Selection maintains a nonadaptive floral polyphenism

Abstract

Adaptive phenotypic plasticity evolves in response to the contrasting selection pressures that arise when organisms face environmental heterogeneity. Despite its importance for understanding how organisms successfully cope with environmental change, adaptive plasticity is often assumed but rarely demonstrated. We study here the adaptive nature of the extreme seasonal within-individual floral polyphenism exhibited by the crucifer Moricandia arvensis, a Mediterranean species that produces two different types of flowers depending on the season of the year. During spring, this species has large, cross-shaped, lilac flowers, while during summer, it develops small, rounded, white flowers. Although floral polyphenism was associated with increased plant fitness, selection moved floral traits away from their local optimum values during the harsh summer. This result strongly suggests that floral polyphenism is not adaptive in M. arvensis. The main factor selecting against floral polyphenism was pollinators, as they select for the same floral morph in all environments. Despite not being adaptive, floral polyphenism occurs throughout the entire distribution range of M. arvensis and has probably been present since the origin of the species. To solve this paradox, we explored the factors causing floral polyphenism, finding that floral polyphenism was triggered by summer flowering. Summer flowering was beneficial because it led to extra seed production and was favored by adaptive plasticity in leaf functional traits. Taken together, our study reveals a complex scenario in which nonadaptive floral polyphenism has been indirectly maintained over M. arvensis evolutionary history by selection operating to favor summer flowering. Our study provides thus strong evidence that nonadaptive plasticity may evolve as a byproduct of colonizing stressful environments.

Keywords: adaptive plasticity; floral plasticity; maladaptation; natural selection; pollinators; within-individual plasticity.

Figure 1.

Fitness gain of plasticity. Fitness gain of plastic individuals quantified as (A) seed production. (B) Fruit production. (C) Ovule production per fruit and (D) Seeds produced per fruit. In (C) and (D), we compared both spring and summer fruits of summer-flowering plants against spring fruits of plants that did not flower during summer and produced flowers only during spring.

Figure 2.

Adaptive value of floral plasticity in Moricandia arvensis. Magnitude and direction of the selection differentials acting on each plant trait during each season (vectors in arrow) and on their reaction norms (spring vectors). The reaction norms are represented as lines joining the phenotypic value in spring and summer. The spring vectors indicates selection acting directly on plasticity, increasing it (raising the slope) or decreasing it (flattening the slope). The length of the vectors indicates the magnitude of the selection, and the color, their statistical significance (gray = nonsignificant, red = significant). The asterisks above each reaction norm indicate significant between-season differences in the magnitude of selection. The color codes in reaction norms are: Blue indicates adaptive plasticity, orange indicates maladaptive plasticity, black indicates adaptatively neutral plasticity, and gray indicates nonsignificant plasticity.

Figure 3.

Selective scenario promoting the evolution of floral plasticity in Moricandia arvensis. Structural equation model relating relative fitness of the plants, the absolute values of the slopes of the reaction norms of floral traits, and the interaction strength of pollinators. Only three functional groups of pollinators (large butterflies, long-tongued large bees and beeflies) were kept in the final model. The model depicted is the definitive model obtained after an iterative process. The number below each trait is the total effect of that trait on fitness ± 1 SE. Dashed lines indicate negative relationships, whereas solid lines indicate positive relationships. Gray lines indicate nonsignificant relationships. Only significant path coefficients are shown (see Supplementary Table S8 for the overall statistical results). *p < 0.05, **p < 0.01, ***p < 0.001, ^ms marginally significant p < 0.1. Line widths are proportional to the magnitude of the effect of each connected variable. Pollinator icons were obtained from divulgare.net under a Creative Common licence.

Figure 4.

Factors shaping summer flowering. Structural equation model exploring the factors promoting summer flowering. Traits included in red ellipses were modeled as latent variables defined by the observed traits connected to them (see Table S10 for the overall statistical results). Dashed lines indicate negative relationships, whereas solid lines indicate positive relationships. Line widths are proportional to the magnitude of the effect of the latent variable on summer flowering or the magnitude of the relationship between traits and their associated latent variable. *p < 0.05, **p < 0.01, ***p < 0.001, ^msmarginally significant p < 0.1. Herbivores are from top to bottom: nectar robbers, seed predators, butterfly chewers, ungulates, large sapsuckers, and leaf chewers. Animal icons were obtained from divulgare.net under a Creative Common licence.