Environments influence the expression of phenotypes of individuals, their progeny, and even their grandprogeny. The duration of environmental effects and how they are modified by subsequent environments are predicted to be targets of natural selection in variable environments.  However, little is known about the genetic basis of the temporal persistence of environmental effects and their stability of expression across subsequent environments, or even the extent to which natural genotypes differ in these attributes of environmental effects.  We factorially manipulated the thermal environment experienced in three successive generations, to quantify the temporal persistence and environmental stability of temperature effects in contrasting genotypes of Arabidopsis thaliana.  We found that genotypes differed in the manner in which environmental effects dissipated across successive generations, the manner in which responses to ancestral environments were stably expressed in present environments, the manner in which ancestral environments altered responses to present environments, and in the manner in which ancestral environments altered fitness in present conditions.  Genetic variation exists in nature for these trait-specific environmental responses, suggesting that the temporal persistence and stability of environmental effects in variable environments have the potential to evolve in response to natural selection imposed by different environments and sequences of environments.

We selected six accessions ("genotypes") of A. thaliana (Supplemental Table S1) that have exhibited variation in phenological (germination and flowering) responses to environmental conditions, and temperature in particular.  These genotypes have been used as founding parents for mapped recombinant inbred lines and have been sequenced in the “1001 Genomes” project (Reiter et al. 1992; Schiff et al. 2001; O’Niel et al. 2008; Bentsink et al. 2010).  The genotypes used are Ag0, Col0, Cvi1, Ler2, Wa1, and Ws1.  See Table S1 for accession numbers and additional information on their location of collection.  Replicates of each genotype were grown in common conditions, with 12h light/12h dark at 22⁰C in replicate growth chambers (Environmental Growth Chambers, Model E7-2, Ohio, USA), in Metromix 360 (Sun Gro, Agawam, MA, USA) in 2-inch plastic pots after three days of cold (4⁰C) stratification to break dormancy.  Mature seeds were harvested and stored in cryo-boxes in a desiccator cabinet at room temperature until use.  Equal numbers of seeds from replicates within the same genotype and treatments were pooled to form the seed pool for the subsequent generation.  

Experimental design and plant growth conditions:

We employed a factorial manipulation of diurnal temperature regime across three generations (Fig. 1A).  This design allowed us to compare the magnitude of effects of diurnal temperature regime in three generations:  present, parental, and grandparental (Fig. 1B-1).  It also allowed us to test whether the response to environments experienced in past generations depends on present environmental conditions (Fig. 1B-2), that is, whether ancestral environmental effects are stably expressed over environments experienced subsequently.  In addition, the design allows us to test whether responding to present environmental conditions depends on ancestral environmental conditions (Fig. 1B-3) and specifically whether ancestral environments reinforce (magnify) or antagonize responses to present environments (Fig. 1B-4).

We used two diurnal 12h light/12h dark thermal regimes, resembling thermal conditions experienced during the growing season of A. thaliana at different locations in its native range, and also resembling thermal conditions in temperate climates that are likely to be experienced during spring and autumn, when seeds are known to germinate. While other aspects of the environment also vary seasonally (day length, precipitation, canopy structure, presence of competitors and herbivores, and many other factors), we have focused on diurnal thermal regime since it is known to strongly influence seasonal developmental timing in A. thaliana. The "High-temperature" thermal regime cycled between 24⁰C day and16⁰C night; the "Low-temperature" thermal regime cycled between18⁰C day and 10⁰C night.  For each planting, twelve replicates from each genotype were planted in 2-inch plastic pots, as described above, with three replicates distributed over four chamber compartments (blocks), using the same chambers as described above.  Plants were grown until senescence, defined as the maturation of  >90% of siliques, and their seeds were then harvested and stored in a desiccator cabinet.  Each generation, seeds from each genotype were pooled across blocks to seed the next generation.  In the final (third) generation, 12 replicates of each of the eight environmental sequences were planted, with three replicates of each genotype from each ancestral treatment in each of the four growth chambers used, for a total of 576 plants (12 replicates x 8 treatments x 6 genotypes).  

Phenotypes measured:

In the third generation, the following phenological traits were measured in daily censuses:  the timing of seed germination, the timing of bolting (the initiation of reproduction), the timing of flowering, and the timing of fruit maturation.  From these, the time intervals between germination and bolting, bolting and flowering, and flowering and fruit maturation were calculated.  Two metrics of size at reproduction were also measured:  the number of leaves at the time of bolting ("number of leaves"), and the length of the largest leaf at the time of bolting ("leaf length").  As a measure of fitness, all seed were collected and weighed, and the total seed biomass was measured with a microbalance. 

Germination assays:

To estimate the proportion of seed germination in the third generation, fresh seeds (fewer than 10 days after harvest) were assayed on 0.7% agar on 35mm plastic petri dishes. Twelve replicate plates were seeded from each genotype and each ancestral environmental combination for each germination temperature regime (same as that used to grow the plants).  Twenty seeds were arranged in a grid and checked for germination every 4 days for 24 days, after which point germination plateaued (or earlier), for a total of 240 seeds per genotype per treatment. Germination was recorded as radicle protrusion from the seed coat, and seeds were checked for viability at the end of 24 days by assessing firmness to touch.  The proportion of viable seeds in each plate that germinated by the end of the experiment was used for statistical analyses.