Selection of Heat Tolerant Lettuce (Lactuca sativa L.) Cultivars Grown in Deep Water Culture and Their Marketability
Round 1
Reviewer 1 Report
line 105 "werer" ?
8 replicates in total seem quite low for statistical Analysis (line 153)
Why is there no maximum Rating (e. g. table 3)?
What is your (internal) control? Is there any?
Criteria (lines 214, 215) not based on standard criteria for lettuce crop quality?
Maybe conclusion a bit more detail with regard to data aquisition (tip burn, bolting, taste)
Author Response
We thank you for reviewing our paper. We understand the time considerations necessary and we very much appreciate you sharing your expertise with us to make our paper better. We have considered your review comments and have addressed them individually. Your original comments are in italics and our responses follow. We believe your comments have much improved our manuscript and we are very grateful.
1. line 105 "werer" ?
changed to “were”
2. 8 replicates in total seem quite low for statistical Analysis (line 153)
Thank you for the comment. We agree. We conducted two experimental runs in the greenhouse (we had several preliminary runs as well). Our results were very consistent, so we decided to essentially pool the data by putting experimental run in the random statement of our SAS model. So, all the data presented from the greenhouse experiment are means or medians of 16 replicates. We have indicated this in our table footnotes.
3. Why is there no maximum Rating (e. g. table 3)?
This is a very good question. In the raw data there are certainly maximum ratings, but we reported medians because the rating data are non-normally distributed by definition. Medians are not sensitive statistics and as such are resistant to move upwards or downwards. For a criterion to have a median rating of “5” almost every response out of 50 would have needed to be a “5”. So, there were several ratings of “5” for certain cultivars for marketability for example, but the median was calculated as “4” due to insensitivity of medians.
4. What is your (internal) control? Is there any?
We elected to not have an experiment-wise “control” in this experiment. We did this for two reasons. The first is that we knew we wanted to analyze growth and development data (size, biomass, bolting, tipburn, etc.) by lettuce groups to account for inherent size differences between groups (Romaine lettuces are generally larger than summer crisps, for example). We didn’t have an example of a standard from each group that could be used as an appropriate control. Second, we wanted to select heat tolerant cultivars without necessarily having to compare them directly to a control. The reason for this is closely related to the reasoning of why we grouped lettuces. Our opinion is that heat tolerance should be within an acceptable range and not based on relative performance to a control, especially in the “worst-case” scenario we grew our lettuce in.
If we had elected to have a control it would have been ‘Rex’. It is the standard hydroponic cultivar grown in our region, but again we didn’t think it was appropriate or necessary in this case. We did include ‘Rex’ in our experiment and were not surprised that it performed well in both the greenhouse experiment and the sensory evaluation.
5. Criteria (lines 214, 215) not based on standard criteria for lettuce crop quality?
This is a very good question. We adapted our tipburn rating scale from standard lettuce rating criteria from USDA-AMS, but we had to modify the scale slightly to account for all damage. Prior to this experiment we did not find a standard rating scale for bolting. We think this is probably due to the fact that any level of bolting is normally unacceptable. Since we were growing our lettuce in a worst-case scenario (the greenhouse was very hot) we felt confident that selecting the lettuces that we did accounted for the very best performers. We also needed to reduce the number of cultivars significantly to make our sensory evaluation manageable for the participants. We had previous experience of recruiting local chefs to taste 20 lettuce cultivars and we quickly realized that was far too many.
6. Maybe conclusion a bit more detail with regard to data aquisition (tip burn, bolting, taste)
Thank you for the suggestion. We have updated our entire manuscript including the conclusion section. Hopefully we have been much clearer in our objectives and conclusions.
Reviewer 2 Report
Although this article contains useful information, adaptable area is very limited as authors pointed out. Considering publication in an international journal, some more general discussions to reach common facts and conclusions are required.
Title
It should be considered that the growth rate or growth characteristics differ depending on the variety. Other words, if you have data on lettuce grown at optimal temperatures, you can evaluate "heat tolerance" (eg, how much growth is not suppressed at high temperatures) by comparing data on lettuce grown at high temperatures is there. This title should be "growth characters under high temperature, not "heat tolerance".
The relationship between the heat tolerance and sensory evaluation is unclear. It should be discussed enough about quality under high temperature environment and sensory evaluation.
Introduction is redundant. Use past papers to explain common facts.
Rather than the needs of limited local farmers, we should set goals that can be shared globally, such as the point of view of selection of varieties and improvement of cultivation methods.
Materials and methods
Is the experiment of L109 "spring and early summer" different from the experiment described in this paper? If they are just preliminary experiments, there is no need to write.
L113-115 Is the concentration of fertilizer a simple base or an oxide base?
L158 I don’t catch the meaning of “RANDOM” and “the GROUP”
L161 T-test? (F-test is adaptable if number of treatments is 3 or more).
L184 Please explain the criterions of sensory evaluation. (I don’t distinguish “texture” and “crispness”)
Table 3. if there are only 2 cultivars, we never use alphabets to indicate significant difference.
Discussion part is very poor. Almost of descriptions are cited from previous articles. Some sentences can be introduced briefly, but here, you have to analyze well your own research results. As I mentioned above, discussion about relationship between heat tolerant and sensory evaluation should be considered well. What about using principal component analysis using both data from cultivation test and sensory test?
Author Response
We thank you for reviewing our paper. We understand the time considerations necessary and we very much appreciate you sharing your expertise with us to make our paper better. We have considered your review comments and have addressed them individually. Your original comments are in italics and our responses follow. We believe your comments have greatly improved our manuscript and we are thankful.
1. Although this article contains useful information, adaptable area is very limited as authors pointed out. Considering publication in an international journal, some more general discussions to reach common facts and conclusions are required.
We appreciate the comment and agree. We have eliminated most mentions of any specific geographical regions in the article. We did keep one reference (L36) to our region in the introduction but re-worded the sentence so that the statement following wouldn’t be perceived as being limited to our particular region. We have amended our manuscript thoroughly.
Title
2. It should be considered that the growth rate or growth characteristics differ depending on the variety. Other words, if you have data on lettuce grown at optimal temperatures, you can evaluate "heat tolerance" (eg, how much growth is not suppressed at high temperatures) by comparing data on lettuce grown at high temperatures is there. This title should be "growth characters under high temperature, not "heat tolerance".
We appreciate the comment and agree in part. We really intended “heat tolerance” to refer to the susceptibility of certain lettuce cultivars to bolting and tipburn. We tested that “tolerance” in a high temperature greenhouse. Our hope was that if a lettuce cultivar performed well in a worst-case scenario we could reasonably assume that it would perform well in a more optimal temperature environment. It is true that we did not grow these lettuces at optimal temperatures in order to make comparisons at supra-optimal temperatures.
We realize that temperature certainly affects growth. However, we would like to distinguish between growth and development. Temperature is the primary environmental factor which drives development while growth is environmentally driven by light and CO2 (both of which were the same for all plants in this experiment). Development is what we are essentially measuring when we record bolting response. We also think that “heat tolerance” is a commonly-used term when considering bolting, in particular. This is reflected well in seed company marketing strategies. Tipburn is not a developmental response, but is certainly exacerbated by supra-optimal temperatures. It is also not a growth characteristic. We provide size index and fresh weight data to give a relative comparison of growth within groups. We compare our growth data with those of other authors because those same data were available for the same cultivars and allowed for an interesting discussion point about heat tolerance/bolting rate possibly being affected by growth rate and/or growing method (soil vs. hydroponics).
Temperature, light, and CO2 were the same for all cultivars in the trial. We were not primarily concerned with growth rate, although it is interesting, but were primarily concerned with development (bolting) and other common damage (tipburn) being triggered (genetically) at supra optimal temperatures. Since plant genetics was the only difference between experimental units, we believe “heat tolerance”, or resistance to bolting and tipburn, is an appropriate term. We have attempted to clarify this in our introduction.
3. The relationship between the heat tolerance and sensory evaluation is unclear. It should be discussed enough about quality under high temperature environment and sensory evaluation.
Thank you for this comment. Our attempt in this paper is not to relate heat tolerance to sensory evaluation. We are attempting to use an evaluation of heat tolerance in a worst-case scenario to select lettuce cultivars that perform relatively well in such circumstances and to then evaluate the sensory characteristics of those cultivars. We have attempted to clarify this throughout our manuscript. It may be the case that heat tolerance is correlated with some sensory attributes. That would likely be a very valuable research question to test, but we did not test that question in the present study. In order to do so we would have included all the original cultivars in the sensory evaluation so that we could correlate sensory ratings with bolting ratings.
Instead, what we did was a step-wise experiment in which we eliminated some lettuces from sensory evaluation because they performed poorly in the greenhouse and eliminated others because consumers didn’t like them for one reason or the other. We were able to recommend seven of eighteen original cultivars which are marketed as “heat tolerant” or “slow bolting” on seed company websites to growers that are looking for the most heat tolerant lettuces to grow. We also have interesting findings in our sensory evaluation which we discuss in more detail thanks to your recommendation.
4. Introduction is redundant. Use past papers to explain common facts.
We agree and have amended our Introduction section. It is now significantly shorter and, we hope, much clearer. Thank you for the very helpful suggestion.
5. Rather than the needs of limited local farmers, we should set goals that can be shared globally, such as the point of view of selection of varieties and improvement of cultivation methods.
Thank you for the comment. We agree that the original language mentioning our specific geographical region was too specific. We have removed those references.
We have included in the manuscript a more well-rounded discussion of how these methods are applicable in many different scenarios globally. We think this makes our manuscript much better and appreciate your direction.
Materials and methods
6. Is the experiment of L109 "spring and early summer" different from the experiment described in this paper? If they are just preliminary experiments, there is no need to write.
Sentence removed.
7. L113-115 Is the concentration of fertilizer a simple base or an oxide base?
Concentrations listed in L 113 are actual concentrations of elemental N, P, K, Ca, and Mg in solution. Fertilizer analyses in L 102-103 are in simple base terms. For example, the label of the fertilizer described in L 102 is 8-15-36 (N-P2O5-K2O). Here, we have simplified the analysis in the N-P-K format.
8. L158 I don’t catch the meaning of “RANDOM” and “the GROUP”
Thank you for bringing our attention to this. We agree that the language here is not as clear as it should be. We have reviewed all the data analysis and have decided to remove the sentence mentioning the GROUP option because we did not actually use it in this particular case. We often use that language to assure readers that heterogeneity of variance was not ignored. In PROC GLIMMIX the GROUP option within a RANDOM statement can be used to correct for heterogeneity of variance (see a more detailed definition below). Our decision points for that are if residual plots or a significant covariance test indicate heterogeneity of variance. In the vast majority of cases, the analysis is not affected at all but we want to assure readers and reviewers that we are aware of the assumption of homogeneity of variance in generalized linear mixed effect models.
From the SAS Help Guide the GROUP option “identifies groups by which to vary the covariance parameters. Each new level of the grouping effect produces a new set of covariance parameters. Continuous variables and computed variables are permitted as group effects. PROC GLIMMIX does not sort by the values of the continuous variable; rather, it considers the data to be from a new group whenever the value of the continuous variable changes from the previous observation. Using a continuous variable decreases execution time for models with a large number of groups and also prevents the production of a large "Class Levels Information" table.”
9. L161 T-test? (F-test is adaptable if number of treatments is 3 or more).
Thank you for catching this. This is correct and we have changed the language in the article.
10. L184 Please explain the criterions of sensory evaluation. (I don’t distinguish “texture” and “crispness”)
This is a very good point. In the discussion section we originally explained why we included “overall texture” in addition to “crispness” but we have now included that explanation and expounded upon it a bit in the revised materials and methods section. Please see L 173-179
11. Table 3. if there are only 2 cultivars, we never use alphabets to indicate significant difference.
Thank you for pointing this out. We agree. We originally used letters to indicate significance in the groups with only 2 cultivars to keep continuity with the rest of the table, but per your suggestion we have decided to remove the lettering and add an explanation in the table footnotes.
12. Discussion part is very poor. Almost of descriptions are cited from previous articles. Some sentences can be introduced briefly, but here, you have to analyze well your own research results. As I mentioned above, discussion about relationship between heat tolerant and sensory evaluation should be considered well. What about using principal component analysis using both data from cultivation test and sensory test?
Thank you for the helpful suggestions. We have made significant changes to the discussion section and have paid more attention to discussion of our own results. We have not directly compared heat tolerance and sensory evaluation because our data do not allow for a true comparison. However, we think the discussion section is much improved and we hope we have been clearer in our objectives and analysis.
Round 2
Reviewer 2 Report
Title
Please consider the title again. In my opinion, “heat tolerance” should be scientifically used based on physiological evidence although seed company use this word easily. Exactly, this article didn’t study “heat tolerance” but carried out “cultivar selection under high temperature condition”, or “selection of heat tolerant cultivar”
Even if you don’t agree my opinion, I don’t understand the relationship between heat tolerance and sensory evaluation.
For example, how about “Selection of heat tolerant lettuce (Lactuca sativa L.) cultivars and their marketability (or palatability, acceptability, etc.).”?
Worldwide adaptability than regional benefit
Authors stress “worst”, however, there are lots of much more “worst” environment in the world. Generally,
Expression of this article seems very regional matter. Please consider to revise details as international article.
Statistical analysis
T-test is available for pairwise comparison and add such as “*” or “**”, for indicating significance at 5% or 1% level, respectively.
If obtained data is “0”, it can no longer be involved in statistical analysis, because the variance is 0.
e.g. [Adriana: 0.0, Rex: 0.0, Skyphos: 0.0, significance: n.s.] is statistically wrong.
Or employ nonparametric analysis.
Did you use ANOVA for sensory evaluation (table 4)? Exactly, data by rating should not employ parametric analysis because mathematical distances between each value may not be even. i.e. is there any mathematical evidence that average of “2.0” and “4.0” becomes “3.0”? I feel there is contradiction between body and table 4 footnote.
L310 interestingly
L312 surprisingly
Subjective expression is not suitable for original article. Authors should discuss at objective viewpoint.
Author Response
We thank you again for reviewing our paper. We have considered your comments and addressed them individually below. We very much hope the following responses will help clear up any confusion. Your original comments are in italics and our responses follow. Thank you again for your helpful review!
1. Please consider the title again. In my opinion, “heat tolerance” should be scientifically used based on physiological evidence although seed company use this word easily. Exactly, this article didn’t study “heat tolerance” but carried out “cultivar selection under high temperature condition”, or “selection of heat tolerant cultivar”
Even if you don’t agree my opinion, I don’t understand the relationship between heat tolerance and sensory evaluation.
For example, how about “Selection of heat tolerant lettuce (Lactuca sativa L.) cultivars and their marketability (or palatability, acceptability, etc.).”?
We have decided to change our title per your suggestion.
2. Worldwide adaptability than regional benefit
Authors stress “worst”, however, there are lots of much more “worst” environment in the world. Generally,
Expression of this article seems very regional matter. Please consider to revise details as international article.
We understand your concern and have removed regional references in our discussion. We only include the location of the experiment so readers will understand where this experiment took place. We believe it is then up to the reader to decide whether or not the results or methodologies apply to their respective location. We have chosen this particular international journal because it is a poplar, widely read, open access journal which means that not only can international audiences access this information, but also interested parties in our region can easily access this journal.
Statistical analysis
3. T-test is available for pairwise comparison and add such as “*” or “**”, for indicating significance at 5% or 1% level, respectively.
All significances are reported at the 5% level.
4. If obtained data is “0”, it can no longer be involved in statistical analysis, because the variance is 0.
e.g. [Adriana: 0.0, Rex: 0.0, Skyphos: 0.0, significance: n.s.] is statistically wrong.
Or employ nonparametric analysis.
We used ESTIMATE statements in PROC GLIMMIX in SAS to determine paired differences among treatments (lettuce cultivars) with the simulate method and we assigned lettering manually. It is important to realize that the Estimate statement does not compare means or medians for the treatments, but compares calculated estimates. This is a valid method for modeling ordinal responses according to Schabenberger and Pierce (2002)*. Data we reported in Table 3 for bolting and tipburn ratings and in Table 4 for sensory ratings are medians. Medians are less sensitive than means. Looking at our raw data we can see that each cultivar had ratings of at least 1 in several cases, but since the median is an insensitive statistic it did not change from 0. If we had reported means they would have been greater than 0.
Further, we did not base our selection of cultivars for the sensory evaluation on statistical analyses. We provided the analyses to spur discussion. We understand the limitations of statistical analyses and as such selected a ratings “ceiling” of median bolting and tipburn ratings ≤ 2.5 and ≤ 1, respectively (L154, L208) as criteria for their inclusion in the sensory evaluation experiment.
*Citations:
Schabenberger, O. and F.J. Pierce. 2002. Contemporary statistical models for the plant and soil sciences. Taylor and Francis, Boca Raton, Florida.
5. Did you use ANOVA for sensory evaluation (table 4)? Exactly, data by rating should not employ parametric analysis because mathematical distances between each value may not be even. i.e. is there any mathematical evidence that average of “2.0” and “4.0” becomes “3.0”? I feel there is contradiction between body and table 4 footnote.
All rating scale responses (such as Bolting Rating, Tipburn Rating, Crispness, Overall Texture, Bitterness, Overall Flavor and Marketability) were analyzed within an ANOVA framework using the experimental and treatment design as described in the materials and methods paragraph describing the statistical analysis. However, the multinomial probability distribution was used instead of the normal probability distribution. This is important because the multinomial probability distribution assumes the response categories are discrete and therefore the distance between categories are unknown. A similar analysis is in Schabenberger and Pierce (2002)* except their analysis used PROC GENMOD (pg. 355). We adapted PROC GLIMMIX to perform the same analysis. Similar to Schabenberger and Pierce (2002)*, we used ESTIMATE statements to determine paired differences among treatments (lettuce cultivars) with the simulate method and we assigned lettering manually. It is important to realize that the Estimate statement does not compare means for the treatments, but compares calculated estimates. Schabenberger and Pierce (2002)* used individual CONTRAST statements for each paired comparison.
We have changed our footnote to more accurately describe our data analysis method.
Original footnote:
7Values in column, within each group, sharing a letter were nonsignficantly different according to Tukey's Honest Significance Difference Test at α=0.05 for Size Index and Head Fresh Weight data and the simulated method at α=0.05 for Bolting and Tipburn rating data.
New footnote:,
7Values in column, within each group, sharing a letter were nonsignficantly different according to Tukey's Honest Significance Difference Test at α=0.05 for Size Index and Head Fresh Weight data and Estimate statements within the simulate method at α=0.05 for Bolting and Tipburn rating data.
We have pasted our SAS code for part of this analysis below.
Proc Glimmix data=sensory;
Title "Crispness";
Class Cultivar;
Model Crispness (descending) = Cultivar / dist=multi link=clogit;
Estimate "144 parameter" intercept 1 Cultivar 1 0 0 0 0 0 0 0 0,
"183 parameter" intercept 1 Cultivar 0 1 0 0 0 0 0 0 0,
"563 parameter" intercept 1 Cultivar 0 0 1 0 0 0 0 0 0,
"634 parameter" intercept 1 Cultivar 0 0 0 1 0 0 0 0 0,
"727 parameter" intercept 1 Cultivar 0 0 0 0 1 0 0 0 0,
"807 parameter" intercept 1 Cultivar 0 0 0 0 0 1 0 0 0,
"833 parameter" intercept 1 Cultivar 0 0 0 0 0 0 1 0 0,
"877 parameter" intercept 1 Cultivar 0 0 0 0 0 0 0 1 0,
"992 parameter" intercept 1 Cultivar 0 0 0 0 0 0 0 0 1;
Estimate "144 vs 183" Cultivar 1 -1 0 0 0 0 0 0 0,
"144 vs 563" Cultivar 1 0 -1 0 0 0 0 0 0,
"144 vs 634" Cultivar 1 0 0 -1 0 0 0 0 0,
"144 vs 727" Cultivar 1 0 0 0 -1 0 0 0 0,
"144 vs 807" Cultivar 1 0 0 0 0 -1 0 0 0,
"144 vs 833" Cultivar 1 0 0 0 0 0 -1 0 0,
"144 vs 877" Cultivar 1 0 0 0 0 0 0 -1 0,
"144 vs 992" Cultivar 1 0 0 0 0 0 0 -1 0,
"183 vs 563" Cultivar 0 1 -1 0 0 0 0 0 0,
"183 vs 634" Cultivar 0 1 0 -1 0 0 0 0 0,
"183 vs 727" Cultivar 0 1 0 0 -1 0 0 0 0,
"183 vs 807" Cultivar 0 1 0 0 0 -1 0 0 0,
"183 vs 833" Cultivar 0 1 0 0 0 0 -1 0 0,
"183 vs 877" Cultivar 0 1 0 0 0 0 0 -1 0,
"183 vs 992" Cultivar 0 1 0 0 0 0 0 0 -1,
"563 vs 634" Cultivar 0 0 1 -1 0 0 0 0 0,
"563 vs 727" Cultivar 0 0 1 0 -1 0 0 0 0,
"563 vs 807" Cultivar 0 0 1 0 0 -1 0 0 0,
"563 vs 833" Cultivar 0 0 1 0 0 0 -1 0 0,
"563 vs 877" Cultivar 0 0 1 0 0 0 0 -1 0,
"563 vs 992" Cultivar 0 0 1 0 0 0 0 0 -1,
"634 vs 727" Cultivar 0 0 0 1 -1 0 0 0 0,
"634 vs 807" Cultivar 0 0 0 1 0 -1 0 0 0,
"634 vs 833" Cultivar 0 0 0 1 0 0 -1 0 0,
"634 vs 877" Cultivar 0 0 0 1 0 0 0 -1 0,
"634 vs 992" Cultivar 0 0 0 1 0 0 0 0 -1,
"727 vs 807" Cultivar 0 0 0 0 1 -1 0 0 0,
"727 vs 833" Cultivar 0 0 0 0 1 0 -1 0 0,
"727 vs 877" Cultivar 0 0 0 0 1 0 0 -1 0,
"727 vs 992" Cultivar 0 0 0 0 1 0 0 0 -1,
"807 vs 833" Cultivar 0 0 0 0 0 1 -1 0 0,
"807 vs 877" Cultivar 0 0 0 0 0 1 0 -1 0,
"807 vs 992" Cultivar 0 0 0 0 0 1 0 0 -1,
"833 vs 877" Cultivar 0 0 0 0 0 0 1 -1 0,
"833 vs 992" Cultivar 0 0 0 0 0 0 1 0 -1,
"877 vs 992" Cultivar 0 0 0 0 0 0 0 1 -1
/ adjust=simulate(seed=123456);
run;
Proc Means data=sensory median;
Class Cultivar;
Var Crispness;
run;
*Citations:
Schabenberger, O. and F.J. Pierce. 2002. Contemporary statistical models for the plant and soil sciences. Taylor and Francis, Boca Raton, Florida.
6. L310 interestingly
L312 surprisingly
Subjective expression is not suitable for original article. Authors should discuss at objective viewpoint.
We agree and have removed the subjective expressions.
