Article activity feed

  1. Reviewer #3 (Public Review):

    In this study, the authors use a pooled GWAS approach combined with a case-control sampling of pairs of apparently drought-damaged and apparently healthy European beech trees in standing populations, to identify locations across the beech genome that may be associated with drought resistance.

    Major strengths:

    • The paired case-control design allows the authors to draw conclusions from standing trees in forests, where the observed phenotypes, though more difficult to interpret, are more relevant than phenotypes observed for tree seedlings under controlled conditions.

    • The paired design avoids many possible complicating factors (conflating influences on the genotype-phenotype relationship). The authors also for example include an important control analysis showing that the genome similarity is comparable among the different groupings, indicating that there are no large and systematic demographic differences among the different groupings. This is not surprising because the trees were taken entirely from a small area in Hessen, Germany.

    • Although the limited geographic coverage itself is not a strength, it does show that genetic differences in stress tolerance or resilience can be detected locally. It is these local differences among trees that may cross-breed that represent the adaptive potential in populations.

    • The genome-wide analysis is conducted to high standards, combining a pool-seq approach pooling individuals from climatic regions and observable stress (e.g. South-stress, South-healthy, North-stress, North-healthy) with high coverage resequencing (20x) of 100 individuals and an analysis of likely linkage among SNPs which significantly varied by stress status.

    • The dataset and the authors' analysis could be a very valuable resource to support ongoing work in forest genetics and climate change, including added value from future monitoring of the same tree populations. The authors point out some of these uses at the end of their Discussion.

    Major weaknesses:

    • The opportunistic use of trees in natural environments (not trees planted for the experiment) is nevertheless a challenge for clear interpretation of the phenotype. The authors aim to identify phenotypes specific to drought stress, which is a complex stress that causes many nonspecific stress responses. They address this by measuring other traits of chosen trees and showing that dried leaves and leaf loss differ, but measures of tree size, canopy closer, and competition do not differ between their drought-stressed and non-stressed classes. Of course there are other causes of dried leaves and leaf loss that cannot be fully excluded from these observations, such as differential damage from insects or pathogens. The authors state that they only chose trees "free from obvious mechanical damage, fungal infestations or other signs of illness", but did not score these phenotypes, measure any correlates, or otherwise record any information regarding these possible confounding factors. It would be more convincing if they had longer-term monitoring data for these trees showing that such differences became apparent after major recent drought events (2018) and were not apparent before these events, if they had other direct measures of water availability and status in the selected trees at relevant timepoints, or if they had measured other more specific indicators of drought stress such as abscisic acid levels at a relevant time point. The authors present long-term climate data from the study area, but no corresponding tree phenotype data to go along with it. This should be addressed.

    • Related to this: the gene candidates (Table S2) include many that are implicated in stress responses, several of these being drought stress responses. In almost all cases these are genes involved in hormonal signaling, protein regulation, growth and development, which are certainly involved in drought responses but are hardly specific to such responses; and there is one involved in spermine synthesis. This plausible involvement, but lack of specificity, is consistent with the gentle gradient of allele frequency changes, and contributions to phenotype prediction, shown in Figs 4 and 5.

    • Also related: the sampling is not representative of the range of Fagus sylvatica. This is most problematic because the genomic locations found to be significant in this association study might not be significant indicators of plant damage in a more diverse dataset. In the worst case, it could be that these trees are so similar in their adaptation to stress, that only small differences in growth and development and environmental signaling can be identified, which each may have a very minor influence; in contrast, if a comparative study were done including more marginal populations which face more frequent drought (while accounting for demographic differences), genes which are generally more important for a robust drought stress response might be identified. This should be discussed with reference to existing literature.

    • The authors do not report the proximity of their pairs (unless I missed this), rather that these were "mutually the closest neighbors with contrasting damage status". Although they do have GPS coordinates for each tree, they do not report how the exact locations were determined and under what conditions, and thus we do not know how precise or accurate these coordinates are. GPS coordinates taken under a leaf-on canopy with a handheld device are likely to have an uncertainty of 10 m and an unknown offset which will differ for each tree due to multipath reflection, so these coordinates cannot be used to judge the relative position of trees in the same population. This can be ameliorated with aerial coordinates, or with a well-anchored reference point in the open combined with a surveying approach to map the trees in reference to this coordinate, but it is not clear if the authors did this. This should be clarified and if necessary, remedied as it is also crucial to the value of the authors' dataset, e.g. for future monitoring of these trees to test predictions.

    • The analysis is rigorous and relevant and suggests important hypotheses regarding the drought resistance or resilience of European beech. However, these hypotheses are neither very precisely articulated, nor tested in this study.

    Did the authors achieve their aims?

    The authors are able to predict membership of an individual in the healthy versus damaged group based on its genotype at significant loci with near 100% accuracy. This does not indicate whether the groupings themselves (healthy versus damaged) are good indicators of tolerance or resilience under drought stress, which may be better supported by other data as well as future monitoring as discussed above, but does support that the GWAS is rigorous.

  2. Reviewer #2 (Public Review):

    This study uses a genome-wide association approach with pool-seq data, which provides a cost-effective alternative to whole genome sequencing of individuals, to dissect the genetic basis of drought resistance in European beech. European beech is an ecologically important forest tree species and drought resistance is a trait that is likely to be becoming increasingly relevant to the survival of these trees as climate change leads to more frequent and prolonged periods of drought. Knowledge of the genetic basis of such traits can help with management of tree populations in the face of increasing threats, especially when such information is used to develop tools for predicting individuals that are likely to have the highest chances of survival, such as was done in this study.

    However, despite the potential importance of this study's findings for those concerned with protecting and managing forests, it is not currently clear that the data analysed meet some of the underlying assumptions of the methods used and essential details of some of the analyses are missing. Whilst this may simply be a case of more information being needed to confirm that the data are suitable for the approaches used, and that all relevant controls and quality checks have been conducted, at the moment the reader cannot be entirely sure that the analyses have been performed in an appropriate way or that the results are robust. Moreover, there are inconsistencies throughout the paper in relation to the number of individuals included in the various analyses, which creates confusion and casts doubt on the rigour of the investigation.

  3. Reviewer #1 (Public Review):


    The authors present a very interesting and appealing approach to relate physiological damages observed after the extreme drought in 2018 to drought resistance genes in the European tree species Fagus sylvatica. Climate change and observed drought damages are a pressing issue for the forestry sector. The species is widespread through Europe and an important timber tree. Sampling took place in Hessen, Germany, in a 90x150km area. The authors used poolGWAS and a recently established reference genome to infer associated SNP loci by contrasting allele frequencies with replicated pools of drought susceptible and resistant trees. The authors also test the detected loci by a linear discriminant analysis based on an additional set of trees from the same region (SNP assay). The authors found systematic and quantitative genomic differences for drought susceptibility in the sampled population based on 7 significant loci located within genes and a few more loci (12) to be located close to genes associated with drought susceptibility in previous studies. The authors conclude that the significant loci found help to accelerate and monitor adaptation of beech to climate change. And they conclude from their results that there is enough genetic variation in beech to adapt to increasing drought and future climate change.


    • The authors used a so-called XP-GWAS or poolGWAS approach, a relatively new (Yang et al. 2015, Zou et al. 2016), time and cost efficient whole genome sequencing method. By using a strictly pair-wise sampling design (XP-GWAS), pitfalls of traditional GWAS studies are avoided. In addition, the authors make use of the recently established reference genome of Fagus sylvatica (however, the used version not yet published). This genetic approach was successfully used in a similar way for crop plants and another broad-leaved tree species (Fraxinus excelsior, Stocks et al. 2020). Sampling effort and sequencing resolution seem to be adequate according to Yang et al. 2015 and yield 106 significantly phenotype associated SNP loci.

    • Environmental variables used in this study contain very detailed climate data from 1950 onwards and mean monthly evaporation potential from 1991 onwards for all sites used for sampling. There are other drought indicators that better correlate with vitality traits such as the ratio of actual versus potential evapotranspiration or minimal site water balance (based on local field capacity) as has been shown for other central European beech forest sites (Braun et al. 2020, Schweiz Z Forestwes 171). But such climate data are difficult to obtain for dense and large-scale samplings without a monitoring background.


    • Small-scale differences in soil water availability and other possible abiotic or biotic factors at the sampling scale of tree pairs are not considered in this study. In the first hypothesis statement, local environmental variation is ruled out by the fact, that the selected trees stand next to each other and, thus, are neighboring trees in the forest. But such small-scale variation should be at least be considered and discussed Literature recommendations: Kätzel 2008, Bolte et al. 2008 (see Sutmöller et al. 2008), Carrière et al. 2019. Furthermore, the (maximum) distance between the pairs of trees is not stated.

    • The selection criteria used for damages versus resistant trees are unclear and the stated selection criteria are not specific to drought stress but rather more general stress indicators (see Wohlgemuth et al. 2020, Schweiz Z Forestwes 171). Traits used seem not to be consistent with protocols from internationally recognized monitoring networks (e.g. ICP forest manual, Attention should be paid also to the fact that observed stress symptoms have a multivariate background. Modelling analysis of long-term data show that other environmental factors such as N deposition are correlated with the changes in health status observed in central European beech forests and show interactions with drought indicators (Braun et al. 2020, Schweiz Z Forstwes 171, see also Pflug et al. 2018).

    • The authors found only few drought-associated loci (7) to be located within genes and a few more (12) to be located close to genes associated with drought susceptibility in previous studies. Although most of the genes found in this study had putative homologs in other plant species, none were involved in a transcriptomic study on drought response in beech saplings (Müller et al. 2017, see discussion paragraph). It is questionable whether these systematic and quantitative genetic differences are large enough to infer that there is a genomic basis for drought resistance in beech and that genetic variation is large enough in this species to cope with future climate change, also with respect to its distribution across Europe.

    The authors study the genomic basis of drought susceptibility and found systematic and quantitative genomic differences. However, the results seem not to be very strong in supporting the conclusions drawn. It is not clear whether the power of the GWAS study is affected by the precision of phenotyping, pool size, selection intensity, marker density or the depth of sequencing. Moreover, geographic limitations of the study and how that limits conclusions with respect to the species range have not been considered. This could, for example, be based in estimates of genetic variation across the species range such as in Magri (2006) New Phytologist and other newer references.

    This study presents an important issue in forestry and forest ecology and implemented a recently developed, time and cost efficient, genetic approach that has only rarely been applied to woody long-lived species.

  4. Evaluation Summary:

    This study uses a genome-wide association approach combining pool-seq data with whole-genome resequencing, which provides a cost-effective means to scale genome-wide association studies to a larger number of individuals, to dissect the genetic basis of drought resistance in several German populations of European beech. European beech is an ecologically important forest tree species and drought resistance is a trait that is likely to be becoming increasingly relevant to the survival of these trees as climate change leads to more frequent and prolonged periods of drought. Knowledge of the genetic basis of such variation existing within current populations can help with management of those populations in the face of increasing threats, especially when such information is used to develop tools for predicting individuals that are likely to have the highest chances of survival, and to suggest hypotheses regarding traits and genetic material which will be important for the future of forests.

    (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 and Reviewer #3 agreed to share their names with the authors.)