illustrate potential bias in models ignoring variation in predictability. We conclude that the variability in the predictability of a trait should not be ignored, and present a coherent approach for its quantification.     

Quantifying natural seasonal variation in mutation parameters with mutation accumulation lines

Mutations create novel genetic variants, but their contribution to variation in fitness and other phenotypes may depend on environmental conditions. Furthermore, natural environments may be highly heterogeneous. We assessed phenotypes associated with survival and reproductive success in over 30,000 plants representing 100 mutation accumulation lines of Arabidopsis thaliana across four temporal environments at a single field site. In each of the four assays, environmental variance was substantially larger than mutational variance. For some traits, whether mutational variance was significantly varied between seasons. The founder genotype had mean trait values near the mean of the distribution of the mutation accumulation lines in all field experiments. New mutations also contributed more phenotypic variation than would be predicted, given phenotypic and sequence‐level divergence among natural populations of A. thaliana. The combination of large environmental variance with a mean effect of mutation near zero suggests that mutations could contribute substantially to standing genetic variation.     

ACROSS‐ENVIRONMENT GENETIC CORRELATIONS AND THE FREQUENCY OF SELECTIVE ENVIRONMENTS SHAPE THE EVOLUTIONARY DYNAMICS OF GROWTH RATE IN IMPATIENS CAPENSIS

Trade‐offs can exist within and across environments, and constrain evolutionary trajectories. To examine the effects of competition and resource availability on trade‐offs, we grew individuals of recombinant inbred lines of Impatiens capensis in a factorial combination of five densities with two light environments (full light and neutral shade) and used a Bayesian logistic growth analysis to estimate intrinsic growth rates. To estimate across‐environment constraints, we developed a variance decomposition approach to principal components analysis, which accounted for sample size, model‐fitting, and within‐RIL variation prior to eigenanalysis. We detected negative across‐environment genetic covariances in intrinsic growth rates, although only under full‐light. To evaluate the potential importance of these covariances, we surveyed natural populations of I. capensis to measure the frequency of different density environments across space and time. We combined our empirical estimates of across‐environment genetic variance–covariance matrices and frequency of selective environments with hypothetical (yet realistic) selection gradients to project evolutionary responses in multiple density environments. Selection in common environments can lead to correlated responses to selection in rare environments that oppose and counteract direct selection in those rare environments. Our results highlight the importance of considering both the frequency of selective environments and the across‐environment genetic covariances in traits simultaneously.     

Selection and inheritance of sexually dimorphic juvenile plumage coloration

Sexually dimorphic plumage coloration is widespread in birds and is generally thought to be a result of sexual selection for more ornamented males. Although many studies find an association between coloration and fitness related traits, few of these simultaneously examine selection and inheritance. Theory predicts that sex‐linked genetic variation can facilitate the evolution of dimorphism, and some empirical work supports this, but we still know very little about the extent of sex linkage of sexually dimorphic traits. We used a longitudinal study on juvenile Florida scrub‐jays (Aphelocoma coerulescens) to estimate strength of selection and autosomal and Z‐linked heritability of mean brightness, UV chroma, and hue. Although plumage coloration signals dominance in juveniles, there was no indication that plumage coloration was related to whether or not an individual bred or its lifetime reproductive success. While mean brightness and UV chroma are moderately heritable, hue is not. There was no evidence for sex‐linked inheritance of any trait with most of the variation explained by maternal effects. The genetic correlation between the sexes was high and not significantly different from unity. These results indicate that evolution of sexual dimorphism in this species is constrained by low sex‐linked heritability and high intersexual genetic correlation.     

Reproductive and post‐reproductive life history of wild‐caught Drosophila melanogaster under laboratory conditions

The life history of the fruit fly (Drosophila melanogaster) is well understood, but fitness components are rarely measured by following single individuals over their lifetime, thereby limiting insights into lifetime reproductive success, reproductive senescence and post‐reproductive lifespan. Moreover, most studies have examined long‐established laboratory strains rather than freshly caught individuals and may thus be confounded by adaptation to laboratory culture, inbreeding or mutation accumulation. Here, we have followed the life histories of individual females from three recently caught, non‐laboratory‐adapted wild populations of D. melanogaster. Populations varied in a number of life‐history traits, including ovariole number, fecundity, hatchability and lifespan. To describe individual patterns of age‐specific fecundity, we developed a new model that allowed us to distinguish four phases during a female's life: a phase of reproductive maturation, followed by a period of linear and then exponential decline in fecundity and, finally, a post‐ovipository period. Individual females exhibited clear‐cut fecundity peaks, which contrasts with previous analyses, and post‐peak levels of fecundity declined independently of how long females lived. Notably, females had a pronounced post‐reproductive lifespan, which on average made up 40% of total lifespan. Post‐reproductive lifespan did not differ among populations and was not correlated with reproductive fitness components, supporting the hypothesis that this period is a highly variable, random ‘add‐on’ at the end of reproductive life rather than a correlate of selection on reproductive fitness. Most life‐history traits were positively correlated, a pattern that might be due to genotype by environment interactions when wild flies are brought into a novel laboratory environment but that is unlikely explained by inbreeding or positive mutational covariance caused by mutation accumulation.     

Evolutionary ecology of egg size and number in a seed beetle: genetic trade-off differs between environments

Abstract In many organisms, large offspring have improved fitness over small offspring, and thus their size is under strong selection. However, due to a trade-off between offspring size and number, females producing larger offspring necessarily must produce fewer unless the total amount of reproductive effort is unlimited. Because differential gene expression among environments may affect genetic covariances among traits, it is important to consider environmental effects on the genetic relationships among traits. We compared the genetic relationships among egg size, lifetime fecundity, and female adult body mass (a trait linked to reproductive effort) in the seed beetle, Stator limbatus , between two environments (host-plant species Acacia greggii and Cercidium floridum ). Genetic correlations among these traits were estimated through half-sib analysis, followed with artificial selection on egg size to observe the correlated responses of lifetime fecundity and female body mass. We found that the magnitude of the genetic trade-off between egg size and lifetime fecundity differed between environments–a strong trade-off was estimated when females laid eggs on C. floridum seeds, yet this trade-off was weak when females laid eggs on A. greggii seeds. Also differing between environments was the genetic correlation between egg size and female body mass–these traits were positively genetically correlated for egg size on A. greggii seeds, yet uncorrelated on C. floridum seeds. On A. greggii seeds, the evolution of egg size and traits linked to reproductive effort (such as female body mass) are not independent from each other as commonly assumed in life-history theory.     

Age and sex affect quantitative genetic parameters for dominance rank and aggression in free‐living greylag geese

Knowledge of the genetic and environmental influences on a character is pivotal for understanding evolutionary changes in quantitative traits in natural populations. Dominance and aggression are ubiquitous traits that are selectively advantageous in many animal societies and have the potential to impact the evolutionary trajectory of animal populations. Here we provide age‐ and sex‐specific estimates of additive genetic and environmental components of variance for dominance rank and aggression rate in a free‐living, human‐habituated bird population subject to natural selection. We use a long‐term data set on individually marked greylag geese (Anser anser) and show that phenotypic variation in dominance‐related behaviours contains significant additive genetic variance, parental effects and permanent environment effects. The relative importance of these variance components varied between age and sex classes, whereby the most pronounced differences concerned nongenetic components. In particular, parental effects were larger in juveniles of both sexes than in adults. In paired adults, the partner's identity had a larger influence on male dominance rank and aggression rate than in females. In sex‐ and age‐specific estimates, heritabilities did not differ significantly between age and sex classes. Adult dominance rank was only weakly genetically correlated between the sexes, leading to considerably higher heritabilities in sex‐specific estimates than across sexes. We discuss these patterns in relation to selection acting on dominance rank and aggression in different life history stages and sexes and suggest that different adaptive optima could be a mechanism for maintaining genetic variation in dominance‐related traits in free‐living animal populations.     

LIFE-CYCLE COMPONENTS OF SELECTION IN ERIGERON ANNUUS: II. GENETIC VARIATION

Genetic variation for seedling and adult fitness components was measured under natural conditions to determine the relative importance of the seedling stage for lifetime fitness in Erigeron annuus. Variation in lifetime reproductive success can result from both the persistent effects of genetic variation expressed among seedlings and from variation in adult fitness components. Analysis of covariance was used to separate the stage specific from the cumulative effects of genetic variance expressed earlier in the life cycle. E. annuus produces seeds through apomixis, which allowed measurement of the fitness of replicate genotypes from germination through the entire life cycle. There were significant differences among genotypes for date of emergence, seedling size, survivorship and fecundity, but heritabilities were low, indicating slow response to selection. For all characters, environmental components of variance were one to two orders of magnitude larger than genetic variance components, resulting in broad sense heritabilities less than 0.1. For seedling size and fecundity, all of the genetic variance was in the form of genotype-environment interactions, often with large negative genetic correlations across environments. In contrast, genotypes differed in mean survivorship through one year, but there were no genotype-environment interactions for viability. Genetic differences in viability were primarily expressed as differences in overwinter survivorship. Genotype × environment interactions among sites and blocks were generated early in the life cycle while the genotype × environment interactions in response to competitive environment (open, annual cover, perennial cover) first appeared in adult fecundity. Genetic variation in lifetime fitness was not significant, despite a fourfold difference in mean fitness among genotypes.     

Quantifying the genetic component of phenotypic variation in unpedigreed wild plants: tailoring genomic scan for within-population use

The study of adaptive genetic variation in natural populations is central to evolutionary biology. Quantitative genetics methods, however, are hardly applicable to long-lived organisms, and current knowledge on adaptive genetic variation in wild plants mostly refers to annuals and short-lived perennials. Studies on long-lived species are essential to explore possible life-history correlates of genetic variation, selection, and trait heritability. In this paper, we propose a method based on molecular markers to quantify the genetic basis of individual phenotypic differences in wild plants under natural conditions. Rather than focusing on inferring individual relatedness to estimate the heritability of phenotypic traits, we directly estimate the proportion of observed phenotypic variance that is statistically accounted for by genotypic differences between individuals. This is achieved by (i) identifying loci that are correlated across individuals with the phenotypic trait of interest by means of an amplified fragment length polymorphism (AFLP)-based explorative genomic scan, and (ii) fitting multiple regression and linear random effect models to estimate the effects of genotype, environment and genotype × environment on phenotypes. We apply this method to estimate genotypic and environmental effects on cumulative maternal fecundity in a wild population of the long-lived Viola cazorlensis monitored for 20 years. Results show that between 56–63% (depending on estimation method) of phenotypic variance in fecundity is accounted for by genotypic differences in 11 AFLP loci that are significantly related to fecundity. Genotype × environment effects accounted for 38% of fecundity variance, which may help to explain the unexpectedly high levels of genetic variance for fecundity found.     

The role of the tolerance–fecundity trade‐off in maintaining intraspecific seed trait variation in a widespread dimorphic herb

Coexistence of species with different seed sizes is a long‐standing issue in community ecology, and a trade‐off between fecundity and stress tolerance has been proposed to explain co‐occurrence in heterogeneous environments. Here we tested an intraspecific extension of this model: whether such trade‐off also explains seed trait variation among populations of widespread plants under stress gradients. We collected seeds from 14 populations of Plantago coronopus along the Atlantic coast in North Africa and Europe. This herb presents seed dimorphism, producing large basal seeds with a mucilaginous coat that facilitates water absorption (more stress tolerant), and small apical seeds without coats (less stress tolerant). We analysed variation among populations in number, size and mucilage production of basal and apical seeds, and searched for relationships between local environment and plant size. Populations under higher stress (higher temperature, lower precipitation, lower soil organic matter) had fewer seeds per fruit, higher predominance of basal relative to apical seeds, and larger basal seeds with thicker mucilaginous coats. These results strongly suggest a trade‐off between tolerance and fecundity at the fruit level underpins variation in seed traits among P. coronopus populations. However, seed production per plant showed the opposite pattern to seed production per fruit, and seemed related to plant size and other life‐cycle components, as an additional strategy to cope with environmental variation across the range. The tolerance–fecundity model may constitute, under stress gradients, a broader ecological framework to explain trait variation than the classical seed size–number compromise, although several fecundity levels and traits should be considered to understand the diverse strategies of widespread plants to maximise fitness in each set of local conditions.     

Characterizing selection in black‐throated blue warblers using a sexual network approach

Our understanding of trait evolution is built upon studies that examine the correlation between traits and fitness, most of which implicitly assume all individuals experience similar selective environments. However, accounting for differences in selective pressures, such as variation in the social environment, can advance our understanding of how selection shapes individual traits and subsequent fitness. In this study, we test whether variation in the social environment affects selection on individual phenotype. We apply a new sexual network framework to quantify each male's social environment as the mean body size of his primary competitors. We test for direct and social selection on male body size using a 10‐year data set on black‐throated blue warblers (Setophaga caerulescens), a territorial species for which body size is hypothesized to mediate competition for mates. We found that direct selection on body size was weak and nonsignificant, as was social selection via the body size of the males' competitors. Analysing both types of selection simultaneously allows us to firmly reject a role for body size in competitive interactions between males and subsequent male fitness in this population. We evaluate the application of the sexual network approach to empirical data and suggest that other phenotypic traits such as song characteristics and plumage may be more relevant than body size for male–male competition in this small passerine bird.     

Detection of major loci associated with the variation of 18 important agronomic traits between Solanum pimpinellifolium and cultivated tomatoes

Wild species can be used to improve various agronomic traits in cultivars; however, a limited understanding of the genetic basis underlying the morphological differences between wild and cultivated species hinders the integration of beneficial traits from wild species. In the present study, we generated and sequenced recombinant inbred lines (RILs, 201 F₁₀ lines) derived from a cross between Solanum pimpinellifolium and Solanum lycopersicum tomatoes. Based on a high‐resolution recombination bin map to uncover major loci determining the phenotypic variance between wild and cultivated tomatoes, 104 significantly associated loci were identified for 18 agronomic traits. On average, these loci explained ~39% of the phenotypic variance of the RILs. We further generated near‐isogenic lines (NILs) for four identified loci, and the lines exhibited significant differences for the associated traits. We found that two loci could improve the flower number and inflorescence architecture in the cultivar following introgression of the wild‐species alleles. These findings allowed us to construct a trait–locus network to help explain the correlations among different traits based on the pleiotropic or linked loci. Our results provide insights into the morphological changes between wild and cultivated tomatoes, and will help to identify key genes governing important agronomic traits for the molecular selection of elite tomato varieties.     

Evolutionary transition between bee pollination and hummingbird pollination in Salvia: Comparing means,variances and covariances of corolla traits

Covariation among traits can modify the evolutionary trajectory of complex structures. This process is thought to operate at a microevolutionary scale, but its long‐term effects remain controversial because trait covariation can itself evolve. Flower morphology, and particularly floral trait (co)variation, has been envisioned as the product of pollinator‐mediated selection. Available evidence suggests that major changes in pollinator assemblages may affect the joint expression of floral traits and their phenotypic integration. We expect species within a monophyletic lineage sharing the same pollinator type will show not only similarity in trait means but also similar phenotypic variance‐covariance structures. Here, we tested this expectation using eighteen Salvia species pollinated either by bees or by hummingbirds. Our findings indicated a nonsignificant multivariate phylogenetic signal and a decoupling between means and variance‐covariance phenotypic matrices of floral traits during the evolution to hummingbird pollination. Mean trait value analyses revealed significant differences between bee‐ and hummingbird‐pollinated Salvia species although fewer differences were detected in the covariance structure between groups. Variance‐covariance matrices were much more similar among bee‐ than hummingbird‐pollinated species. This pattern is consistent with the expectation that, unlike hummingbirds, bees physically manipulate the flower, presumably exerting stronger selection pressures favouring morphological convergence among species. Overall, we conclude that the evolution of hummingbird pollination proceeded through different independent transitions. Thus, although the evolution of hummingbird pollination led to a new phenotypic optimum, the process involved the diversification of the covariance structure.