Role of Morpho-Phenological Traits in Passive Resistance to Fusarium Head Blight in Wheat
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsDear Editors and Authors,
I read with interest the manuscript entitled "Role of Morphological and Phenological Traits in Passive Resistance to Fusarium Head Blight in Wheat". In this study, a complex of eight morpho-phenological traits and their association with overall FHB resistance was investigated. The aims of this study were as follows: to evaluate the variation in the morpho-phenological traits in a spring wheat set of 322 cultivars across two years; to evaluate FHB severity under artificial inoculation in two-year field trials; to conduct Spearman correlation analysis among morpho-phenological traits and FHB severity; and to apply random forest regression (RFR) analysis to determine the contribution of the morpho-phenological traits for FHB resistance, measuring prediction accuracy between RFR-predicted and observed FHB severity scores and the relevant importance of the traits. The subject of the article is important and has great relevance for the scientific environment of the study area. Therefore, the manuscript needs some adjustments so that it can then be forwarded to the publication process. The manuscript has the potential for publication in this journal Agriculture and needs the following adjustments:
ABSTRACT
Lack of quantitative data: The abstract describes the correlations in a qualitative way ("strong negative association", "low correlation"). For a scientific article, it is imperative to present the correlation coefficients (r) and the significance levels (p-value). Without the numbers, the reader cannot assess the real strength of the findings.
Model Accuracy: When mentioning the 65% accuracy in the "Random Forest" regression analysis, it is crucial to specify whether this value refers to the training set or cross-validation. If it is in training, 65% is low; if it is in validation, it is acceptable. The distinction changes the interpretation of the quality of the model.
INTRODUCTION
Definition of resistance x escape: The introduction mixes the concepts of "Passive Resistance" (physical barriers, such as ear density or waxiness) with "Disease Escape" (phenological asynchrony). It is essential that the authors differentiate these concepts clearly. The escape is not a genetic resistance per se, but rather a temporal chance.
Limitation of spray inoculation: The authors state that the spray is efficient to assess type I resistance. However, they should explicitly recognize that, by forcing the inoculum under pressure, this method can "overcome" the natural physical barriers of the plant (such as the closure of the glumes), underestimating the true passive resistance that would occur in a natural infection in the field.
MATERIAL AND METHODS
Critical time discrepancy (2025 data): This is the most serious methodological flaw. The manuscript indicates that disease severity was measured in 2022/2023, but "Anther Extrusion" (a highly environmentally sensitive trait) was measured in 2025.
Suggestion: Correlating disease from one year with phenotyping from another year assumes that the Genotype x Environment (GxE) interaction is null, which is biologically incorrect for floral traits. The authors need to statistically justify that the extrusion of anthers is stable between years or admit this as a severe limitation.
Use of plastic bags: The method describes covering the ears with plastic bags for 48 hours after inoculation.
Tip: The use of bags creates a saturated artificial wet chamber. This cancels out important effects of passive resistance, such as plant height (which affects aeration and distance from the soil inoculum) and ear laxity (which affects drying speed). By creating a perfect environment for the fungus, the authors may have masked the morphological differences they wanted to test. This needs to be discussed.
RESULTS
Plant height and correlation: The results show an inconsistent correlation of plant height with disease (significant in one year, not significant in the next).
Suggestion: This contradicts the vast literature on FHB. The authors should discuss whether this lack of correlation is due to the inoculation method (artificial spray) that eliminates the advantage of tall plants in avoiding the natural spore splashes of the soil.
Phenology and escape: It is observed that later genotypes had less disease (moderate negative correlation).
Suggestion: It is vital to analyze the climate data from the experimental years. If there was a dry period during flowering of the late varieties, the result is purely "Disease Escape" and not genetic resistance. I recommend overlaying the precipitation graph with bloom dates to prove or disprove the escape.
DISCUSSION
Rht genes (dwarfism): Discussion of Rht alleles indicates that dwarf plants are more susceptible.
Suggestion: Further discussion: Does susceptibility come from reduced height (microclimate) or from a pleiotropic effect of the Rht gene (which alters hormonal sensitivity to gibberellins/jasmonic acid)? Because height per se did not correlate well in the results, it is likely that susceptibility comes from the compact density of the ear caused by the Rht gene, rather than height. This physiological distinction would greatly enrich the article.
Study Limitations: A robust limitations section was missing. The authors should list the use of data from different years and artificial inoculation as factors that restrict the extrapolation of results to actual farmers' field conditions.
CONCLUSIONS
Finish with a sentence about integration. Conclude that "passive resistance" is not a one-size-fits-all solution, but must be pyramided with physiological resistance genes (Type I and II) to obtain truly durable wheat varieties.
Author Response
Thank you very much for your thorough review and valuable comments. Please find point-by-point responses below and the attached revised manuscript, in which all changes are visible. The new and updated text is highlighted in red colour.
Comment 1: Lack of quantitative data: The abstract describes the correlations in a qualitative way ("strong negative association", "low correlation"). For a scientific article, it is imperative to present the correlation coefficients (r) and the significance levels (p-value). Without the numbers, the reader cannot assess the real strength of the findings.
Response 1: Thank you for pointing it out. We agree. Therefore, we have included the correlation coefficients and significance levels in the abstract: “Correlation analysis demonstrated a strong negative association between FHB severity and phenological traits: days to heading (r=-0.43, p<0.001), days to flowering (r=-0.39, p<0.001) and a low to medium negative correlation between FHB resistance and spike length (r=-0.29, p<0.001), spikelets per spike (r=-0.26, p<0.001) in average across two years. Furthermore, there was a significant negative but weak association between anther extrusion and FHB severity (r=-0.21, p<0.001).”
Comment 2: Model Accuracy: When mentioning the 65% accuracy in the "Random Forest" regression analysis, it is crucial to specify whether this value refers to the training set or cross-validation. If it is in training, 65% is low; if it is in validation, it is acceptable. The distinction changes the interpretation of the quality of the model.
Response 2: We agree that this is important information and should be mentioned. This accuracy was noticed in cross-validation set; therefore we have modified the text accordingly: “Random forest regression analysis demonstrated that a complex of eight morpho-phenological traits predicted FHB severity with an accuracy of 65% in 2023 and 57% in cross-validation sets across two years.”
Comment 3: Definition of resistance x escape: The introduction mixes the concepts of "Passive Resistance" (physical barriers, such as ear density or waxiness) with "Disease Escape" (phenological asynchrony). It is essential that the authors differentiate these concepts clearly. The escape is not a genetic resistance per se, but rather a temporal chance.
Response 3: Your comment is valid. Passive resistance consists of different components, and from some perspectives, “disease escape” might be excluded from “passive resistance” and treated as a separate indirect component of resistance. As you noted, it belongs to phenological asynchrony. However, we decided not to raise this topic in this study and to follow the classic general definition of passive resistance, according to which phenological traits (heading and flowering) are classified as passive resistance. We have included more references regarding passive resistance in the text (Mesterhazy, 1995; Zhang et al., 2020; Nannuru et al., 2022).
Comment 4: Limitation of spray inoculation: The authors state that the spray is efficient to assess type I resistance. However, they should explicitly recognize that, by forcing the inoculum under pressure, this method can "overcome" the natural physical barriers of the plant (such as the closure of the glumes), underestimating the true passive resistance that would occur in a natural infection in the field.
Response 4: Thank you for pointing this out. We have included limitations of this study in the Discussion section: “However, some limitations of this study should be noted. The trials were performed using artificial inoculation, and the inoculated spikes were covered with plastic bags for 48 h. This resulted in artificially created non-natural conditions, which were highly favorable for fungal penetration and overcoming Type I resistance, thereby causing higher disease pressure than it would be under natural conditions. Therefore, the contribution of passive resistance may be more essential under natural conditions, and the results of this study cannot be directly transformed to the field scale”.
Comment 5: Critical time discrepancy (2025 data): This is the most serious methodological flaw. The manuscript indicates that disease severity was measured in 2022/2023, but "Anther Extrusion" (a highly environmentally sensitive trait) was measured in 2025.
Suggestion: Correlating disease from one year with phenotyping from another year assumes that the Genotype x Environment (GxE) interaction is null, which is biologically incorrect for floral traits. The authors need to statistically justify that the extrusion of anthers is stable between years or admit this as a severe limitation.
Response 5: We agree that including anther extrusion observations from a year that is outside the experiment is a methodological weakness. However, we applied the calculation of BLUEs, modeling year-to-year differences. The following information was added to the Materials and Methods section: “The best linear unbiased estimators (BLUEs) were calculated using META-R v.6.04, where genotypes were fitted as fixed effects, and environments, replicates, and genotype-by-environment interactions were treated as random effects. This enabled the production of BLUEs adjusted for year effects, modeling year-to-year differences as random variance”.
In the discussion, we mentioned: “It is noteworthy to mention that the anther extrusion was measured in 2025 instead of 2022. Although BLUEs were calculated to adjust for year effects, the results may potentially be biased by environmental conditions specific to 2025.”
Comment 6: Use of plastic bags: The method describes covering the ears with plastic bags for 48 hours after inoculation.
Tip: The use of bags creates a saturated artificial wet chamber. This cancels out important effects of passive resistance, such as plant height (which affects aeration and distance from the soil inoculum) and ear laxity (which affects drying speed). By creating a perfect environment for the fungus, the authors may have masked the morphological differences they wanted to test. This needs to be discussed.
Response 6: The limitations of artificial inoculation and covering with plastic bags were mentioned in the response 4.
Comment 7: Plant height and correlation: The results show an inconsistent correlation of plant height with disease (significant in one year, not significant in the next).
Suggestion: This contradicts the vast literature on FHB. The authors should discuss whether this lack of correlation is due to the inoculation method (artificial spray) that eliminates the advantage of tall plants in avoiding the natural spore splashes of the soil.
Response 7: The effect of disease pressure was mentioned in the discussion section: “The dependence of passive resistance on disease pressure was noticed by Mesterhazy (1995), who reported that under a strong artificial disease background, non-significant differences in plant resistance to Fusarium head blight (FHB) were observed among various plant height classes [18]. Buerstmayr and Buerstmayr (2022) also reported the minimal effect of plant height on FHB under greenhouse conditions”. In addition, we discussed that “The potential effect of semi-dwarfing alleles may not have been detected because of the imbalanced group sizes, which resulted in reduced statistical power in our study.”
Comment 8: Phenology and escape: It is observed that later genotypes had less disease (moderate negative correlation).
Suggestion: It is vital to analyze the climate data from the experimental years. If there was a dry period during flowering of the late varieties, the result is purely "Disease Escape" and not genetic resistance. I recommend overlaying the precipitation graph with bloom dates to prove or disprove the escape.
Response 8: Meteorological conditions during flowering are demonstrated in the figure 1. Different conditions for late and early maturing genotypes are mentioned in the discussion: “In our results, we observed that the late-maturing genotypes showed lower severity than the early maturing genotypes. This might be because the late genotypes, compared to the early maturing genotypes, escaped the peak infection period. Weather conditions at the time of inoculation significantly affect the spread of fungal infection, which may increase severity and promote the accumulation of mycotoxins”.
Comment 9: Rht genes (dwarfism): Discussion of Rht alleles indicates that dwarf plants are more susceptible.
Suggestion: Further discussion: Does susceptibility come from reduced height (microclimate) or from a pleiotropic effect of the Rht gene (which alters hormonal sensitivity to gibberellins/jasmonic acid)? Because height per se did not correlate well in the results, it is likely that susceptibility comes from the compact density of the ear caused by the Rht gene, rather than height. This physiological distinction would greatly enrich the article.
Response 9: We have updated the discussion concerning Rht genes: “However, over time, researchers have reported the impact of both Rht gib-berellin-insensitive alleles (Rht-B1b and Rht-D1b) on FHB susceptibility. A similar negative pleiotropic effect on FHB resistance was found for gib-berellin-sensitive alleles of Rht8. Miedaner and Voss (2008), observed an increase in susceptibility due to Rht-B1b, Rht-D1b, and Rht8c alleles by 19%, 52%, and 19%, respec-tively [34]. Many studies suggest that susceptibility to FHB does not result from reduced plant height per se, but appears to be a direct effect of the Rht-D1b loci [32,72–75]. Miedaner et al., 2022 found that Rht24b did not affect FHB resistance proving that susceptibility to Rht genes is not conferred solely by reduced height [74].” In addition, the group sizes with Rht and non-Rht alleles were imbalanced. That could mask pleotropic effects of semi-dwarfing alleles. The effects of disease pressure and imbalanced group sizes are discussed in Discussion section. Ear density did not have large importance on FHB resistance in our study.
Comment 10: Study Limitations: A robust limitations section was missing. The authors should list the use of data from different years and artificial inoculation as factors that restrict the extrapolation of results to actual farmers' field conditions.
Response 10: Limitations of the study have now been explicitly discussed in the Discussion section: “However, some limitations of this study should be noted. The trials were performed using artificial inoculation, and the inoculated spikes were covered with plastic bags for 48 h. This resulted in artificially created non-natural conditions, which were highly favorable for fungal penetration and overcoming Type I resistance, thereby causing higher disease pressure than it would be under natural conditions. Therefore, the contribution of passive resistance may be more essential under natural conditions, and the results of this study cannot be directly transformed to the field scale.”
Comment 11: Finish with a sentence about integration. Conclude that "passive resistance" is not a one-size-fits-all solution, but must be pyramided with physiological resistance genes (Type I and II) to obtain truly durable wheat varieties.
Response 11: Thank you for your comment. We have accordingly extended the conclusion section: “However, the effectiveness of passive resistance depends highly on environmental conditions. Breeding for passive resistance based on morpho-phenological traits cannot be applied as a standalone strategy. Wheat breeding for FHB resistance requires the integration of both active and passive resistance mechanisms to develop durable resistance.”
Reviewer 2 Report
Comments and Suggestions for AuthorsTo the Editor,
The manuscript titled "Role of Morphological and Phenological Traits in Passive Resistance to Fusarium Head Blight in Wheat" by Shayan Syed et al. presents a comprehensive multi-year assessment of how eight morpho-phenological traits contribute to passive resistance against Fusarium head blight (FHB) in a diverse panel of 332 spring wheat genotypes. The study combines well-designed field inoculation experiments with an appropriate suite of statistical tools, including non-parametric analyses, mixed-model BLUEs, and Random Forest regression.
Overall, the manuscript is very readable and interesting and addresses an important question in wheat pathology and breeding, but it should be improved. The authors provide strong evidence that phenological traits—particularly days to heading and flowering—along with anther extrusion, are key contributors to reduced FHB severity under field conditions. The application of machine learning provides additional interpretive value and supports the conclusion that a complex of passive resistance traits explains a substantial portion of the phenotypic variation.
I believe major revisions are necessary before it can be accepted. Several issues require clarification, including the integration of anther extrusion data collected in 2025, a deeper interpretation of genotype × environment interactions, and a more thorough analysis of Rht allele effects given their known association with FHB susceptibility. Additionally, the discussion of spike architecture and the interpretation of the Random Forest results would benefit from refinement and stronger contextualization. These improvements will enhance the clarity, coherence, and impact of the manuscript.
The issues that I would like the authors to improve or clarify include the following:
- The title is accurate but wordy. Please, consider shortening for clarity.
- The abstract is well written but too long; repetition of statements about correlations reduces impact. Also, please add exact ranges/r values for correlation to avoid ambiguity (will be better as “strong negative association” and “low to medium negative correlation”).
- The introduction is well written, but several references describe well-known FHB history, and it is better to reduce redundancy. Epidemic history could be combined into a single paragraph. Since objectives 1 and 2 both refer to trait evaluation, could the authors take into consideration combining them to shorten the number of objectives and so make them more concise?
- Tables and figures should be simplified for broader readability. Some statistical tables (e.g., Table 1) contain redundant information and could be streamlined for clarity. Figures are generally clear, but all abbreviations should be defined in figure captions. The readability of Figure 1 (weather patterns) would benefit from clearer axis labels or separation of genotype count vs. meteorological data.
- Materials and Methods: please explain why anther extrusion (AE) was measured in 2025, while the study is described as a 2022–2023 analysis. Are AE 2025 readings comparable to 2022–2023 phenology? Were environmental differences accounted for?
- Results: referring to Rht alleles, authors should explicitly acknowledge Type II error risk rather than lightly mentioning imbalance. Also, sample size imbalance (only 6–7% dwarfing alleles) makes statistical comparisons underpowered.
Correlation are well presented, but please take into consideration adding p-values directly to the correlation figure for clarity.
Tables referring to genotype effects contain too much raw information for the main text. Consider moving parts to supplementary material.
- Discussion: The authors are linking findings to literature, but my suggestion is to reduce over-citation of basic concepts and focus more on interpreting the study’s own data.
Spike density discussion is too brief given its contradictory results. The authors should expand this section with methodological explanations (artificial vs natural inoculation, microclimate differences, and G×E effects).
Also, referring to Rht genes, I consider that much stronger discussion is needed regarding underpowered sample size. Please, add a sentence stating that conclusions cannot be generalized due to sample imbalance.
- Conclusions are well summarized but slightly overclaim predictive power. Please add exact % variance explained values
Author Response
Thank you very much for taking the time to review this manuscript. Please find point-by-point responses below and the attached revised manuscript, in which all changes are visible. The new and updated text is highlighted in red colour.
Comment: The title is accurate but wordy. Please, consider shortening for clarity.
Response: We shortened the title slightly, replacing “Morphological and Phenological traits” with “Morpho-phenological traits”.
Comment: The abstract is well written but too long; repetition of statements about correlations reduces impact. Also, please add exact ranges/r values for correlation to avoid ambiguity (will be better as “strong negative association” and “low to medium negative correlation”).
Response: We reviewed the abstract and have done some changes to avoid redundancy. Moreover, we added the exact values along with verbal (weak or strong) association for clarity.
Comment: The introduction is well written, but several references describe well-known FHB history, and it is better to reduce redundancy. Epidemic history could be combined into a single paragraph. Since objectives 1 and 2 both refer to trait evaluation, could the authors take into consideration combining them to shorten the number of objectives and so make them more concise?
Response: The objectives 1 and 2 have been merged to avoid any repetition. Besides, to reduce redundancy, we excluded lines 33-34 from the 1st paragraph in introduction.
Comment: Tables and figures should be simplified for broader readability. Some statistical tables (e.g., Table 1) contain redundant information and could be streamlined for clarity. Figures are generally clear, but all abbreviations should be defined in figure captions. The readability of Figure 1 (weather patterns) would benefit from clearer axis labels or separation of genotype count vs. meteorological data.
Response: The columns “Residual portion” and “p-value” were deleted to avoid redundancy. The captions were extended in Figure 1. More extended captions should improve clarity in Figure 1.
Comment: Materials and Methods: please explain why anther extrusion (AE) was measured in 2025, while the study is described as a 2022–2023 analysis. Are AE 2025 readings comparable to 2022–2023 phenology? Were environmental differences accounted for?
Response: Anther extrusion was not scored in 2022 and evaluated in 2025 because of unplanned and unforeseen circumstances. To adjust for environmental differences, BLUEs were calculated. We mentioned in methodology: “The best linear unbiased estimators (BLUEs) were calculated using META-R v.6.04, where genotypes were fitted as fixed effects, and environments, replicates, and genotype-by-environment interactions were treated as random effects. This enabled the production of BLUEs adjusted for year effects, modeling year-to-year differences as random variance”.
Additionally, in the discussion, we mentioned: “It is noteworthy to mention that the anther extrusion was measured in 2025 instead of 2022. Although BLUEs were calculated to adjust for year effects, the results may potentially be biased by environmental conditions specific to 2025.”
Comment: Results: referring to Rht alleles, authors should explicitly acknowledge Type II error risk rather than lightly mentioning imbalance. Also, sample size imbalance (only 6–7% dwarfing alleles) makes statistical comparisons underpowered.
Response: We agree. Type II error was acknowledged and discussed in the Discussion section: “The potential effect of semi-dwarfing alleles may not have been detected because of the imbalanced group sizes, which resulted in reduced statistical power in our study. The ratios between the non-Rht and Rht-B1b and non-Rht and Rht-D1b groups were 1:13 and 1:15, respectively. This imbalance between the wild type and semi-dwarfing allele groups increased the risk of failing to detect a true effect (Type II error).”
Comment: Correlation are well presented, but please take into consideration adding p-values directly to the correlation figure for clarity.
Response: The level of significance for each coefficient is depicted in the graph. Adding p-values will not provide essential information, and may instead introduce redundancy and reduce clarity.
Comment: Tables referring to genotype effects contain too much raw information for the main text. Consider moving parts to supplementary material.
Response: Thank you for the suggestion. We have carefully revised it. The columns “Residual portion” and “p-value” were removed. The rest of the information provides essential information to interpret the results. Therefore, we believe that it is appropriate to keep it in the main text.
Comment: Discussion: The authors are linking findings to literature, but my suggestion is to reduce over-citation of basic concepts and focus more on interpreting the study’s own data.
Response: We appreciate your suggestion; we have made some changes in discussion section to avoid over-citation.
Comment: Spike density discussion is too brief given its contradictory results. The authors should expand this section with methodological explanations (artificial vs natural inoculation, microclimate differences, and G×E effects).
Response: We agree that this spike density could be discussed in more detail; however, the discussion section is already lengthy. Therefore, we tried to focus more on key findings.
Comment: Also, referring to Rht genes, I consider that much stronger discussion is needed regarding underpowered sample size. Please, add a sentence stating that conclusions cannot be generalized due to sample imbalance.
Response: Thank you for your comment. We have mentioned “Therefore, the findings concerning the effect of semi-dwarfing alleles cannot be generalized.” (lines 593-594)
Comment: Conclusions are well summarized but slightly overclaim predictive power. Please add exact % variance explained values
Response: We have included R2 in the figure 4 a,b and updated the conclusions: “Random forest regression analysis demonstrated that eight to seven morpho-phenological traits predicted FHB severity in cross-validation sets with an accuracy of 65% in 2023, 28% under high disease pressure in 2022, and 57% across two years, explaining 42%, 8%, and 32% of the phenotypic variance (R²), respectively.”
Reviewer 3 Report
Comments and Suggestions for AuthorsThis manuscript presents a well-structured and methodologically sound investigation into the role of morphological and phenological traits in passive resistance to Fusarium Head Blight (FHB) in spring wheat. The study leverages a large and diverse panel of 332 genotypes, multi-year field trials, robust statistical analyses, and machine learning approaches. The topic is highly relevant to wheat breeding for disease resistance. The manuscript is generally clear, the experiments are well-described, and the conclusions are largely supported by the data. However, several areas require clarification, methodological justification, and more nuanced interpretation before publication.
- The manuscript emphasizes evaluating “overall resistance” using spray inoculation, which is stated to assess both Type I and Type II resistance. However, the distinction between “overall resistance,” “Type I,” and “Type II” becomes blurred in the interpretation. For instance, traits like anther extrusion are strongly linked to Type I resistance. The discussion should more clearly frame how the measured “FHB severity” relates to the classic Type I/II framework. A clearer operational definition of “overall resistance” as measured in this study would strengthen the manuscript.
2.The discrepancy between Spearman correlation coefficients and Random Forest (RF) permutation importance scores for certain traits is intriguing and noted by the authors. However, the explanation provided remains somewhat speculative. To strengthen this key finding, consider adding partial dependence plots (PDPs) or individual conditional expectation (ICE) plots for the top predictors from the RF model. This would visually demonstrate the nature of the relationships and greatly enhance the interpretation.
3.The authors correctly identify the issue of imbalanced group sizes (only ~12% of genotypes carry semi-dwarfing alleles) and the potential for a Type II error. This limitation severely constraints the conclusions that can be drawn about Rht effects in this specific panel.
4.“The effect of semi-dwarfing alleles could be negligible because of the imbalanced group sizes in our study” should be rephrased. The data does not show the effect is negligible; it shows the study was underpowered to detect an effect if one exists. It is recommended to soften claims about Rht effects and present this primarily as an observation on the allele frequency in this particular germplasm set, with the statistical limitation clearly stated.
5.The stark difference between 2022 (high disease pressure) and 2023 (low disease pressure) is a strength, clearly showing the context-dependency of passive resistance. However, the prediction accuracy was very low (28%) in 2022. The discussion should more deeply address why the model failed under high pressure. Was it simply because passive mechanisms are overwhelmed, or could it be that the relationship between traits and severity becomes non-linear or even reversed under extreme stress? This is a critical point for breeders who need to know the limits of passive resistance.
6.Epsilon-squared values are presented, which is good for non-parametric data. However, for traits like “Days to heading” and “Days to flowering” in Table 3, the genotype effect is reported as negative (ε² = -0.07, -0.10). Epsilon-squared, as a measure of variance explained, should not be negative. This suggests a potential error in calculation or application of the formula for these specific cases. This must be checked and corrected.
7.Figure 1: The figure is conceptually good but visually very busy. Consider creating two separate panels for temperature/precipitation and genotype counts, or using a dual-axis plot with greater care to distinguish the data types.
8.The abstract states "332 spring wheat genotypes" but later in the aims (page 3) and results, "322" is sometimes used. Ensure consistency (332 is correct based on Methods).
9.The transition from general FHB background to the specific gap addressed by the study is good. Consider adding a sentence explicitly stating that most studies focus on individual traits or small combinations, whereas this study aims to evaluate a complex of eight traits simultaneously using advanced analytics.
10.Clarify the sowing date(s) for each year, as this is the reference point for "days to heading/flowering."
11.Anther extrusion for 2025 (AE_2025) is included in a table focused on 2022 and 2023. Consider a separate note or table for the 2025 AE data to avoid confusion.
12.The caption for Figure 3 mentions "across 3 years," but the plots are for 2022 and 2023. FHB severity data for 2025 is not mentioned. Correct the caption.
13.The discussion is comprehensive but somewhat lengthy. It could be tightened by focusing more on synthesizing the novel findings of this study rather than extensively recounting literature that agrees or disagrees.
14.When discussing contradictory findings in the literature (e.g., on awns, heading date), explicitly link back to how your results, especially from the RF model which accounts for interactions and context, might help reconcile these differences.
Author Response
Thank you very much for your comprehensive review and valuable comments. Please find point-by-point responses below and the attached revised manuscript, in which all changes are visible. The new and updated text is highlighted in red colour.
Comment 1: The manuscript emphasizes evaluating “overall resistance” using spray inoculation, which is stated to assess both Type I and Type II resistance. However, the distinction between “overall resistance,” “Type I,” and “Type II” becomes blurred in the interpretation. For instance, traits like anther extrusion are strongly linked to Type I resistance. The discussion should more clearly frame how the measured “FHB severity” relates to the classic Type I/II framework. A clearer operational definition of “overall resistance” as measured in this study would strengthen the manuscript.
Response 1: Thank you for the suggestion, we incorporated your recommendations and defined the definition of “overall resistance”: “Overall resistance combines type I and II, representing resistance to both fungal penetration and further disease spread. Therefore, the evaluation of overall resistance is preferable in practical breeding, as it provides information about the two types of resistance.”
Comment 2: The discrepancy between the Spearman correlation coefficients and Random Forest (RF) permutation importance scores for certain traits is intriguing and noted by the authors. However, the explanation provided remains somewhat speculative. To strengthen this key finding, consider adding partial dependence plots (PDPs) or individual conditional expectation (ICE) plots for the top predictors from the RF model. This would visually demonstrate the nature of the relationships and greatly enhance the interpretation.
Response 2: Thank you very much for your valuable comment. We included partial dependence plots (PDPs) and individual conditional expectation (ICE) plots to analyse relationships between morpho-phenological traits and predicted FHB severity. A new subtitle 3.8 is added into the Results section: “3.8. Relationships between morpho-phenological traits and predicted FHB severity.
PDP and ICE plots were built to reveal how the learning algorithm produced predictions in the Random Forest model. The correspondence between the observed morpho-phenological values, their range, and their effects on the predicted FHB severity are shown in Figures 6–7. The direction and size effects are indicated by deviations from zero on the y-axis. The shift into negative values on the y-axis corresponds to a predicted reduced FHB severity, whereas positive values demonstrate that observations at a specific range were associated with increased FHB severity. Threshold effects were detected for certain traits. For example, awn length had a strong effect on FHB severity only in the range from 8 to 9. Anther extrusion, days to flowering, and days to heading showed threshold effects. Plants that had anther extrusion scores of 4–5, as well as late flowering/heading, were strongly associated with reduced FHB severity in the Random Forest model (Figures 6-7).”
Discussion section has been extended as well: “In our study, a threshold effect was evident for anther extrusion: observations with values up to two scores had nearly zero effects, while values above three scores were associated with a rapidly increasing impact on FHB resistance. Additionally, thresholds were observed for days to flowering and heading (Figures 6-7). The curvature of the PDP lines may indicate that the relationships between the traits and FHB severity captured by the Random Forest model were predominantly non-linear (Figures 6-7).”
Comment 3: The authors correctly identify the issue of imbalanced group sizes (only ~12% of genotypes carry semi-dwarfing alleles) and the potential for a Type II error. This limitation severely constraints the conclusions that can be drawn about Rht effects in this specific panel.
Response 3: We agree. Therefore, we do not focus on the effects of Rht alleles in this study. We have mentioned: “Therefore, the findings concerning the effect of semi-dwarfing alleles cannot be generalized.” (lines 593-594). These findings discussed in Discussion, but not mentioned in Conclusions.
Comment 4: “The effect of semi-dwarfing alleles could be negligible because of the imbalanced group sizes in our study” should be rephrased. The data does not show the effect is negligible; it shows the study was underpowered to detect an effect if one exists. It is recommended to soften claims about Rht effects and present this primarily as an observation on the allele frequency in this particular germplasm set, with the statistical limitation clearly stated.
Response 4: Thank you for pointing this out. We have rephrased the sentence: “The potential effect of semi-dwarfing alleles may not have been detected because of the imbalanced group sizes, which resulted in reduced statistical power in our study.”
Comment 5: The stark difference between 2022 (high disease pressure) and 2023 (low disease pressure) is a strength, clearly showing the context-dependency of passive resistance. However, the prediction accuracy was very low (28%) in 2022. The discussion should more deeply address why the model failed under high pressure. Was it simply because passive mechanisms are overwhelmed, or could it be that the relationship between traits and severity becomes non-linear or even reversed under extreme stress? This is a critical point for breeders who need to know the limits of passive resistance.
Response 5: We agree, it is an important finding. We discussed already extensively the different weather conditions. The different patterns in responses depending on disease pressure were noticed in previous studies. We mentioned: “Under high disease pressure, passive resistance can be overcome by Fusarium pathogens. The dependence of passive resistance on disease pressure was noticed by Mesterhazy (1995), who reported that under a strong artificial disease background, non-significant differences in plant resistance to Fusarium head blight (FHB) were observed among various plant height classes [18]. Buerstmayr and Buerstmayr (2022) also reported the minimal effect of plant height on FHB under greenhouse conditions [75]. In contrast to 2022, plant height showed a significant negative correlation with FHB severity in 2023, indicating that long-stature plants have a greater chance of avoiding initial FHB infection under low disease pressure. Similar conclusions have been drawn by other researchers [27,29–32,38,69,76].”. After that we have added a summarizing statement: “The findings of previous studies, supported by our results, suggest that passive resistance creates obstacles or barriers to fungal development. However, these physical barriers are not universal and have a threshold that can be overcome by pathogen if the disease pressure is sufficiently high.”. This is an interesting hypothesis that the genetic pattern of resistance and interaction between fungus and plant might change linearity or become reversed depending on disease pressure; however, in our opinion, it would be too speculative to make such a suggestion in our study.
Comment 6: Epsilon-squared values are presented, which is good for non-parametric data. However, for traits like “Days to heading” and “Days to flowering” in Table 3, the genotype effect is reported as negative (ε² = -0.07, -0.10). Epsilon-squared, as a measure of variance explained, should not be negative. This suggests a potential error in calculation or application of the formula for these specific cases. This must be checked and corrected.
Response 6: Thank you for pointing this out. The genotype effect was 0 for “Days to heading” and “Days to flowering”. We have updated the table.
Comment 7: Figure 1: The figure is conceptually good but visually very busy. Consider creating two separate panels for temperature/precipitation and genotype counts, or using a dual-axis plot with greater care to distinguish the data types.
Response 7: We agree that including too many elements/axes makes the graph more difficult to understand quickly. However, there is a trade-off; if additional graphs are made, it will be more difficult to compare weather conditions with flowering and heading dynamics. We have extended the captions to improve clarity. And we would prefer to keep a more complex, but more convenient graph that allows direct comparison of the different parameters.
Comment 8: The abstract states "332 spring wheat genotypes" but later in the aims (page 3) and results, "322" is sometimes used. Ensure consistency (332 is correct based on Methods).
Response 8: It was a misprint. We have corrected it.
Comment 9: The transition from general FHB background to the specific gap addressed by the study is good. Consider adding a sentence explicitly stating that most studies focus on individual traits or small combinations, whereas this study aims to evaluate a complex of eight traits simultaneously using advanced analytics.
Response 9: Thank you for your comment. The text has been updated: “However, in this study, we investigated a complex of eight morpho-phenological traits simultaneously and their association with overall FHB resistance using advanced analytics.”
Comment 10. Clarify the sowing date(s) for each year, as this is the reference point for "days to heading/flowering."
Response 10: The sowing dates have been added to the materials and methods.
Comment 11: Anther extrusion for 2025 (AE_2025) is included in a table focused on 2022 and 2023. Consider a separate note or table for the 2025 AE data to avoid confusion.
Response 11: We have modified Table 2. A separate row “AE (2025)” is added and a capture below table: “*AE measurements of 2025” was added to provide additional clarity.
Comment 12: The caption for Figure 3 mentions "across 3 years," but the plots are for 2022 and 2023. FHB severity data for 2025 is not mentioned. Correct the caption.
Response 12: The caption has been corrected.
Comment 13: The discussion is comprehensive but somewhat lengthy. It could be tightened by focusing more on synthesizing the novel findings of this study rather than extensively recounting literature that agrees or disagrees.
Response 13: We agree that the discussion may feel somewhat lengthy. The discussion conserning Rht genes were reduced. However, the further reduction may result in the loss of essential information.
Comment 14: When discussing contradictory findings in the literature (e.g., on awns, heading date), explicitly link back to how your results, especially from the RF model which accounts for interactions and context, might help reconcile these differences.
Response 14: Thank you for your comment. We analysed contradictory findings of awn length using PDP and ICE plots. It has been included in the discussion section: “PDP and ICE plots indicate that the group with long awns was associated with higher FHB resistance. This contradicts the general conception of passive resistance but can be explained by the distribution of awned and awnless genotypes. A strong imbalance was observed between the awned and awnless genotype groups. Approximately 9.8% of genotypes exhibited long and dense awns, whereas about 90% were awnletted and awnless types. Moreover, famous standards of FHB resistance, such as Sumai 3, Wangshuibai, and FHB-resistant lines N894037 and SHA3/CBRD from CIMMYT, fell into the smallest group with long awns. As a result, the increased resistance of this group may be caused by major genes (Fhb1, Fhb2, Fhb4, and Fhb5), which masked the true effects of awns in passive resistance.”
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThis article has been corrected.
Reviewer 2 Report
Comments and Suggestions for AuthorsPlease ensure that all citations follow the journal's formatting guidelines.
Reviewer 3 Report
Comments and Suggestions for AuthorsThe author has already revised the manuscript quite well.
