Locally adaptive temperature response of vegetative growth in Arabidopsis thaliana

Read the full article See related articles


We investigated early vegetative growth of natural Arabidopsis thaliana accessions in cold, non-freezing 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 towards cold acclimation and winter survival.

Article activity feed

  1. Evaluation Summary:

    This work investigates natural variation in plant growth plasticity in Arabidopsis thaliana. Notably, the paper shows that plants from cold regions show less growth response to temperature than plants from warmer regions and this variation in response is consistent with local adaptation. These results have clear relevance for those specialized in working on A. thaliana life history.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

  2. Reviewer #1 (Public Review):

    Clauw et al combine large-scale phenotyping, trait-environment correlation analyses, transcriptome profiling, GWAS, and tests of adaptive differentiation to examine the potential for local adaptation in growth among >200 accessions of A. thaliana sampled from a wide range of locations. They show accessions from northern latitudes produced larger seedlings yet slower growth given exposure to colder temperatures compared to accessions from southern latitudes. They provide clear evidence that growth is polygenic and exhibits rather large broad-sense heritabilities, identify candidate loci underlying plant growth, and provide evidence for adaptive differentiation in initial size and growth rate at one of the temperatures examined.

    There are a number of strengths and a few weaknesses of this work. Strong points are the depth of data generated - the authors take size twice per day using 7k plants (5 replicates of 249 accessions, 2 temperature treatments across 3 repeated experiments). They model growth with a power-law function, noting that growth slows with increasing size, and then use these estimates of growth (initial size, growth rate at two temperatures, and response to changes in temperature) to assess correlations to climate (exclusively focusing on mean temperature during the coldest quarter). Another strength is the combination of this dense phenotypic analysis with metabolome, transcriptome, and GWAS analyses to examine the functional genetic basis underlying the potential growth differences between northern and southern populations. They likewise perform a test of adaptive differentiation.

    This work, while extensive, has a few areas that need improvement. Of major concern is that in its current form, it lacks important and useful evolutionary/ecological framing and hypotheses - why is an assessment of local adaptation for growth useful, interesting, and what specific question is this work tackling? Most people likely assume growth is locally adapted. Without broader framing, it is difficult to understand the novelty or importance of the work. There is a rich literature that the authors could tap into (e.g. growth is a key part of life history and trade-offs between growth and survival or reproduction, given environmental cues, is certainly adaptive, yet this has not been considered at such a broad and multi-level analysis before), which would make the relevance of the current manuscript more broadly appreciated. The overall arching hypothesis of the study should be more easily identified in the introduction.

    Next, most take-homes from the phenotypic work are shown from simple regressions yet the methods state that more developed statistical models were employed. We need to see the results of the analyses that drive the assertions. Additionally, more rigorous analyses should be provided: it appears from the regressions that populations located predominantly in Asia (the populations that experience the coldest winter temperatures) are driving the correlations between growth and climate, and even potentially responsible for the adaptive differentiation result. Digging into this further will be important. Further, the wide range of growth plasticity (ie growth rate temperature response) in North Sweden populations call into question some of the simple conclusions drawn in the text. This too should be examined more closely.

  3. Reviewer #2 (Public Review):

    The manuscript's main claim is that there is local adaptation for growth rate response to cold. Specifically, the growth rate of accessions from cold regions is less reduced by cold than the growth rate of accessions from warmer regions (Fig. 4). Additionally, the results of a Qst-Fst-like test suggest that the difference in growth rate response is adaptive. The manuscript hypothesizes that northern accessions actively inhibit their growth because slow growth may be beneficial in cold climates.

    The manuscript has an impressive phenotyping dataset demonstrating variation in initial growth, growth in each treatment, and temperature growth response. The manuscript also clearly demonstrates that these phenotypes are correlated to winter temperature and that the variation in these responses is larger than expected due to drift, consistent with local adaptation. We see these findings as the strongest aspects of the paper.

    A major weakness of the paper is the gene expression analysis. As presented here (Figure 5), it is not clear how these results support the conclusion of adaptation. Furthermore, the methods and results are not clearly described. However, we did find the analysis of how gene cluster expression changed across treatments in accessions from different climates (Figure S5) more interesting. There is promise for these results if they are more clearly presented and discussed, to contribute to the overall argument of the paper by comparing how plastic expression changes relate to evolved expression changes.

    An additional weakness of the paper is the current framing. The introduction immediately begins with a reference to A. thaliana life history and does not link these results to broader questions about the evolution of plasticity that would be interesting to more readers. The manuscript would be improved by connecting growth rate response to temperature with ideas about the evolution of plasticity in the introduction and discussion. For example, active and passive plasticity are defined in the introduction but not returned to later in the discussion, despite the fact that the careful phenotyping done in this paper is a good example of how to distinguish between active and passive plasticity. Similarly, throughout the paper, there could be clearer links between specific results, like the heritability estimates and the GWAS results, and the main argument about plasticity.

  4. Reviewer #3 (Public Review):

    The study reports on a large experiment using a powerful design and thorough control of variation. I am particularly impressed with the full triplication, which makes for an exemplary design. Controlling the amount of soil and thus its density is also very relevant here. The modelling and the characterisation of trait variation are thorough and mostly clear although figures 2 and 3 are not very informative.

    Unfortunately implementing a GWAS and RNA-seq analysis did not yield much insight, with no gene or locus to single out. The study does not provide novel molecular insights and instead places a focus on the different levels of plasticity found across population. This calls for the refocusing of the paper on the trait variation and how the treatment response differs across populations and locations of origin.

    - Issue with the conceptual framework
    Whilst the modelling is thorough, I have some reservations about the conceptual framework and its biological interpretation.
    The plants are first grown for 15 days at 22°C before the treatment (6°C vs. 16°C) starts. The size reached after these 15 days, inferred as the intercept of the power curve is interpreted as an absolute measure of early growth.
    In my mind, M_0 confounds early growth with growth at a warm, near-optimal temperature. The authors need to justify that this is not another measure of plasticity for growth under warmer conditions (ie. that growth during the first 15 days is constant and not/less temperature-dependent). Then there is a high correlation between r's at 6°C, 16°C and their reaction norms. This is quite trivial.

    - Lack of clarity in the presentation of the data
    The authors mention metabolomic data (eg. line 148) that are nowhere to be found in the manuscript. There is no clear disclosure of which eight accessions were retained for RNA-seq. I gather I could use the x-axis label on fig. 5, but this is too much work-out to be clear.
    The authors have tested two modelling frameworks and retained a power-law model. This is justified based on goodness-of-fit of these models but there is too much information about the model that is not being used in the results. It would be better to show the one, best model in the results and keep the alternative model and the justification for not using it in the methods.

    - Some heuristics need further justification or explanation
    A fair bit of the introduction is spent on presenting cold acclimation molecular mechanism but the lowest temperature tested here is 6°C. This is above growth cessation and per se requires a justification for how this temperature is expected to induce cold acclimation response in A. thaliana.
    It is unclear what were the 241 cold-acclimation genes selected and how was enrichment tested? Give the degrees of freedom of the test.
    The testing of adaptive differentiation presents in Fig. 7 the values along PC5&6, I assume that PC1 to 4 were not as visual, but that is also a heuristic that needs to be made explicit.
    Finally, most analyses take into account population stratification using a "correction for structure". It is unclear how the K matrix was computed (how the SNP were selected and what kind of relatedness calculation was implemented). Was it always the same K matrix across analysis (I suspect those used for the GWAS and for the adaptive differentiation test are not the same).