Locally adaptive temperature response of vegetative growth in Arabidopsis thaliana

  1. Pieter Clauw  Is a corresponding author
  2. Envel Kerdaffrec
  3. Joanna Gunis
  4. Ilka Reichardt-Gomez
  5. Viktoria Nizhynska
  6. Stefanie Koemeda
  7. Jakub Jez
  8. Magnus Nordborg  Is a corresponding author
  1. Gregor Mendel Institute of Molecular Plant Biology, Austrian Academy of Sciences, Vienna BioCenter, Austria
  2. Department of Biology, University of Fribourg, Switzerland
  3. Max Planck Institute of Molecular Cell Biology and Genetics, Germany
  4. Plant Sciences Facility, Vienna BioCenter Core Facilities GmbH, Austria

Abstract

We investigated early vegetative growth of natural Arabidopsis thaliana accessions in cold, nonfreezing temperatures, similar to temperatures these plants naturally encounter in fall at northern latitudes. We found that accessions from northern latitudes produced larger seedlings than accessions from southern latitudes, partly as a result of larger seed size. However, their subsequent vegetative growth when exposed to colder temperatures was slower. The difference was too large to be explained by random population differentiation, and is thus suggestive of local adaptation, a notion that is further supported by substantial transcriptome and metabolome changes in northern accessions. We hypothesize that the reduced growth of northern accessions is an adaptive response and a consequence of reallocating resources toward cold acclimation and winter survival.

Editor's evaluation

The article combines genetic and phenotypic approaches to show convincing evidence of local adaptation in early vegetative growth of Arabidopsis lineages sampled from a wide range of locations. The authors show larger initial size and slower growth of northern accessions compared to southern accessions when exposed to cold temperatures, suggesting that northern accessions potentially reallocate resources for winter survival. This study is commendable for its scope and comprehensive analysis of local adaptation of a highly polygenic trait in a model weed.

https://doi.org/10.7554/eLife.77913.sa0

Introduction

Plants use a wide variety of life history strategies in adaptation to their local environment. These strategies have evolved to maximize fitness, but are constrained by trade-offs between components such as growth, survival, and reproduction (Lande, 1982; Stearns, 1992). While most life history studies investigate differences between species, there is also variation found within species, including in Arabidopsis thaliana, where life history variation has been linked to climate parameters (Estarague et al., 2022; Vasseur et al., 2018; Sartori et al., 2019). Less clear is how trade-offs are shaping this variation. In this article, we consider vegetative growth, a key component of life history, and use transcriptome and metabolome data to help explore potential trade-offs. We did this specifically in cold temperatures meant to simulate natural conditions in the northern regions of the species distribution.

Local adaptation studies in A. thaliana have found important roles for life history traits such as seed dormancy and flowering time (Takou et al., 2019). Temperature is a major regulator of these traits and local populations are adapted to their local climate (Martínez-Berdeja et al., 2020; Simpson and Dean, 2002; Hepworth et al., 2018). Plant growth is also affected by temperature, and previous studies have detected genetic variation underlying growth-related traits (Bac-Molenaar et al., 2015; Marchadier et al., 2019), as well as signals of polygenic adaptation (Wieters et al., 2021). How vegetative growth is adapted to local temperatures remains unclear, however.

Growth can be seen as the end sum of a vast number of physiological processes. All of these are genetically determined but can also be heavily influenced by environmental conditions (Bac-Molenaar et al., 2015; Fritz et al., 2018). Growth is therefore not only genetically a complex trait but also enormously plastic. The most straightforward environmental effect is when conditions are so adverse that growth reaches a physiological limit, making it impossible for the plant to grow any further. This is called ‘passive plasticity’ (Forsman, 2015; van Kleunen and Fischer, 2005). Yet, when survival is at stake, it may also be in the interest of the plant to actively inhibit growth upon deteriorating environmental conditions (Claeys and Inzé, 2013), called ‘active plasticity’ (Forsman, 2015; van Kleunen and Fischer, 2005). Since vegetative growth ultimately determines photosynthetic surface and thus energy input that can be invested in the next generation, it is directly related to fitness, and is typically in trade-off with survival. Allocation of resources towards either growth or survival is thus an important balance to keep, and plants are expected to be adapted to constantly perceiving and responding to specific environmental changes as cues for looming adverse conditions.

Cold acclimation is a well-studied mechanism in plants, in which decreasing temperatures induce freezing tolerance in preparation for winter (Thomashow, 1999; Hughes and Dunn, 1996). This temperature response is typically studied at 4°C, but has been observed in temperatures up to 12°C (Bond et al., 2011). The increased freezing tolerance is accomplished by changing membrane composition, producing cryoprotective polypeptides such as COR15A (Artus et al., 1996; Steponkus et al., 1998) and accumulating compatible solutes with cryoprotective properties such as raffinose, sucrose, and proline (Nanjo et al., 1999; Gilmour et al., 2000; Taji et al., 2002). Main regulators of cold acclimation are CBF1/DREB1b, CBF2/DREB1c, and CBF3/DREB1a, three AP2/ERF transcription factors, for which allelic variation in CBF2 has been linked to natural variation in freezing tolerance (Oakley et al., 2014; Park et al., 2018; Alonso-Blanco et al., 2005). Nothing is known, however, about whether natural variation in freezing tolerance regulation also influences a trade-off with growth responses to cold temperatures.

Here, we investigated the role of growth in adaptation to cold temperatures by comparing vegetative growth of 249 accessions (Figure 1) grown in daily maximum temperatures of 16 and 6°C for a period of 3 weeks following seedling establishment (Figure 1—figure supplement 1). Rosette growth of each plant was measured twice a day during temperature treatments using automated phenotyping. The experiment generated rosette growth estimates at a high temporal resolution in two ecologically realistic temperature conditions in a wide set of accessions, allowing us to look for patterns of local adaptation.

Figure 1 with 1 supplement see all
Geographic origin of the 249 accessions.

Map color shows winter temperature (mean temperature of coldest quarter). Accessions are colored according to subpopulation (1001 Genomes Consortium, 2016). Accessions from the warmest and coldest regions are from Greece and the Himalayas, respectively.

Figure 1—source data 1

List of all 249 accessions with indication of the 8 accessions used for the transcriptome analysis.

https://cdn.elifesciences.org/articles/77913/elife-77913-fig1-data1-v1.csv

Results

Estimating plant growth

Our highly replicated experiment yielded dense (two measurements per day) time-series growth data for over 7000 individual plants (5 replicate plants × 249 accessions × 2 treatments × 3 replicate experiments). These data were used to model plant growth and estimate growth parameters for further analysis. Unlimited growth should be exponential, but plant growth is known to slow down with increasing size, and therefore a power-law function, dMdt=rMβ, with β<1 is typically a better fit than a pure exponential function (for which β=1 — in the equation, M is the size, r is the growth rate, and β is a scaling factor that allows rate of size increase to change with size). Growth according to a power-law function typically describes early stages of plant growth especially well (Paine et al., 2012), and our rosette size measurements were no exception (see ‘Materials and methods’). To calculate the rosette size from a power-law function at a given time point, only three parameters are required: the initial size (M0), growth rate (r), and β. Note that (M0) is the rosette size at the start of the temperature treatment 14 days after stratification (Figure 1—figure supplement 1) and is thus not affected by the temperature treatment. We used a nonlinear mixed model to obtain estimates for the initial size, growth rates, and β. Accession was added as fixed effect for initial size and growth rate, temperature and accession × temperature interactions were added as fixed effects for growth rate only. β was considered to be constant over accessions and temperatures. The ‘temperature response’ of the growth rate was calculated for each accession as the slope between the growth rate at 16 and 6°C. As expected, all accessions grew faster when it was warmer. The observed phenotypic variation (Figure 2—figure supplement 1) is to a large extent explained by genetic variation; broad-sense heritabilities are 0.41 for initial size, and 0.57 and 0.32 for growth rate at 16 and 6°C, respectively.

Growth parameters correlate with the environment of origin

If growth rates are locally adaptive, they may reflect the environment of origin of each accession. To investigate this, we correlated our estimated growth rates with climate data. The climate variables showing the strongest correlations with the different growth parameters were linked to winter temperatures (Figure 2—figure supplement 2), also when correlations were corrected for population structure (Figure 2—figure supplement 2). In particular, the mean temperature during the coldest quarter (henceforth referred to as ‘winter temperature’) was most strongly correlated with our parameter estimates, and we focus on it in what follows.

Initial size

Accessions from colder climates generally had higher initial rosette size (M0), 2 weeks after germination, than accessions from warmer climates (r=0.39), but then grew more slowly during the temperature experiment – regardless of temperature regime (Figure 2, Figure 2—figure supplement 3). Because during the first 2 weeks the plants were growing at 21°C, it is impossible to disentangle early growth from growth at a warmer temperature.

Figure 2 with 6 supplements see all
Correlations of growth parameters with winter temperature.

(A) Initial size. (B) Growth rate at 16°C. (C) Growth rate at 6°C. (D) Temperature response of growth rate. Colors indicate genetically defined subpopulations of the accessions (1001 Genomes Consortium, 2016).

One reason for this pattern is likely to be the differences in seed size between accessions. Using unpublished seed size measurements for a subset of 123 Swedish accessions from previous experiments, we found that seed size is positively correlated with initial size (r=0.28; Figure 2—figure supplement 5A), and also with winter temperature (r=0.75; Figure 2—figure supplement 5B), at least for the subset of 123 Swedish accessions. In a random-effect model, winter temperature explained 32.7% of the variation in initial size, whereas seed size explained 11.9%. Winter temperature is still significantly associated with initial size when seed size is taken into account (pvalue<1e04). The precise role of seed size in initial growth is surely a topic that would benefit from further studies, but for the purpose of this study, it is clear that seed size alone cannot explain the geographic pattern we observe for M0 and that there must be a role for variation in growth rate during the very initial phases of seedling growth.

Growth rates

While the initial sizes correlate negatively with winter temperature, we observed the opposite relation for the growth rates. Despite being larger initially, accessions from colder climates grew more slowly during both the 16°C (r=0.33) and 6°C (r=0.28) treatments (Figure 2). The higher growth rates of accessions from warmer climates prove that resources are not limiting, suggesting that the northern lines are actively inhibiting their growth, and that growing slower may be beneficial in colder climates, perhaps in preparation for winter. Accessions from colder climates were also less sensitive to the temperature experiment in the sense that the temperature response of the growth rate increased with winter temperature of origin (r=0.28; Figure 2D). Even though accessions from the Asian and north Swedish subpopulations were more variable in their growth rate temperature response (Figure 2—figure supplement 4A), the correlations still hold when removing either Asian or northern and southern Swedish subpopulations (Figure 2—figure supplement 6B, Figure 2—figure supplement 4D), and when looking specifically within the northern and southern Swedish subpopulations (Figure 2—figure supplement 6).

Cold acclimation response

Just like the observed geographic pattern of the growth rates, metabolite measurements taken at the final day of our experiment and presented in an earlier publication (Weiszmann et al., 2020) showed clear differences between accessions from cold and warm regions, and many of these differences involved metabolites with a known role in cold acclimation. Since the transcriptomic component of cold acclimation is well studied, we analyzed the expression profiles of 251 previously described cold-acclimation genes (Figure 3—source data 1) in eight accessions that were representative in terms of their growth and metabolome profiles (Figure 3—figure supplement 1). The selected genes are described in the literature as being differentially expressed upon exposure to cold, and their expression is under control of at least one of the known transcription factors regulating cold acclimation: CBF1, CBF2, CBF3, HSFC1 (Park et al., 2015), or ZAT12 (Vogel et al., 2005). In our experiment, expression of these genes is likewise more affected by temperature than expected by chance (Figure 3; χ2-test: pvalue<0.001) and separates the two accessions from the coldest region (northern Sweden) from the rest in the 16°C treatment, and the three accessions from the warmest regions (Spain and Azerbaijan) from the rest in the 6°C treatment. Expression of different subsets of the selected cold-acclimation genes shows clear correlations with winter temperature of origin (Figure 3—figure supplement 2, Figure 3—figure supplement 3). In particular, the genes that were previously found to be upregulated upon cold exposure showed higher expression in accessions from cold climates (Figure 3—figure supplement 4). Since the expression of these cold-acclimation genes has been linked to the strength of cold acclimation in previous experiments (Park et al., 2015; Vogel et al., 2005), these accessions likely differ in their ability to cope with freezing temperatures upon cold treatment.

Figure 3 with 5 supplements see all
Expression of 251 previously described cold-acclimation genes.

Expression is shown as the gene-wide z-scores of the normalized counts. The z-scores allow for grouping genes with a similar expression behavior over the different accessions in both temperatures. The top bar indicates winter temperature (°C) for each accession’s origin. Both dendrograms along y-axis and x-axis, respectively, show hierarchical clustering of genes, and of accessions in both temperatures.

Figure 3—source data 1

Cold-acclimation genes and their expression cluster membership as shown in Figure 3.

https://cdn.elifesciences.org/articles/77913/elife-77913-fig3-data1-v1.csv

Growth is polygenic and shows signs of local adaptation

We used genome-wide association to investigate the genetic architecture underlying variation for the different growth parameters. As expected, these traits appear to be highly polygenic, and there were no genome-wide significant associations (Figure 4—figure supplement 1). The strongest association was found for overall growth rate at 16°C (Figure 4A). Inflated significance levels after correcting for population structure are consistent with what we would expect from a polygenic trait (Figure 4B) and were also observed for the other traits, except for growth rate at 6°C (Figure 4—figure supplement 1). Plausible candidates within 10 kb of the most significant SNP (chr5: 23334281; log10(pvalue)=6.85) include CIPK21 and MYB36. CIPK21 encodes a CBL-interacting protein kinase that is upregulated in cold conditions and is involved in the salt and osmotic stress response (Pandey et al., 2015). MYB36 is a key regulator of root endodermal differentiation (Liberman et al., 2015). Slightly more distant, 22 kb away, is COL5, encoding a transcription factor that is part of the gene network that is regulated by AN3, a regulator of cell proliferation in leaf growth (Vercruyssen et al., 2014).

Figure 4 with 1 supplement see all
Genome-wide association study (GWAS) results for the growth rate at 16°C.

(A) Manhattan plot showing the significance of the association between the phenotype and each of the tested SNPs (MAF>10%). The Bonferroni-corrected threshold is shown with a dashed red line. (B) QQ-plot showing the relation between observed and expected log10(pvalue) distributions. Red line shows the observed relationship. The gray line and band show the expected relationship under the null hypothesis of no differentiation between both distributions.

To test for potential polygenic adaptation, we compared the phenotypic divergence to the expected neutral genome-wide genetic divergence. This can be done using a QST-FST test (Prout and Barker, 1993; Whitlock, 2008; Spitze, 1993); however, this test is not well suited for species with complex population structure, and so we used a variation that was designed to detect adaptive differentiation for traits measured in structured GWAS panels (Josephs et al., 2019). Instead of looking at divergences between predefined populations, this method uses principal components (PCs) of the genetic relatedness matrix as axes of potential adaptive differentiation. Adaptive differentiation is then detected as a correlation between the focal phenotype and any of these relatedness PCs that is significantly different than expected under neutrality.

Adaptive differentiation was detected for initial size and for growth rate at 16°C and its temperature response. These traits show adaptive differentiation along different genetic axes (Figure 5). Initial size shows significant adaptive differentiation along PC6 (pvalue<0.05), whereas growth rate at 16°C and its temperature response showed significant adaptive differentiation along PC5 (pvalues<0.05). Adaptive differentiation was not significant along the other axes of genetic differentiation (PC1–4, PC7–10). The adaptive differentiation for initial size along PC6 seems to stem from higher initial sizes in Swedish accessions compared to central European accessions. The adaptive differentiation along PC5 seems to be driven by the lower growth rate temperature responses in Asian and northern Swedish accessions in contrast to higher growth rates in a subset of southern Swedish accessions. The accessions in our set that come from northern Sweden and Asia hail from the coldest climates. Thus, these results suggest adaptive differentiation driven by adaptation to cold winters. Given the seemingly strong influence from the Asian accessions, we repeated the analysis without them. Also in this analysis we detected significant adaptive differentiation (pvalues<0.05) for initial size, growth rate at 16°C, and its temperature response (Figure 5—figure supplement 1).

Figure 5 with 1 supplement see all
Adaptive differentiation of initial size, growth rate at 16°C, and the temperature response of growth rate along different axes of genetic differentiation.

Plots represent the phenotypes and axes of genetic differentiation for which we detected significant adaptive differentiation; initial size and PC6 (A), growth rate in 16ºC and PC5 (B), and the growthrate's temperature response and PC5 (C). Accessions are colored according to their respective admixture groups, as specified in 1001 Genomes Consortium, 2016. The gray ribbon represents the expected correlation between phenotype and axis of genetic differentiation under neutrality with a 90% confidence interval. The neutral expectation is based on axes of genetic differentiation within populations (see ‘Materials and methods’ and Josephs et al., 2019 for further details). The blue line represents the observed correlation between phenotype and axis of genetic differentiation. Percentages refer to the genetic variation explained by the respective principal component.

Discussion

This study explores natural variation of rosette growth in nonfreezing temperatures. We detect genetic variation for the different growth parameters, and environmental correlations that suggest local adaptation. GWAS analyses reveal, not surprisingly, a polygenic trait architecture. We speculated that the slower growth measured in accessions from colder climates reflects relocation of resources from growth towards cold acclimation. Both metabolome and gene expression data are consistent with accessions from colder climates preparing for a harsh winter. In our temperature experiment, we see that the growth of northern lines is affected less than southern lines by switching from 16°C to 6°C.

Our conclusion that slower growth is likely adaptive in populations facing fiercer winters is in line with recent results of Wieters et al., 2021, who concluded that the reduced growth in northern lines was adaptive and not a consequence of an accumulation of deleterious mutations at the species border. If slower growth were indeed a consequence of accumulated deleterious mutations, we would expect to see slower growth also during the initial seedling establishment, which we measured here as the initial size. On the contrary, we saw a fast seedling establishment for accessions from colder regions. We speculate that the fast seedling establishment is a potential adaptation for short growth seasons, which often coincide with colder climates (high latitude or high altitude). This fast seedling establishment seems to be partly supported by larger seeds. These larger seeds may provide more nutrients to initiate faster seedling establishment, while this is of less importance in warmer climates with longer growth seasons. Further work is needed to disentangle initial growth from seed size effects and confirm that there is a causal relationship between seed size and fast seedling establishment, whether this is due to seed nutrient storage, and whether it is adaptive.

The adaptation of growth to local climates is likely to be influenced by a trade-off with cold acclimation. General growth-survival trade-offs have long been observed and are described in general ecological strategy schemes such as Grime’s C-S-R triangle (Grime, 1979) and the leaf–height–seed scheme (Westoby, 1998). Specific trade-offs between growth and cold/frost survival were observed for wheat (Hayes and Aamodt, 1927; Quisenberry, 1931), alfalfa (Castonguay et al., 2006), Dactylis glomerata (Bristiel et al., 2018), and multiple tree species (Koehler et al., 2012; Loehle, 1998; Molina-Montenegro et al., 2012; Savage and Cavender-Bares, 2013). Here, we observed higher expression of genes involved in cold acclimation in accessions from colder regions. This is clearest at 6°C, but is also happening at 16°C, suggesting that activation of cold acclimation is stronger in accessions from cold climates and may be triggered more easily. More generally, it establishes that geographic differentiation exists in the expression of known cold-response genes. Accessions from warm climates may instead activate cold acclimation as a stress response rather than as a preventive measure. Even though this is based on a limited set of 8 accessions, metabolome measurements in all 249 accessions lead to the same conclusion. Metabolites involved in cold acclimation such as raffinose, sucrose, and proline were found in higher concentrations in accessions from colder climates (Weiszmann et al., 2020). We believe that accessions from colder environments are relocating more energy and resources from growth towards preparations for upcoming freezing temperatures, which is a clear example of active plasticity (Forsman, 2015; van Kleunen and Fischer, 2005). Even though we have no direct survival measurements, we speculate that this results in stronger cold acclimation and consequently increased freezing tolerance in the accessions from colder regions. Indeed, accessions originating from colder environments show increased freezing tolerance upon cold acclimation (Zhen et al., 2011; Zuther et al., 2012; Hannah et al., 2006; Horton et al., 2016). This fits with observations of northern and colder regions favoring slower growing, more stress-tolerant plants (Vasseur et al., 2018; Estarague et al., 2022). Also, biogeographic studies in A. thaliana found that winter temperatures are a major determinant of suitable habitats for this species (Hoffmann, 2002; Yim et al., 2022), and reciprocal transplant experiments detected an important role for freezing tolerance in fitness variation in northern sites (Ågren and Schemske, 2012). The high variability we observed in our data does, however, show that there is more at play than selection for cold resistance alone. What these factors are we can only speculate about. Phenotypes are shaped by a mixture of neutral and adaptive processes, with a plethora of trade-offs between traits. Investigating phenotypes at different organismal scales in specific and realistic environments will further elucidate how phenotypes and, ultimately, life history strategies are shaped. We speculate that the reduced growth plasticity observed for accessions from colder climates is due to the stronger growth reduction at 16°C in these accessions compared to accessions from warmer climates. They may well be anticipating winter, whereas accessions from warmer climates do not, and hence show a stronger difference between 16 and 6°C. Although both metabolite and gene expression data suggest an involvement of cold acclimation, we note that it is impossible to rule out that accessions from colder climates grow slower due to reduced resource efficiency as they become larger or, for example, increase leaf thickness (Adams et al., 2016). Further work is needed to understand the mechanism underlying the growth response observed here.

There is strong evidence from QTL mapping that genetic variation in the CBF2 gene is one of the drivers for adaptation to freezing stress (Oakley et al., 2014; Gehan et al., 2015). Here, we looked at growth phenotypes and did not detect associations with the CBF loci. In the transcriptome analysis, we did pick up a role for CBF and other known cold-acclimation genes. The most significant locus detected in our GWAS analysis (for growth rate at 16°C) lies in the vicinity of COL5, a gene that is part of a leaf growth regulatory network (Vercruyssen et al., 2014) and whose expression is induced by both cold treatment and CBF1, CBF2, and CBF3 overexpression (Park et al., 2015). It is however unclear what its exact regulatory role in growth in cold conditions might be. In summary, we detected adaptive differentiation for growth between accessions from warm and cold climates. Our transcriptome data and previous metabolome data suggest that resources are relocated from growth to cold acclimation in accessions from colder regions. This allows these accessions to be fully prepared for the coming of winter.

Materials and methods

Plant growth and phenotyping

Request a detailed protocol

Seeds of 249 natural accessions (Figure 1—source data 1) of A. thaliana described in the 1001 genomes project (1001 Genomes Consortium, 2016) were sown on sieved (6 mm) substrate (Einheitserde ED63). Pots were filled with 71.5 g ± 1.5 g of soil to ensure homogenous packing. The prepared pots were all covered with blue mats (Junker et al., 2014) to enable a robust performance of the high-throughput image analysis algorithm. Seeds were stratified (4 days at 4°C in darkness) after which they germinated and left to grow for 2 weeks at 21°C (relative humidity: 55%; light intensity: 160μmolm2s1; 14 hr light). The temperature treatments were started by transferring the seedlings to either 6 or 16°C. To simulate natural conditions, temperatures fluctuated diurnally between 16–21°C, 0.5–6°C, and 8–16°C for the 21°C initial growth conditions and the 6 and 16°C treatments, respectively (Figure 1—figure supplement 1). Light intensity was kept constant at 160μmolm2s1 throughout the experiment. Relative humidity was set at 55% but in colder temperatures it rose uncontrollably to maximum 95%. Daylength was 9 hr during the 16 and 6°C treatments. Each temperature treatment was repeated in three independent experiments. Five replicate plants were grown for every genotype per experiment. Plants were randomly distributed across the growth chamber with an independent randomization pattern for each experiment. During the temperature treatments (14–35 days after stratification), plants were photographed twice a day (1 hr after/before lights switched on/off), using an RGB camera (IDS uEye UI-548xRE-C; 5MP) mounted to a robotic arm. Rosette areas were extracted from the plant images using Lemnatec OS (LemnaTec GmbH, Aachen, Germany) software. Plant growth profiles were visually inspected, and datapoints with smaller rosette areas than earlier time points (negative growth) were discarded from further analyses. At 35 days after stratification, whole rosettes were harvested, immediately frozen in liquid nitrogen, and stored at –80°C until further analysis.

Nonlinear modeling

Request a detailed protocol

Nonlinear modeling was used to describe plant growth in a minimum number of parameters. In a first step, we constructed a simple nonlinear model with plant size being explained by either the exponential (Equation 1a, Equation 1b) or the power-law function (Equation 2a, Equation 2b), with individual plant as a random effect for each of the model parameters; M0, r, and β. With β being only present in the power-law model. Models were constructed using the nlsList and nlme functions from the nlme package (3.1.152; Pinheiro et al., 2021) for R (4.0.3; R Development Core Team, 2017). Exponential and power-law SelfStart functions were used from Paine et al., 2012. Based on Akaike information criterion and likelihood ratio test generated by the anova function (Table 1), we decided to use the power-law model for further analyses.

(1a) dMdt=rM
(1b) Mt=M0ert
(2a) dMdt=rMβ
(2b) Mt=(M01-β+rt(1-β))1/(1-β)
Table 1
ANOVA table for the comparison between the exponential and power-law model with degrees of freedom (df), Akaike information criterion (AIC), Bayesian information criterion (BIC), and log-likelihood (logLik) for each model.

The likelihood ratio statistic (L.ratio) and p-value are given for the likelihood ratio test that was used to compare these models.

ModeldfAICBIClogLikTestL.ratiop-Value
Exponential16–668551.6–668488.1334281.8--
Power-law210–870605.0–870499.1435312.51 vs. 2202.061<0.001

In a second step, we constructed a model with fixed effects for the different power-law parameters. For initial size (M0), we added accession as fixed effect. Temperature treatment only started from the initial time point onwards, and thus could not have an effect on the initial plant size. The growth rate, on the other hand, should be affected by temperature; therefore, we included accession, temperature, and their interaction as fixed effects for growth rate (r). No fixed effects were added for β. The idea here is that it is an adjustment factor for decreasing growth rates (when β<1) with increasing plant sizes, which is general for plant growth, or at least for our data in this case. Individuals nested within experiment were added as random effects for each of the model parameters. The correlation structure intrinsic to measuring the same individuals over time was accounted for by adding the first-order continuous autoregressive correlation structure (corCAR1). The estimated fixed effects of this model were then used to obtain initial size estimates for each accession and growth rate estimates for each accession in both temperatures. These estimates were used for all further analyses apart from broad-sense heritability calculations (see below). For each accession, we calculated the growth rate response as the slope between the growth rate at 6°C and the growth rate at 16°C. The slope was obtained from linear regression with the lm function in R (4.0.3; R Development Core Team, 2017) using temperature as an ordered categorical variable (6°C < 16°C).

Climate correlations

Request a detailed protocol

The different phenotypes were correlated with each of the different (bio)climate variables downloaded from https://www.worldclim.org (Fick and Hijmans, 2017). Correlations were calculated as Pearson’s correlations using the cor function in R (4.0.3; R Development Core Team, 2017). Population structure may confound the correlation between phenotype and climate. Therefore, we included additional phenotype–climate correlations with correction for population structure (Figure 2—figure supplement 3). For the population–structure-corrected correlations, we used a mixed-effects model as implemented in the lmekin function from the coxme (2.2.16; Therneau, 2020) package with phenotype as dependent variable, climate variable as fixed effects, and the kinship matrix as random effect. The kinship matrix was based on the SNPs from the 1001 genomes consortium (1001 Genomes Consortium, 2016) and was calculated using ’mixmogam’ (https://github.com/bvilhjal/mixmogam; Vilhjalmsson, 2019) based on Kang et al., 2010. In this analysis, phenotype and climate variables were standardized, so that regression coefficients were comparable to correlation coefficients. Even though the strength and significance of the correlations weaken upon population structure correction, the growth parameters still demonstrate the same pattern, being most strongly correlated with winter temperatures.

Seed size correlations

Request a detailed protocol

We used the seeds produced by Kerdaffrec et al., 2016 and limited our measurements to the set of 123 Swedish accessions that overlapped with our growth dataset. After seed stratification for four days at 4°C in darkness, mother plants were grown for 8 weeks at 4°C under long-day conditions (16 hr light; 8 hr dark) to ensure proper vernalization. Temperature was raised to 21°C (light) and 16°C (dark) for flowering and seed ripening. Seeds were kept in darkness at 16°C and 30% relative humidity, from the harvest until seed size measurements. For each genotype, three replicates were pooled and about 200–300 seeds were sprinkled on 12 × 12 cm2, transparent Petri dishes. Image acquisition was performed as described in Exposito-Alonso et al., 2018 by scanning dishes on a cluster of eight Epson V600 scanners. The resulting 1200 dpi .tiff images were analyzed with the ImageJ software (2.1.0/1.53c). Images were converted to eight-bit binary images and thresholded with the setAutoThreshold("Defaultdark”) command, and seed area was measured in mm2 by running the Analyse Particles command (inclusion parameters: size=0.04-0.25circularity=0.70-1.00). All scripts used for image processing are available at https://github.com/vevel/seed_size; Kerdaffrec, 2022. The variance decomposition for initial size into variance explained by winter temperature and seed size was done with a random-effect model where initial size was explained by winter temperature and seed size, using the lmer and VarCorr functions from the lme4 package (1.1.27.1; Bates et al., 2015) in R (4.0.3; R Development Core Team, 2017). The seed size-corrected correlation between initial size and winter temperature was estimated with the lme function from the nlme package (3.1.152; Pinheiro et al., 2021) in R (4.0.3; R Development Core Team, 2017), and the correction for seed size was done by including seed size as a random effect.

Transcriptome profiling

Request a detailed protocol

35 days after stratification, rosette tissue of all plants were harvested and flash frozen in liquid nitrogen. Random samples from each replicate experiment for both temperatures were taken for eight accessions to profile the transcriptome with RNA-sequencing. The eight accessions were selected to represent the climatic variation in the full panel (Figure 3—figure supplement 1, Figure 1—source data 1). Total RNA was extracted using the KingFisher Duo Prime System (Thermo Fisher Scientific) together with a high-performance RNA bead isolation kit (Molecular Biology Service, VBC Core Facilities, Vienna). To determine the quantity of RNA, we used Fluorometer Qubit 4 (Invitrogen) and Qubit RNA BR Kit (Invitrogen). For each sample, 1 µg of total RNA was treated with the poly(A) RNA Selection Kit (Lexogen) and eluted in 12 µl of Nuclease-Free Water. Libraries were prepared according to the manufacturer’s protocol in NEBNext Ultra II Directional RNA Library Prep Kit (New England Biolabs) and individually indexed with NEBNext Multiplex Oligos for Illumina (New England Biolabs). The quantity and quality of each amplified library were analyzed by using Fragment Analyzer (Agilent) and HS NGS Fragment Kit (Agilent). Libraries were sequenced with an Illumina HiSeq2500 in paired-end mode with read length of 125 bp. Sequencing was performed by the Next Generation Sequencing Facility at Vienna BioCenter Core Facilities (VBCF), member of the Vienna BioCenter (VBC), Austria. Samples were distributed over four independent libraries. This was due to failed samples that needed replacement. Detailed info on which samples belong to which library is listed in SRA (https://www.ncbi.nlm.nih.gov/sra/PRJNA807069). Gene expression was quantified by using quasi-mapping in salmon (1.2.1; Patro et al., 2017). The salmon indices were built separately for each accession as we incorporated the SNP variation from the 1001 Genomes Consortium, 2016 into the reference transcriptome. The heatmap was built using pheatmap (1.0.12; Kolde, 2019) in R (4.0.3; R Development Core Team, 2017). The clustering was done by complete clustering on Euclidean distances for both rows and columns. The gene clusters were defined by cutting the dendrogram in seven groups. With a -test, we tested for overrepresentation of a temperature effect on the expression of the 251 selected cold-acclimation genes (d f=1) compared to the remaining 18,784 background genes. We used the chisq.test function in R (4.0.3; R Development Core Team, 2017). Differential expression analysis was conducted with the DESeq2 package (1.30.0; Love et al., 2014) in R (4.0.3; R Development Core Team, 2017). A full model was used, with expression depending on replicate+accession+temperature+replicate:temperature+accession:temperature , after which significance of each model coefficient was defined with a negative binomial Wald test. Differential expression for each accession was then extracted by specifying the respective contrasts using the lfcShrink function in DESeq2 with the adaptive shrinkage estimator (Stephens, 2017). Genes were considered differentially expressed when the adjusted p-value was <0.05.

Metabolome profiling

Request a detailed protocol

Besides transcriptome profiling of eight accessions, we also conducted metabolome profiling on all 249 accessions. Samples for metabolome measurements were taken from the same experiments described in this study and, just like the transcriptome samples, were taken 35 days after stratification. Results and detailed methodology are described in Weiszmann et al., 2020.

Broad-sense heritabilities

Request a detailed protocol

Broad-sense heritabilities (H2) were calculated as the ratio between phenotypic variation explained by genotype (Vg) and the total phenotypic variation (Vp), which is the sum of Vg and phenotypic variation explained by environment (Ve). These variances were obtained from a mixed model by including accession as a random effect (estimate for Vg). Because our accession estimates were corrected for experiment effects, we removed the variance explained by experiment by including experiment as a fixed effect. In order to estimate the variance within each accession (Ve), the dependent variables in this model were the growth parameter estimates for each individual plant, in contrast to the estimates for each accession that were used in all other analyses. These individual plant estimates were obtained from the same model, but took the random effects into account. The variance explained by accession was then taken as an estimate for Vg, the residual variance was taken as Ve. For initial size, we calculated heritability over all experiments, and growth rate heritabilities were calculated for each temperature independently. The mixed model was constructed with the lmer function in the lme4 package (1.1.27.1; Bates et al., 2015) in R (4.0.3; R Development Core Team, 2017).

Genome-wide association mapping

Request a detailed protocol

Genome-wide association mapping was done for each of the growth parameters in both temperatures and also the temperature response for the growth rate. We used a mixed model with phenotype as dependent variable, genotype as fixed effect, and genetic relatedness as random factor. Nonimputed SNPs obtained from the 1001 Genomes Consortium, 2016 were used as genotypes. This model was run in GEMMA (0.98.3; Zhou and Stephens, 2012), with kinship matrix calculated as the centered relatedness matrix, as implemented in GEMMA.

Testing for adaptive differentiation

Request a detailed protocol

Adaptive differentiation was tested with the method described by Josephs et al., 2019 and the accompanying quaint package (0.0.0.9; https://github.com/emjosephs/quaint; Josephs, 2020) in R (4.0.3; R Development Core Team, 2017). The kinship matrix was calculated using the make_k function in the quaint package. Genetic PCs were then calculated from the eigen decomposition of the kinship matrix. Adaptive differentiation of each phenotype along the first 10 PCs was tested with the calcQpc function in the quaint package. PCs 11–248 were used to build the expected phenotypic differentiation under neutrality.

Supplementary information

Request a detailed protocol

Scripts can be found at https://github.com/picla/growth_16C_6C/; Clauw, 2022. Scripts for seedsize analysis can be found at https://github.com/vevel/seed_size; Kerdaffrec, 2022.

All RNA-sequencing were uploaded to SRA under http://www.ncbi.nlm.nih.gov/bioproject/807069. All generated phenotyping data are filed under 10.5281/zenodo.6076948.

Data availability

Scripts can be found at https://github.com/picla/growth_16C_6C/, copy archived at swh:1:rev:f32793d07a068a6c49c876536f684ffebdcd9e6b. Scripts for seedsize analysis can be found in https://github.com/vevel/seed_size, copy archived at swh:1:rev:748523031aa64601720328ab49e02009a6fb70da. All RNA-sequencing were uploaded to SRA under http://www.ncbi.nlm.nih.gov/bioproject/807069. All generated phenotyping data are filed under https://doi.org/10.5281/zenodo.6076948.

The following data sets were generated
    1. Clauw P
    (2022) NCBI BioProject
    ID PRJNA807069. Arabidopsis transcriptome in 16C and 6C.

References

    1. Castonguay Y
    2. Laberge S
    3. Brummer EC
    4. Volenec JJ
    (2006)
    In Advances in Agronomy
    203–265, Alfalfa winter hardiness: A research retrospective and integrated perspective, In Advances in Agronomy, Academic Press.
    1. Grime JP
    (1979) Primary strategies in plants
    Transactions of the Botanical Society of Edinburgh 43:151–160.
    https://doi.org/10.1080/03746607908685348
    1. Hayes HK
    2. Aamodt OS
    (1927)
    Inheritance of winter hardiness and growth habit in crosses of marquis with minhardi and minturki wheats
    Journal of Agricultural Research 35:223–236.
  1. Book
    1. Quisenberry KS
    (1931)
    Inheritance of Winter Hardiness, Growth Habit, and Stem-Rust Reaction in Crosses Between Minhardi Winter and H-44 Spring Wheats
    U.S. Department of Agriculture.
  2. Software
    1. R Development Core Team
    (2017) R: A Language and Environment for Statistical Computing
    R Foundation for Statistical Computing, Vienna, Austria.
  3. Book
    1. Stearns SC
    (1992)
    The Evolution of Life Histories
    Oxford University Press.

Decision letter

  1. Regina S Baucom
    Reviewing Editor; University of Michigan, United States
  2. Christian R Landry
    Senior Editor; Université Laval, Canada
  3. Regina S Baucom
    Reviewer; University of Michigan, United States

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Locally adaptive temperature response of vegetative growth in Arabidopsis thaliana" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Regina S Baucom as Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Christian Landry as the Senior Editor.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

(1) The work needs significant re-framing for broad relevance in the introduction. As is currently written, the manuscript is perhaps of more interest to a plant biology journal. Re-framing will be essential for eLife's broad audience. Each reviewer provides suggestions to this end.

(2) Each major conclusion needs better justification (ie need to see the results of analysis, F-values, P-values). Many of the conclusions were presented without such results; a supplemental document is likely needed in this regard. Reviewers felt results from the phenotype data, metabolome data, transcriptome data etc all were a bit weak in this regard.

(3) A number of conclusions drawn were questioned by reviewers, with suggestions for re-analysis and clarity surrounding analyses. Each of these in turn should be considered.

Reviewer #1 (Recommendations for the authors):

I enjoyed reading this work; it is a very extensive set of data nicely pulled together into a story and it will add a useful examination of plant growth adaptation to the literature. I list areas of improvement below according to line number.

Introduction

1. Lines 29-46 lack a clear and novel hypothesis that the authors are addressing; I suggest reframing here in terms of life-history adaptation and trade-offs between growth for reproduction and survival.

2. Line 48 for clarity, remove ‘upcoming’ and ‘consequently’.

Results

3. Presentation of sample sizes is sometimes unclear. Line 71 rephrase to 'Our replicated experiment yielded dense…' and '(5 replicates X 249 accessions X 2 treatments X 3 experiments)'.

4. In Line 74-80, unclear why authors chose power law here – is this based on their own data or previous analyses? Methods make it apparent that authors have modeled this themselves, yet there is no presentation of model fit analysis in the results. This section really needs data in support of conclusions – report AIC, etc, explaining why power law is appropriate. Or, develop this in supplemental text.

5. Lines 91-94. The authors jump back and forth between using accession as a fixed effect in some analyses, and then random effect in others. Explanation of these choices would improve the manuscript.

6. Lines 105-115. This is a pretty significant section yet the data are not shown – estimates of the strength of the correlation between growth parameters and winter temperature if authors assert 'strongly correlated' should be shown in text as a range. Figures S1 and S2 are very nicely displayed but the assertion of a strong correlation should have a number associated with it so that readers can form their own conclusions. Typically 'strong' correlations are 0.5+ but often 0.7+. Lower than 0.4 are most often considered moderate.

7. Lines 117-145. It appears from the regressions that Asian populations are doing a lot of the work. This is certainly biologically relevant (although not discussed enough in the discussion!) so I am not suggesting the authors should remove these populations but they should report how much of their overall conclusions are driven by the populations experiencing the coldest temperatures. How much are these populations driving overall trends?

Additionally, the data in this section are discussed predominantly as the outcome of regressions, with R2 values reported on the figure. In the methods, it is clear that authors modeled effects using fixed and mixed analyses of variance, yet none of these results are apparent in text or in supplemental. Readers, myself included, want to see effect sizes, F- and p-values associated with the claims made; without them, the work appears analyses-lite.

How much of the variation in size was due to seed size? Numbers are necessary to back up claims such as 'seed size alone' etc. What proportion of the variance is leftover once seed size is corrected? I would suggest modeling seed size as a fixed covariate, unclear why the choice to model it as a random effect was made.

Line 138. Why aren't resources limiting? Do the authors mean resources in this particular growth chamber study aren't limiting?

Lines 141-145. The conclusion here is hard to accept given that the Swedish populations show a huge range in response to temperature changes.

8. Lines 147+. A metabolome experiment is mentioned yet there was not a metabolomics section in the methods, nor was there clear data analysis of a metabolome experiment. I assume authors are referencing a previously published dataset but details should be present in this work as well.

9. Line 196. Remove 'so-called'.

10. Lines 208+. The influence of the Asian populations should be assessed in the test of adaptive differentiation.

11. Figure 2. This figure is difficult to interpret. I like the idea of having a conceptual map of the experimental design, but the 'seedling establishment' banner is odd (perhaps put this text on the x-axis during seedling establishment phase and not within the temps on the y axis?). The inlay showing the light/temp acclimation period is hard to interpret, why are the y-axes off-kilter from the overall figure? Y-axis should be labeled 'Temperature'.

Discussion

12. Line 227 'Despite high plasticity' <- unclear where this conclusion comes from – current data of temperature response? If so, then plasticity needs to be discussed at more length in the results. Or, if authors are concluding this from previously published work, a citation is needed.

13. Lines 237-239. This statement is difficult to accept given the wide variation among Swedish populations.

14. Lines 245-255. The discussion of seedling establishment is a bit long especially since I didn't see seedling establishment data presented per se. How was fast seedling establishment measured and was it compared between northern and southern latitudes?

Methods.

15. Most comments above on data presentation can be applied to the methods.

Line 419 'each replicate' is unclear – I believe the authors meant replicate experiment here not replicate plants. The overall transcriptome sampling is unclear – how many total libraries were developed?

Was there a metabolome experiment? If so it should be explained briefly in the methods.

Reviewer #2 (Recommendations for the authors):

1. The metabolome analysis is mentioned in the discussion (lines 270-272) but not reported in the results.

2. There are a number of crucial pieces of information missing from the presentation of the gene expression results. How were the dendrograms in figure 5 made? How were these clusters generated? Is this figure only for the 251 genes or for a larger set of genes? What was the background set of genes for the Chi-square reported in line 161? Please provide additional details to answer these questions.

3. The information presented in Figure S5 seems very relevant and interesting to the main results, but this figure could be improved. It is very hard right now to read the trend line between the temperature of origin and expression for each cluster/treatment, and this would be clearer if the lines were a contrasting color.

4. We were confused about the statistical tests reported in Figure S5, since they do not appear to be described in the methods. It seems like this figure reports the correlation between the temperature of origin and expression for every gene in the cluster. This analysis would be pseudoreplication since the expression of different genes from the same genotype is not independent. A more appropriate analysis would be to estimate the correlation of temperature of origin with the mean expression or eigengene for all genes in the cluster. Alternatively, the correlation between the temperature of origin and expression could be independently calculated for each gene in a cluster, and then the number of significant correlations and/or mean slope of these correlations within each cluster could be reported.

5. In the paragraph beginning in line 261, it is clear how a slower growth rate in accessions from cold regions at 6C is consistent with cold acclimation literature but not how the reduced growth rate plasticity in these accessions is connected to the literature. The discussion could be added to explain why the accessions from cold regions also grow more slowly when never exposed to cold (expand on line 138) and why they might have less plasticity in the growth rate.

Additional line comments:

6. The paragraph starting on line 47 could use more context for how the information relates to the broader argument.

7. Line 123 should cite where the seed size information is from.

8. It would be useful to report the p-value and effect size of the "most significant" association (line 179).

9. What are the "inflated significance levels" referred to on lines 180-181?

10. Should include statistics to justify the model choice in lines 348-351.

11. Lines with potential typos: 344, 360, 462.

12. Figures should have larger axes and labels, and all panels should be labeled within a figure. For example, label the panels in Figures3 and 4.

13. In figure 4, what impact does the x-axis outlier (coldest temp) have on the patterns we see? It would be nice to know that the outlier is not driving the trends and to have an explanation of why that accession is much colder than the rest (for example, where is this accession from?).

14. The Qst vs Fst analysis was used appropriately but there could be an additional explanation for how to interpret figure 7 in regards to how the gray fields are calculated

15. Unclear what "high plasticity" means in line 227.

Reviewer #3 (Recommendations for the authors):

Line 35: remove so-*

Line 47-58: the emphasis on CBFs is maybe excessive here as you don't provide compelling evidence for their involvement. Maybe swap for an introduction of the main ecological strategy of Nordic vs Southern adapted accessions?

Line 74-82: Provide the AIC/BIC/deviance or RMSE to justify the use of power vs exponential growth model in materials, but keep it simple here so that the interpretation of the parameter is clearer.

Figure 2: Upon vernalisattion? I believe you mean cold stratification?

Line 94: This is peripheral but I wonder if there is no biological insight to be gained from fitting a β for each accession. Wouldn't this be informative about the rate of resource accumulation?

Line 103-128: I understand that some metrics are provided in the figures associated with these sections but it would be clearer to provide some quantitative support (and tests) in the main text.

Line 147: how does the metabolite measurement fit?

Line 154: which accessions? Give the rationale for selection based on Figure S4.

Line 162: the null hypothesis of the test is not known.

Line 171-174: I find this argument circular: these genes were identified as differentially expressed in the very same accession I presume (although this is not explicitly stated in the paper).

Line 134: relocation of resources towards… I think there could be some information in analysing β's inferred for each accession here.

Line 360: typo

Line 372: Put the rest of the section in the heritability section.

Line 392: grammar issue.

Line 462: grammar again.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Locally adaptive temperature response of vegetative growth in Arabidopsis thaliana" for further consideration by eLife. Your revised article has been evaluated by Christian Landry (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

Reviewer #2 (Recommendations for the authors):

The clarity of the manuscript has been improved particularly in the descriptions of the methods and figure design. A few additional analyses to explore subsets of the data that drive trends in growth rate are included which strengthens the overall conclusion that temperature response of vegetative growth is locally adaptive. However, the manuscript has not addressed two of our initial concerns about the broad framing of the manuscript and the gene expression analyses.

1. The manuscript has been reframed in the context of ecological and life history strategies with a focus on the growth and survival tradeoff, based on feedback from us and other reviewers. While the growth rate results are potentially consistent with what is expected for a growth/survival tradeoff, there is no evidence for the other side of the tradeoff, survival. The manuscript should make this clear. In addition, the current framing includes references to active and passive plasticity, but we are not convinced that faster growth of accessions from warm temperatures "proves" that slow growth of accessions from cold temperatures is active plasticity (lines 125-127) as opposed to another explanation such as less efficient use of resources in accessions from cold temperatures at a later life stage. Additionally, the manuscript notes that "plant growth is known to slow down with increasing size" when justifying its model choice for estimating growth rates. With this in mind, is it a surprise that the populations with initial larger rosettes now show slower growth?

2. The gene expression results should be better integrated to support the conclusion that growth rate response is locally adapted. The main results described in the text are (1) the cold acclimation genes are more likely to be differentially expressed between temperatures compared to background genes. This is unsurprising since these genes were initially identified as being differentially expressed in the cold. (2) Gene expression clustering separates accessions from warmer and colder climates and the expression of some clusters is correlated with the temperature of origin. It is unclear if this pattern differs from background genes or the trait divergence expected under neutrality. (3) Genes that were up-regulated in cold (from published experiments) are also generally upregulated in accessions from cold climates. This last result is potentially the clearest link to the adaptive story, but it is not emphasized here and it's also unclear to us if upregulating these genes is a sign of adaptive or maladaptive plasticity. Overall, there is not a clear link between these results and the conclusion about adaptation to cold temperatures.

In addition, while Figure 3 —figure supplement 3 does address our concern about pseudoreplication, Figure 3 —figure supplement 2 still reports misleading p values that should not be included due to pseudoreplication.

Line Comments:

The statements made in lines 41-43 should have citations.

Lines 154-156 seem circular.

Line 181 change "both" to "all".

Line 198-199. The phrase "across the genome we detect adaptive differentiation for certain growth parameters." seems unsupported by the data. While the Qst-Fst results show evidence that phenotypic variation is adaptive, there is still no evidence that this trait is shaped by loci 'across the genome'.

Line 273 should read 160 µmolm−2s-1?

It is not clear how the growth rate temperature response was calculated. Line 91-92 suggests that it is "the slope between the growth rate at 16ºC and 6ºC" but examining specific points in figure 2B, C, and D makes it look like something else is going on. For example, the far left point has a growth rate of ~0.16 in 16C and ~0.052 in 6C, but the response is ~0.072 when we would expect something closer to 0.1 based on the text.

Please include percent variance explained on figures with PCs as axes (Figure 5 and Figure 5 supplement).

Figure 1—figure supplement 1 is somewhat confusing with the y axis referring to the experiment timeline on the x-axis and to the inset graph of a single day.

The inclusion of the previously published metabolite data is still somewhat confusing. The results report that metabolite measurements were different between accessions from different climates (line 137) but these results are not presented in a figure anywhere. If the authors want to present previously published data that has been reanalyzed to this paper in a figure, that would be fine. However, without that, it seems like this statement should be moved to the discussion.

https://doi.org/10.7554/eLife.77913.sa1

Author response

Essential revisions:

1) The work needs significant re-framing for broad relevance in the introduction. As is currently written, the manuscript is perhaps of more interest to a plant biology journal. Re-framing will be essential for eLife's broad audience. Each reviewer provides suggestions to this end.

We agree with this point, and have revised both Introduction and Discussion accordingly. See below for details.

2) Each major conclusion needs better justification (ie need to see the results of analysis, F-values, P-values). Many of the conclusions were presented without such results; a supplemental document is likely needed in this regard. Reviewers felt results from the phenotype data, metabolome data, transcriptome data etc all were a bit weak in this regard.

This was mostly due to poor presentation, and we have revised accordingly throughout. In particular, we have expanded Materials and methods.

3) A number of conclusions drawn were questioned by reviewers, with suggestions for re-analysis and clarity surrounding analyses. Each of these in turn should be considered.

Again we agree, and we have provided additional analyses to address these concerns. Most importantly, we reanalyzed the data with outlier groups removed to confirm that the conclusions still hold. Please see below for details.

Reviewer #1 (Recommendations for the authors):

I enjoyed reading this work; it is a very extensive set of data nicely pulled together into a story and it will add a useful examination of plant growth adaptation to the literature. I list areas of improvement below according to line number.

Introduction

1. Lines 29-46 lack a clear and novel hypothesis that the authors are addressing; I suggest reframing here in terms of life-history adaptation and trade-offs between growth for reproduction and survival.

Thank you for this suggestion. We have incorporated stronger links to the life-history/ecological strategy literature, in both the Introduction and the Discussion.

2. Line 48 for clarity, remove ‘upcoming’ and ‘consequently’.

The sentence has been rewritten.

Results

3. Presentation of sample sizes is sometimes unclear. Line 71 rephrase to 'Our replicated experiment yielded dense…' and '(5 replicates X 249 accessions X 2 treatments X 3 experiments)'.

Done

4. In Line 74-80, unclear why authors chose power law here – is this based on their own data or previous analyses? Methods make it apparent that authors have modeled this themselves, yet there is no presentation of model fit analysis in the results. This section really needs data in support of conclusions – report AIC, etc, explaining why power law is appropriate. Or, develop this in supplemental text.

This section has been expanded, and the revised manuscript contains the Anova table (Table S1) with the results of the comparison between an exponential and a power-law model. This table gives degrees of freedom, AIC, BIC and the log-likelihood for each model, and the likelihood ratio and p-value for the comparison of both models. We mention in the text that the power-law model is a standard model for these kinds of data.

5. Lines 91-94. The authors jump back and forth between using accession as a fixed effect in some analyses, and then random effect in others. Explanation of these choices would improve the manuscript.

This is a misunderstanding, again presumably due to unclear writing. Accession was consistently used as a fixed effect in the main power-law model. This model produced estimates for the different growth parameters for each accession.

We suspect the confusion comes from the description of the broad sense-heritabilities. To calculate these, we used growth parameter estimates for each individual plant. These estimates were obtained from the same power-law model, but estimates now take the random effect for individual plants into account, so that we could partition the variance into genetic and “environmental” variance and calculate heritability.

We have revised the two relevant Materials and methods sections (“Non-linear modeling” and “Broad-sense heritabilities”) to make this clearer.

6. Lines 105-115. This is a pretty significant section yet the data are not shown – estimates of the strength of the correlation between growth parameters and winter temperature if authors assert 'strongly correlated' should be shown in text as a range. Figures S1 and S2 are very nicely displayed but the assertion of a strong correlation should have a number associated with it so that readers can form their own conclusions. Typically 'strong' correlations are 0.5+ but often 0.7+. Lower than 0.4 are most often considered moderate.

The correlation coefficients for growth parameters and winter temperature are actually given in Figure 4. We have added a figure reference to the text, and also toned down the language. Our main point is that growth parameters are most strongly correlated with winter temperature.

7. Lines 117-145. It appears from the regressions that Asian populations are doing a lot of the work. This is certainly biologically relevant (although not discussed enough in the discussion!) so I am not suggesting the authors should remove these populations but they should report how much of their overall conclusions are driven by the populations experiencing the coldest temperatures. How much are these populations driving overall trends?

Excellent question. As noted above, we tried this, and it turns out not to matter. We did a similar analysis for the Swedish accessions with similar results – see comments below. We note this in the text, and provide supplemental figures (Figure 2—figure supplement 4,6).

Additionally, the data in this section are discussed predominantly as the outcome of regressions, with R2 values reported on the figure. In the methods, it is clear that authors modeled effects using fixed and mixed analyses of variance, yet none of these results are apparent in text or in supplemental. Readers, myself included, want to see effect sizes, F- and p-values associated with the claims made; without them, the work appears analyses-lite.

The mixed-model analysis was only used for the population structure-corrected climate correlations in Figure 2—figure supplement 3. The other analyses are standard regressions. We adjusted the Materials and methods section on climate correlations to clarify this.

How much of the variation in size was due to seed size? Numbers are necessary to back up claims such as 'seed size alone' etc. What proportion of the variance is leftover once seed size is corrected? I would suggest modeling seed size as a fixed covariate, unclear why the choice to model it as a random effect was made.

We added a variance decomposition to see how much of the variation in initial size is explained by winter temperature and by seed size (32.7% and 11.9% of the variation, respectively). Together with the significant correlation when correcting for seed size, this shows that winter temperature is still strongly correlated with initial size when seed size is taken into account.

We chose to add seed size as a random effect since we only wanted an estimate for the correlation between initial size and winter temperature.

Line 138. Why aren't resources limiting? Do the authors mean resources in this particular growth chamber study aren't limiting?

Yes, this is a reference to the higher growth rates of the accessions from warmer climates. Those higher growth rates prove that resources are not limited and thus accessions with slower growth could in principle also exhibit similarly high growth. We clarified this in the text.

Lines 141-145. The conclusion here is hard to accept given that the Swedish populations show a huge range in response to temperature changes.

The Swedish accessions are indeed showing a very large spread for the temperature response of the growth rate. We added a supplementary figure (Figure 2—figure supplement 6) showing the coefficient of variance for each of the subpopulations and its correlation with winter temperature, plots of correlation between the growth rate's temperature response and winter temperature for all except the Swedish subpopulations, and the same correlation for only the Swedish subpopulations. These plots show that accessions from colder temperatures still have lower plasticity for this trait, also within the Swedish subpopulations the conclusion still holds.

8. Lines 147+. A metabolome experiment is mentioned yet there was not a metabolomics section in the methods, nor was there clear data analysis of a metabolome experiment. I assume authors are referencing a previously published dataset but details should be present in this work as well.

The metabolome experiment that we mention was indeed described in a previous study. We added a small selection describing this in Materials and methods, with a clear reference to the respective paper.

9. Line 196. Remove 'so-called'.

Done

10. Lines 208+. The influence of the Asian populations should be assessed in the test of adaptive differentiation.

Yes, thanks for this suggestion. We assessed the influence of the Asian accessions on the detected adaptive differentiation by excluding them. It is important to note here that the principal components needed to be recalculated after doing this, and are thus not the same as the principal components in the original analysis.

In summary, we still detect adaptive differentiation (Figure 5—figure supplement 1). For initial size there is significant adaptive differentiation along 3 different axes of genetic differentiation (PC1, PC5 and PC9). PC1 is mainly splitting northern Swedish and to a lesser extent also southern Swedish accessions from the rest, with both showing higher initial sizes compared to the other accessions. PC5 is differentiating southern Swedish and to some extent northern Swedish from central European accessions. PC9 is more enigmatic, splitting up two subgroups of the western European accessions. For both growth rate at 16ºC and its temperature response we detect adaptive differentiation along PC5, splitting up southern Swedish, northern Swedish and central European accessions. Linking back to the initial analysis, we can say that there is adaptive differentiation for initial size, growth rate at 16ºC and its temperature response. Asian accessions are driving part of this adaptive differentiation, however, also for the Swedish accessions we detect significant signals of being differently adapted in terms of these growth parameters.

This analysis is added to the manuscript (Figure 5—figure supplement 1, line 267-271).

11. Figure 2. This figure is difficult to interpret. I like the idea of having a conceptual map of the experimental design, but the 'seedling establishment' banner is odd (perhaps put this text on the x-axis during seedling establishment phase and not within the temps on the y axis?). The inlay showing the light/temp acclimation period is hard to interpret, why are the y-axes off-kilter from the overall figure? Y-axis should be labeled 'Temperature'.

We have implemented your suggestions for improving the figure, however, since reviewer #3 did not find this figure very informative, we have moved it to supplement.

Discussion

12. Line 227 'Despite high plasticity' <- unclear where this conclusion comes from – current data of temperature response? If so, then plasticity needs to be discussed at more length in the results. Or, if authors are concluding this from previously published work, a citation is needed.

Correct, we don't draw conclusions about plasticity. We removed this statement since it has no real use and will cause confusion.

13. Lines 237-239. This statement is difficult to accept given the wide variation among Swedish populations.

Since the correlations without the Swedish accessions are still holding (see earlier comment), we think that this claim can still be made.

14. Lines 245-255. The discussion of seedling establishment is a bit long especially since I didn't see seedling establishment data presented per se. How was fast seedling establishment measured and was it compared between northern and southern latitudes?

We clarified that we measured seedling establishment based on the initial size, one of the growth parameters from the power-law model.

Methods.

15. Most comments above on data presentation can be applied to the methods.

Line 419 'each replicate' is unclear – I believe the authors meant replicate experiment here not replicate plants. The overall transcriptome sampling is unclear – how many total libraries were developed?

Adjusted 'replicate' to 'replicate experiment'. Information about libraries has been added.

Was there a metabolome experiment? If so it should be explained briefly in the methods.

Materials and methods now contains a section on the metabolome measurements.

Reviewer #2 (Recommendations for the authors):

1. The metabolome analysis is mentioned in the discussion (lines 270-272) but not reported in the results.

The metabolome analysis had been published. For completeness, we added a specific section in Materials and methods. See also earlier comments by Reviewer #1.

2. There are a number of crucial pieces of information missing from the presentation of the gene expression results. How were the dendrograms in figure 5 made? How were these clusters generated? Is this figure only for the 251 genes or for a larger set of genes? What was the background set of genes for the Chi-square reported in line 161? Please provide additional details to answer these questions.

Yes: we have clarified the gene expression results. The caption of Figure 5 (now Figure 3) now explicitly states that it contains the 251 cold acclimation genes. Further details on the clustering method, chi-square test and differential expression analysis are added to Materials and methods.

3. The information presented in Figure S5 seems very relevant and interesting to the main results, but this figure could be improved. It is very hard right now to read the trend line between the temperature of origin and expression for each cluster/treatment, and this would be clearer if the lines were a contrasting color.

Due to image conversion, the contrast in the figure (now Figure 3—figure supplement 2) in the submitted pdf is pretty bad indeed, and we apologize for not noticing this. The original figure does provide a good contrast between the individual gene lines and the trend line. We further improved the contrast by adding a black shadow around the trend lines. This has been fixed.

4. We were confused about the statistical tests reported in Figure S5, since they do not appear to be described in the methods. It seems like this figure reports the correlation between the temperature of origin and expression for every gene in the cluster. This analysis would be pseudoreplication since the expression of different genes from the same genotype is not independent. A more appropriate analysis would be to estimate the correlation of temperature of origin with the mean expression or eigengene for all genes in the cluster. Alternatively, the correlation between the temperature of origin and expression could be independently calculated for each gene in a cluster, and then the number of significant correlations and/or mean slope of these correlations within each cluster could be reported.

Agreed. We have added a supplemental figure (Figure 3—figure supplement 3) that shows the proportion of genes in each cluster that significantly correlates with winter temperature.

5. In the paragraph beginning in line 261, it is clear how a slower growth rate in accessions from cold regions at 6C is consistent with cold acclimation literature but not how the reduced growth rate plasticity in these accessions is connected to the literature. The discussion could be added to explain why the accessions from cold regions also grow more slowly when never exposed to cold (expand on line 138) and why they might have less plasticity in the growth rate.

We have added a small paragraph to the Discussion where we speculate about the meaning of the reduced growth plasticity in accessions from cold regions.

Additional line comments:

6. The paragraph starting on line 47 could use more context for how the information relates to the broader argument.

We added an explanation that it is not clear how natural variation in cold acclimation is in a trade-off with growth responses in cold temperatures. We do think it is important to have a brief introduction to cold acclimation and its regulators, since that information was used to select the candidate genes involved in cold acclimation.

7. Line 123 should cite where the seed size information is from.

Seed size measurements came from other experiments done in the group and were never published. Methodology is described in Materials and methods and we made clear in the text that the data comes from unpublished experiments.

8. It would be useful to report the p-value and effect size of the "most significant" association (line 179).

P-value and effect size are added to the results.

9. What are the "inflated significance levels" referred to on lines 180-181?

The text now refers explicitly to the QQ-plot in Figure 4B.

10. Should include statistics to justify the model choice in lines 348-351.

We added the ANOVA table on which we based our model choice. See also comment by reviewer #1.

11. Lines with potential typos: 344, 360, 462.

Fixed: thanks for pointing these out.

12. Figures should have larger axes and labels, and all panels should be labeled within a figure. For example, label the panels in Figures3 and 4.

All panels are now labeled and font sizes have been increased.

13. In figure 4, what impact does the x-axis outlier (coldest temp) have on the patterns we see? It would be nice to know that the outlier is not driving the trends and to have an explanation of why that accession is much colder than the rest (for example, where is this accession from?).

For the impact of the accessions from the coldest regions (Asian and Swedish accessions) on the climate correlations, see our response to Reviewer #1. Briefly, the results hold if they are excluded from the analysis. We have added geographic information for the accessions coming from the warmest and coldest regions to the caption of Figure 1.

14. The Qst vs Fst analysis was used appropriately but there could be an additional explanation for how to interpret figure 7 in regards to how the gray fields are calculated

We have added more details in the calculation of the neutral expectation in the caption of the figure (now Figure 5), with reference to Materials and methods and the original description of this method in Josephs et al., 2019.

15. Unclear what "high plasticity" means in line 227.

This was removed, see comment #12 from Reviewer #1.

Reviewer #3 (Recommendations for the authors):

Line 35: remove so-*

Changed

Line 47-58: the emphasis on CBFs is maybe excessive here as you don't provide compelling evidence for their involvement. Maybe swap for an introduction of the main ecological strategy of Nordic vs Southern adapted accessions?

We have broadened the Introduction with respect to life-history/ecological strategies and also framed the paragraph on cold acclimation thusly.

Line 74-82: Provide the AIC/BIC/deviance or RMSE to justify the use of power vs exponential growth model in materials, but keep it simple here so that the interpretation of the parameter is clearer.

The ANOVA table of the model comparison is now added to Materials and methods.

Figure 2: Upon vernalisattion? I believe you mean cold stratification?

Indeed, thanks for spotting this.

Line 94: This is peripheral but I wonder if there is no biological insight to be gained from fitting a β for each accession. Wouldn't this be informative about the rate of resource accumulation?

The problem with fitting a β for each accession is that β and growth rate (r) describe the speed of rosette size increase together and they are not independent. With that independence I mean that the same size growth pattern can be described by different combinations of r and β. In other words, multiple solutions to the same problem. For this reason we took a pragmatic approach to model r as accession-specific while β is kept fixed. This choice lets r maximally describe the growth-pattern differences between accessions. What, exactly, is physiologically determining β is an interesting question, but outside the scope of this manuscript.

Line 103-128: I understand that some metrics are provided in the figures associated with these sections but it would be clearer to provide some quantitative support (and tests) in the main text.

The correlation coefficients and p-value are now also in the text.

Line 147: how does the metabolite measurement fit?

This refers to the resemblance in geographic pattern, we clarified this in the text.

Line 154: which accessions? Give the rationale for selection based on Figure S4.

The 8 accessions were selected to represent the climatic variation in the full panel. This is clarified in the caption of Figure S8 and in Materials and methods.

Line 162: the null hypothesis of the test is not known.

This is now clarified in Materials and methods.

Line 171-174: I find this argument circular: these genes were identified as differentially expressed in the very same accession I presume (although this is not explicitly stated in the paper).

No, the 251 genes were selected based on published functional studies, so there is no circularity. This has been clarified.

Line 134: relocation of resources towards… I think there could be some information in analysing β's inferred for each accession here.

See response to earlier comment.

Line 360: typo

Fixed.

Line 372: Put the rest of the section in the heritability section.

This section was moved to the description of the heritability calculation, hopefully making it easier to follow (see also comment by Reviewer #1).

Line 392: grammar issue.

Fixed.

Line 462: grammar again.

Fixed.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Reviewer #2 (Recommendations for the authors):

The clarity of the manuscript has been improved particularly in the descriptions of the methods and figure design. A few additional analyses to explore subsets of the data that drive trends in growth rate are included which strengthens the overall conclusion that temperature response of vegetative growth is locally adaptive. However, the manuscript has not addressed two of our initial concerns about the broad framing of the manuscript and the gene expression analyses.

1. The manuscript has been reframed in the context of ecological and life history strategies with a focus on the growth and survival tradeoff, based on feedback from us and other reviewers. While the growth rate results are potentially consistent with what is expected for a growth/survival tradeoff, there is no evidence for the other side of the tradeoff, survival. The manuscript should make this clear. In addition, the current framing includes references to active and passive plasticity, but we are not convinced that faster growth of accessions from warm temperatures "proves" that slow growth of accessions from cold temperatures is active plasticity (lines 125-127) as opposed to another explanation such as less efficient use of resources in accessions from cold temperatures at a later life stage. Additionally, the manuscript notes that "plant growth is known to slow down with increasing size" when justifying its model choice for estimating growth rates. With this in mind, is it a surprise that the populations with initial larger rosettes now show slower growth?

We agree that we cannot make hard claims on the survival aspect, therefore we were always careful in our wording. We now also added explicitly that we did not measure survival itself. (line 238). At the same time, we note (line 245-246) that reciprocal transplant experiments in Arabidopsis thaliana did detect an important contribution of freezing tolerance to fitness variation in sites with strong winters.

We also added a disclaimer paragraph (line 255-259) to make clear that a reduced resource efficiency in later development may cause a similar growth reduction. We still believe that our data is pointing towards active inhibition in trade-off with cold acclimation, however, further work is required to nail down the exact resource relocation and the regulation of this growth inhibition.

2. The gene expression results should be better integrated to support the conclusion that growth rate response is locally adapted. The main results described in the text are (1) the cold acclimation genes are more likely to be differentially expressed between temperatures compared to background genes. This is unsurprising since these genes were initially identified as being differentially expressed in the cold. (2) Gene expression clustering separates accessions from warmer and colder climates and the expression of some clusters is correlated with the temperature of origin. It is unclear if this pattern differs from background genes or the trait divergence expected under neutrality. (3) Genes that were up-regulated in cold (from published experiments) are also generally upregulated in accessions from cold climates. This last result is potentially the clearest link to the adaptive story, but it is not emphasized here and it's also unclear to us if upregulating these genes is a sign of adaptive or maladaptive plasticity. Overall, there is not a clear link between these results and the conclusion about adaptation to cold temperatures.

In addition, while Figure 3 —figure supplement 3 does address our concern about pseudoreplication, Figure 3 —figure supplement 2 still reports misleading p values that should not be included due to pseudoreplication.

1) We agree that this is unsurprising, but in our opinion it is important to show that also in this specific setup this group of genes is more affected by temperature than expected by chance.

2) We added figure 3 - figure supplement 5, which is showing that for most clusters (except cluster F in 16ºC), the correlation coefficients are stronger than expected from random sets of background genes. For this we performed 10,000 permutations using random sets of background genes for each of the clusters, in each temperature.

That the correlations are stronger than expected under neutrality is in our opinion proven by the fact that the correlations are (for most clusters) more significant than expected by chance, which was now tested by doing 10.000 permutations of winter temperature correlations for each of the clusters of cold acclimated genes. These p-values are now corrected in figure 3 —figure supplement 2, which is also tackling the problem of the misleading p-values in that figure.

2) These results could be interpreted many ways, however, they do demonstrate that the expression of known cold-response genes differ geographically, consistent with local adaptation. We now note this in the Discussion (line 229-230).

Line Comments:

The statements made in lines 41-43 should have citations.

We added citations.

Lines 154-156 seem circular.

We don't think this sentence is circular. The first part of the sentence refers to independent experiments, the second to our own results. We adjusted the sentences to make this more clear.

Line 181 change "both" to "all".

We changed it to 'These', referring to the traits and temperatures discussed in the previous sentence. 'all' would be confusing since there was no adaptive differentiation for all traits in all temperatures.

Line 198-199. The phrase "across the genome we detect adaptive differentiation for certain growth parameters." seems unsupported by the data. While the Qst-Fst results show evidence that phenotypic variation is adaptive, there is still no evidence that this trait is shaped by loci 'across the genome'.

We agree that there is no direct evidence for this and removed the statement.

Line 273 should read 160 µmolm-2s-1?

Indeed, thanks for spotting this.

It is not clear how the growth rate temperature response was calculated. Line 91-92 suggests that it is "the slope between the growth rate at 16ºC and 6ºC" but examining specific points in figure 2B, C, and D makes it look like something else is going on. For example, the far left point has a growth rate of ~0.16 in 16C and ~0.052 in 6C, but the response is ~0.072 when we would expect something closer to 0.1 based on the text.

A value closer to 0.1 is indeed expected when the categorical variables '6C' and '16C' are recoded to 0 and 1. Because we wanted to obtain the slope from 6C towards 16C, we specified temperature to be an ordered factor with 6C the lower order and 16 the higher order. The linear regression used to define the slope works a little bit differently with ordered categorical variables, encoding the categorical variables not to 0 and 1, but to -0.70 and 0.70. The slopes we obtained are thus different but perfectly correlated with the slopes obtained from non-ordered categorical variables. We specified the use of order categorical variables in Materials and methods (line 318-321).

Please include percent variance explained on figures with PCs as axes (Figure 5 and Figure 5 supplement).

The percentages of the genetic variation explained by each of the principal components are added to the respective axis labels.

Figure 1—figure supplement 1 is somewhat confusing with the y axis referring to the experiment timeline on the x-axis and to the inset graph of a single day.

The insert is now put in the upper right corner of the figure, this hopefully will take away the confusion and make clearer that the insert axes stand on their own.

The inclusion of the previously published metabolite data is still somewhat confusing. The results report that metabolite measurements were different between accessions from different climates (line 137) but these results are not presented in a figure anywhere. If the authors want to present previously published data that has been reanalyzed to this paper in a figure, that would be fine. However, without that, it seems like this statement should be moved to the discussion.

We think that mentioning the findings of the previously published metabolite data is important to introduce why we look into the expression of cold acclimation genes. Therefore we prefer to keep this sentence for its introductory value in the gene expression analysis.

We added to the text that the metabolite measurements are presented in a previous publication.

https://doi.org/10.7554/eLife.77913.sa2

Article and author information

Author details

  1. Pieter Clauw

    Gregor Mendel Institute of Molecular Plant Biology, Austrian Academy of Sciences, Vienna BioCenter, Vienna, Austria
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Writing - original draft
    For correspondence
    pieter.clauw@gmi.oeaw.ac.at
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9677-8727
  2. Envel Kerdaffrec

    Department of Biology, University of Fribourg, Fribourg, Switzerland
    Contribution
    Resources
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-8667-6850
  3. Joanna Gunis

    Gregor Mendel Institute of Molecular Plant Biology, Austrian Academy of Sciences, Vienna BioCenter, Vienna, Austria
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  4. Ilka Reichardt-Gomez

    Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7964-9352
  5. Viktoria Nizhynska

    Gregor Mendel Institute of Molecular Plant Biology, Austrian Academy of Sciences, Vienna BioCenter, Vienna, Austria
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  6. Stefanie Koemeda

    Plant Sciences Facility, Vienna BioCenter Core Facilities GmbH, Vienna, Austria
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  7. Jakub Jez

    Plant Sciences Facility, Vienna BioCenter Core Facilities GmbH, Vienna, Austria
    Contribution
    Supervision, Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6481-4383
  8. Magnus Nordborg

    Gregor Mendel Institute of Molecular Plant Biology, Austrian Academy of Sciences, Vienna BioCenter, Vienna, Austria
    Contribution
    Conceptualization, Supervision, Funding acquisition, Project administration, Writing - review and editing
    For correspondence
    magnus.nordborg@gmi.oeaw.ac.at
    Competing interests
    Reviewing editor, eLife
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7178-9748

Funding

No external funding was received for this work.

Acknowledgements

We thank past and current members of the Nordborg group for their help in setting up and harvesting these experiments. Thanks also to Daniele Filiault for her valuable comments on the manuscript.

Senior Editor

  1. Christian R Landry, Université Laval, Canada

Reviewing Editor

  1. Regina S Baucom, University of Michigan, United States

Reviewer

  1. Regina S Baucom, University of Michigan, United States

Publication history

  1. Received: February 15, 2022
  2. Preprint posted: February 17, 2022 (view preprint)
  3. Accepted: July 13, 2022
  4. Version of Record published: July 29, 2022 (version 1)

Copyright

© 2022, Clauw et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 898
    Page views
  • 205
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Pieter Clauw
  2. Envel Kerdaffrec
  3. Joanna Gunis
  4. Ilka Reichardt-Gomez
  5. Viktoria Nizhynska
  6. Stefanie Koemeda
  7. Jakub Jez
  8. Magnus Nordborg
(2022)
Locally adaptive temperature response of vegetative growth in Arabidopsis thaliana
eLife 11:e77913.
https://doi.org/10.7554/eLife.77913
  1. Further reading

Further reading

    1. Ecology
    2. Evolutionary Biology
    Zinan Wang, Joseph P Receveur ... Henry Chung
    Research Article

    Maintaining water balance is a universal challenge for organisms living in terrestrial environments, especially for insects, which have essential roles in our ecosystem. Although the high surface area to volume ratio in insects makes them vulnerable to water loss, insects have evolved different levels of desiccation resistance to adapt to diverse environments. To withstand desiccation, insects use a lipid layer called cuticular hydrocarbons (CHCs) to reduce water evaporation from the body surface. It has long been hypothesized that the waterproofing capability of this CHC layer, which can confer different levels of desiccation resistance, depends on its chemical composition. However, it is unknown which CHC components are important contributors to desiccation resistance and how these components can determine differences in desiccation resistance. In this study, we used machine learning algorithms, correlation analyses, and synthetic CHCs to investigate how different CHC components affect desiccation resistance in 50 Drosophila and related species. We showed that desiccation resistance differences across these species can be largely explained by variation in CHC composition. In particular, length variation in a subset of CHCs, the methyl-branched CHCs (mbCHCs), is a key determinant of desiccation resistance. There is also a significant correlation between the evolution of longer mbCHCs and higher desiccation resistance in these species. Given that CHCs are almost ubiquitous in insects, we suggest that evolutionary changes in insect CHC components can be a general mechanism for the evolution of desiccation resistance and adaptation to diverse and changing environments.

    1. Ecology
    Thomas P Smith, Shorok Mombrikotb ... Thomas Bell
    Research Article

    How complex microbial communities respond to climatic fluctuations remains an open question. Due to their relatively short generation times and high functional diversity, microbial populations harbor great potential to respond as a community through a combination of strain-level phenotypic plasticity, adaptation, and species sorting. However, the relative importance of these mechanisms remains unclear. We conducted a laboratory experiment to investigate the degree to which bacterial communities can respond to changes in environmental temperature through a combination of phenotypic plasticity and species sorting alone. We grew replicate soil communities from a single location at six temperatures between 4°C and 50°C. We found that phylogenetically and functionally distinct communities emerge at each of these temperatures, with K-strategist taxa favored under cooler conditions and r-strategist taxa under warmer conditions. We show that this dynamic emergence of distinct communities across a wide range of temperatures (in essence, community-level adaptation) is driven by the resuscitation of latent functional diversity: the parent community harbors multiple strains pre-adapted to different temperatures that are able to ‘switch on’ at their preferred temperature without immigration or adaptation. Our findings suggest that microbial community function in nature is likely to respond rapidly to climatic temperature fluctuations through shifts in species composition by resuscitation of latent functional diversity.