Response to joint selection on germination and flowering phenology depends on the direction of selection

Abstract

Flowering and germination time are components of phenology, a complex phenotype that incorporates a number of traits. In natural populations, selection is likely to occur on multiple components of phenology at once. However, we have little knowledge of how joint selection on several phenological traits influences evolutionary response. We conducted one generation of artificial selection for all combinations of early and late germination and flowering on replicated lines within two independent base populations in the herb Campanula americana. We then measured response to selection and realized heritability for each trait. Response to selection and heritability were greater for flowering time than germination time, indicating greater evolutionary potential of this trait. Selection for earlier phenology, both flowering and germination, did not depend on the direction of selection on the other trait, whereas response to selection to delay germination and flowering was greater when selection on the other trait was in the opposite direction (e.g., early germination and late flowering), indicating a negative genetic correlation between the traits. Therefore, the extent to which correlations shaped response to selection depended on the direction of selection. Furthermore, the genetic correlation between timing of germination and flowering varies across the trait distributions. The negative correlation between germination and flowering time found when selecting for delayed phenology follows theoretical predictions of constraint for traits that jointly determine life history schedule. In contrast, the lack of constraint found when selecting for an accelerated phenology suggests a reduction of the covariance due to strong selection favoring earlier flowering and a shorter life cycle. This genetic architecture, in turn, will facilitate further evolution of the early phenology often favored in warm climates.

Keywords: Campanula americana; Campanulastrum americanum; artificial selection; bivariate selection; correlated response; flowering time; germination time; life history evolution; maternal effects; realized heritability; reproductive phenology.

Figure 1

Schematic of design of selection experiment. A population with relatively early germination and flowering time was crossed to one with late germination and flowering. F1s were grown and crossed to form a variable F2 generation to serve as a base population for selection. Base populations were divided into two replicate lines. Each line was selected for the four combinations of early and late germination and flowering time as well as a control, where individuals were selected at random. Germination and flowering time were scored on offspring of F2 to determine response to selection. The process was repeated for two pairs of populations, creating independent replicates to assess the response to selection

Figure 2

Deviation of the mean number of days to germination and flower of each selection treatment from the mean of the control (±SE) in Campanula americana following one generation of artificial selection for early germination and early flowering (E_gE_f), early germination and late flowering (E_gL_f), late germination and early flowering (L_gE_f), and late germination and late flowering (L_gL_f). Means are the average of two replicate lines for each base population. Asterisks indicate results of contrasts comparing each selection treatment with the control. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001

Figure 3

Realized heritability (±SE) of days to germination and days to flowering in Campanula americana following artificial selection for early germination and early flowering (E_gE_f), early germination and late flowering (E_gL_f), late germination and early flowering (L_gE_f), and late germination and late flowering (L_gL_f) averaged across two replicate lines for each of the base population