I have read the revision of the paper - it greatly improved, but I have some further suggestions that might improve the statistical presentation of results.

(1) I find it confusing at what stages the slow vs fast ramping regime was analysed in one analysis, and when these were analysed separately. From what I understand - the first analysis of knockdown times was done separately, and then at what stage/in which analyses the slow/fast selected lines were pooled? Does "Selection" fixed effect always mean slow vs fast, or does it mean selected vs control? It should be clearer and the naming of factors should leave no space for ambiguity here...

(2) Random effects - from what I see you had 1 random effect (line) - how was it tested in lme4 using LRT? This package does not allow for fitting a model without random effects, which would be needed to compare it with the one with the line effect. Also - you report only 1 result of LRT - was it the only analysis where random effect were used?

(3) I think analysing all static assay temperatures is superfluous (not to mention the multiple comparisons it generates, I think such results, if not corrected for false-positive discovery rate inflation (e.g., via p.adjust() in R), may be anticonservative). What I suggest is sticking to the knockdown time ~ temp relationship for each fly and line, and using a mixed effect ANCOVA to test hypotheses. In such analysis, the intercept differences would reflect changes in knockdown time, and changes in slope (interaction of temperature and selection regime) would reflect potential effects of selection on CT_max; such a model would also allow for random slopes (variation in t~temp slope variation across replicate lines).

Minor

L18-20 There's something wrong with the word order in this sentence

L111-112 What are the "confounding effects" here? This prediction is not clear at all.

L143 at a rate, not TO a rate (similar typo in several other places)

L234 glmer, not glm?

L346-347 In what species and/or study type?

L358 assay, not ASSAYED

I suggest one more read looking for other typos or leftovers from previous version.

I read the paper with great interest - it is an nice experiment that targets the artificial (co)evolution of traits associated with thermal sensitivity in the fruit fly. However, I have some comments and questions that may be valid for the interpretation of the results.

1) So you created 100 isofemale lines - but then their identiy is largely lost, where they maintained (i.e., 100 lines, then each split into 3 reps and 4 groups) throughout the experiment? It is confusing as it is very difficult to find ones way through the maze of N values, and how different sample sizes come together to describe the power of the study. Please provide: a flow-chart/image of your experiemnt, where at each key stage you specify the number of enclosures (bottles/cages/vials), number of lines and number of individual flies breeding/participating in assays.

2) Also on the question of lines and replicates: what where the results for random effects? Variance estimates etc - this information is entirely missing from the paper. Can you provide open data and code (I strongly suggest thi is available before next review)? Where there changes in variance components between selection regimes?

3) Presentation of results: please provide CIs for all relevant estimates (instead of SE, it emphasizes interval-based conclusions). Also - what are LTR_5 etc in the desication results section? Why are some presented as just LTR and some with a chi-squared statistic?

4) Is it possible that actually selection you imposed was also selecting for desication resistance (so dessication resistance would not evolve in correlation but under direct selection?) I mean - your selection procedure could target both individual's ability to resist thermal stress but also to survive dessication. Would that be a valid explanation?

5) Since it is a selection experiment: I miss basic estimates (in the founding population) of standing genetic variance in traits that were then studied as evolving under selection. Do you have such estimates? Would isofemale lines be inbred enough to provide such estimates? Would genetic variance in base population be enough to justify the observed response to directional selection?

6) What is the interpretation of the polygon are in Fig 2B? You refer to it as "area occupied in the thermal sensitivity landscape" - but is it just a measure of error in linear relationship between z and CT_max? Are CT_max and z expected to be thermally related? If so - this area measure would largely be a statistical byproduct of uncertain values...

The text needs some typo fixing and English editing.

Adaptive response to global warming, although species specific, generally depends on the population genetic variability and thermal stress intensity. The Author of this paper clearly emphasize the significance and background of research within that topic in introduction. Natural populations are exposed to multiple environmental stressors, and it is known that increased tolerance to one stressor can boost tolerance to another. The major controbution of the present study is that focus is on the effect of variable heat stress intensity on the correlated responses of resistance traits, such as the desiccation and starvation resistance in a Chilean population of D. subobscura. The experimental design is given clearly, with sufficient details, including descripiton of methods performed. Statistical analyses are appropriate and  I do not find any missing interpeatation in the results. All Tables and Figures are readable and clear.

The conclusions of this study are adequately supported by the results. The obtained results show the correlated response to thermal stress selection for the studied resistance traits in D.subobscura under the goven experimental design, which demonstrates that the evolutionary response to tolerate higher temperatures also confers the capacity to tolerate other stress such as desiccation and starvation. As Author correctly states, these correlated responses depended on the intensity of thermal selection and sex, which could limit the capacities to transfer these findings to natural scenarios. However, it does not downplay the the value of experimental evolutionary approach to explore and to understand the adaptive responses of natural populations to global warming.

The chosen species D. subobscura has been proven to be ideal to study „contemporary evolution“ as natural experiment, since it has adapted to New world from Palearctic and spreaded quickly by adapting to new enviroments. In that respect, my only remark (suggestion) for this paper is that discussion should take account the results of the genetic variabilty bacground and possible causes of evolutionary and cros evolution response obtained in this particular species. In D. subobacura, the inversion coadaptation studies related to thermal selection have been studied throughly in the light of climatic change. Thus, some of the results, relevant for this research I think should be commented. The Author says that further evidence is needed such as quantitative genetic or genome-wide analysis studies to elucidate the genetic basis of the cross-tolerance evolution in D. subobscura, but I think that they already exist and could shed some light to the results based on the hypothesis in this paper. This species has unique inversion polymorphism, and some of the gene arrangements within individual chromosomes have been proved as useful and informative markers for adaptaion under thermal stess.