This manuscript presents an important and timely assessment of diversity, equity, and inclusion (DEI) in the cloud physics community, combining a metadata analysis of 6,987 peer-reviewed papers (1970–2020) with a community survey of 198 respondents. The topic is highly relevant, and the effort to integrate quantitative and qualitative perspectives is commendable. However, several conceptual and structural issues limit the manuscript’s clarity and impact in its current form. In particular, the terminology used to describe researchers from the Global South, the framing of gender analysis, and the interpretation of small-sample survey results require substantial revision. Addressing these concerns would considerably strengthen the analytical rigor, inclusivity, and global sensitivity of the paper.

Major Comments

1. Terminology – “Tropical Scientists”: In my opinion, the use of the term “tropical scientists” is highly problematic. I find it both unnecessary and ill-considered that the authors have chosen to introduce a new label that lacks conceptual grounding and adds no clarity to the discussion. As someone working within this geographic zone, I find this terminology technically inaccurate and, frankly, counterproductive—it conflates latitude with social and institutional realities, and risks exoticizing researchers from the Global South. In my view, this framing reflects a misunderstanding of the issue it seeks to address. Established terms such as Global North/South, high-, middle-, or low-income countries (as defined by the UN or World Bank), or developed/developing countries—though imperfect—are far more appropriate, as they rest on recognized metrics and are widely understood in academic and policy contexts. I strongly recommend removing the term entirely and clarifying the distinction between tropical study sites and researchers from the Global South, which are not interchangeable categories.
2. Use of “Minorities” vs “Marginalized”: The manuscript frequently uses the term “minorities.” Given the global context of the study, where numerical minorities may differ by region, the term “marginalized” would be more appropriate and inclusive. It emphasizes systemic disadvantage rather than numerical representation. This change would also align better with the ethical framing of equity in the paper’s conclusion.
3. Gender Analysis and Non-Binary Representation: The gender analysis appears to be strictly binary (men/women). While this may have been constrained by data availability (e.g., use of genderize.io), the discussion section should at least acknowledge and briefly discuss the exclusion of trans and non-binary individuals. For instance, the authors could note the limitations of algorithmic gender inference, and mention issues such as misgendering, the use of dead names in publication records, and the broader underrepresentation of trans scientists in the geosciences. This would strengthen the inclusivity and awareness reflected in the DEI framing.
4. Survey Interpretation and Sample Size Limitations: The metadata analysis is the stronger and more robust component of the paper. However, the survey-based results (N=198) present significant limitations when further divided by gender, country, and ethnicity. When broken down into subcategories, the sample size per group, in my opinion, becomes too small for reliable interpretation or for drawing broad conclusions about regional or demographic patterns. I suggest restructuring the Results section so that the metadata analysis comes first—forming the backbone of the findings—followed by the survey results as complementary qualitative insight. The discussion could then focus on how these two strands inform each other, rather than treating both as equally representative.

In this study, the authors present the current state and over time development of Diversity, Equity, and Inclusion (DEI) in the cloud physics community by considering two different data sets: metadata of 6,987 published papers between 1970 and 2020, and the survey responses from 198 participants. This research is significant for the advancement of the geosciences as it helps identify areas that need improvement. However, there are some issues with the framing of certain aspects, including the terminology used. Therefore, my general comments align with the Rev1.

General comments:

• Underrepresented instead of the word minority or marginalised: Minority means that the smaller number or part. As Rev1 mentioned, minorities differ from region to regionIt is important to note that labelling the global south as a minority is a Western-centric viewpoint. The groups considered in this context are not smaller in number; they are underrepresented, overlooked, or lacking visibility. Additionally, "marginalised" implies insignificance or a peripheral status, which also doesn’t adequately convey the issue at hand.
• Terminology tropical scientists/countries: The term "tropical" carries connotations that could present an "exotic" portrayal, which is not appropriate here. I completely agree with Rev1. Additionally, framing a comparison of “tropical countries” with so so-called global north is a false dichotomy. One is defined by economic and historical factors under neo-colonial structures, while the other is geographical.
• The gender analysis presented is binary and fails to include non-binary and transgender individuals. This omission is not addressed in the paper. Although the potential for misgendering due to the algorithm is acknowledged, it still contains uncategorized (gender-blind) names. Furthermore, such gender-inferencing algorithms often struggle with the accuracy of non-Western names.

I will try to elaborate on what I mentioned in the general comments by giving specific examples from the first couple of pages, which are valid throughout the entirety of the paper.

line 46: ".... minority representation..." This gap exists for any underrepresented group, not just minorities. You might not be a minority, but underrepresented, you might be a minority if you are looking from a Western lens. We are great in numbers, but our work or existence is not represented or acknowledged.

line 65: same issue here with the use of minorities

line 70: Women are underrepresented but larger in number, exactly like the scientists from the so-called global south. They are underrepresented, and there is a lack of acknowledgement. 

line 81: In the cited paper, this is mentioned as "USA, UK, ... However, these are not necessarily the most collaborative countries, if measured by their proportion of collaborative output... The analysis provided evidence that countries rated high in terms of scientific development were more likely to collaborate." This does not necessarily mean low income with low scientific development. 

line 86&96: Egypt and South Africa are not situated in the tropics but the subtropics. When comparing these regions with the global north, framing it as a dichotomy could be misleading, aka a false dichotomy. Additionally, the term "tropical" carries connotations that could present an "exotic" portrayal, which is not appropriate here.

line 189: Latitude range includes Australia, which is considered to be a "global north" country. If the intention is to analyse underrepresentation, this could lead to misrepresentation in the analysis. Although the countries are highlighted later in the table, it would be beneficial to clarify this point earlier on.

This paper presents an analysis into the current climate and feelings of belonging within the specific geoscience community of cloud physics. It utilizes two different approaches to address this question: a meta-analysis of peer-reviewed publications and an online survey. The question addressed in the paper is a unique contribution to the field, however there are several significant issues that need to be revised and addressed before publication. In addition to the previous two referee comments above, I will add the following:

1. There has been a surplus of more recent work that highlights the their point in lines 45-47 (Although various programs, policies, and inclusive environments are being developed to address issues such as gender inequality in academia, many gaps in other minority representation still remain (e.g., Hispanic, black, and indigenous communities; Bernard and Cooperdock, 2018; NCSES, 2019; Gewin, 48 2020; Odekunle, 2020). Please include an updated literature review from the last 5 years in this.
2. I agree with the previous two referees that “marginalized” is a more appropriate term than minority. Based on the descriptors provided, another alternative would be to say “historically excluded” groups. This is a better descriptor that is commonly used and recognizes the fact that these groups have been intentionally excluded by the system, as opposed to just individual actions.
3. In line 55 – able-bodied individuals are also significantly more represented in all fields than in the general population. There is literature on this as well and should be included in the literature review.
4. Lines 58-59 (The lack of diversity and inclusion across academic communities manifests itself in systematic bias of under-represented minorities). I disagree with how this comment is written – the system was intentionally designed to prevent lack of diversity and inclusion from the very beginning. The system is the problem, which further exacerbates the lack of diversity and inclusion we see today.
5. I think it would also be beneficial to expand on the background of cloud physics as a research field. It seems that your assumption is that everyone in this field trained in geosciences, or is there a combination of physics, chemistry, geography, etc? Each of these fields has presented its own variations in why they have issues related to DEI and should be included as a background variable when considering the climate of cloud physics.
6. Your response rates are not clearly defined and need to be included. How many individuals were included in the ICCP listserv and the other announcement sent? How many participants actually registered for the 2021 meeting? Comparing this number to other membership numbers is not sufficient – especially when the data is anonymous and you cannot guarantee that individuals did not take the survey more than one time. I don’t agree with the statement that it represents 30-47% of the community as a whole – you should only report on response rates from who the survey was disseminated to.
7. Lines 137-139: you say the number of respondents who were Indian was higher – but how do you know that if the data is anonymous? Please clarify.
8. Lines 206-208: What are the actual stats presented by Dutt, 2020? The ratio of white/non-white seems higher than what the ratio is in the US.
9. Figure 1: The color purple is used to signify two different things – please use different colors for the “total” versus “other” gender category.
10. The paper published by Marin-Spiotta et al (2023) in Earth’s Future provides survey results on the workplace climate in the geosciences. You should include this in your discussion/analysis as a comparison since their survey results include over 2000 survey responses.
11. Did you ask if survey respondents had experienced bullying or harassment within a specified time frame or ever? This should be further clarified and justified on either approach.
12. Line 324 – the leaky pipeline analogy is problematic. Berhe et al (2024) describes this in their Nature Geoscience paper and recommends a better analogy.
13. Line 335-336. I disagree that you have the numbers to state that you are representing the cloud physicists globally, based on my previous comment about response rates. This should be removed.
14. Given the broad demographics of who this survey was disseminated to, I think you should expand on potential limitations of your findings on feeling included. Your question was simply “Do you feel included in the cloud physics community?” This is really vague and could be interpreted in so many different ways – I am less inclined to say that this is a valuable finding with this approach. What do you mean when you say included? Just in the workplace, in professional societies, when in school? This is not well defined.
15. Why did you only ask about working with women in your survey? This is not inclusive to the entire gender spectrum.
16. What does “sensitive to gender issues” mean? Again as above – this is so vague and could be taken to mean so many things. I don’t agree that this should be presented in this dataset, it’s not a well designed survey question.
17. Question 40 in your survey – were you only talking about race/ethnicity here? Or was this applicable to any and all identities? Not clearly defined, which is problematic.
18. You inherently introduced a lot of bias in your survey by using the term “tropical” to describe an entire subset of people. This is problematic.
19. You say that the majority of people said no to question 67. Why did you not include an additional clarifier on that to explain their perspective? There could be many reasons why participants selected no, I don’t think it’s appropriate to speculate on that in your paper.

This review is both significant and timely. However, it demonstrates the tendency within academia to use archaic language and framing when analyzing geographic, racial, gender, and sexuality DEI issues globally. To address these, significant revisions are needed before publication. Specifically, the manuscript will be more impactful if the authors address the implicit, inherited, and adopted racism and sexism entrenched in the sciences.

Gender Methodology Issues

Lines 182-186: "It is important to note that genderize.io assigns the gender based on the person's first name. Although this could be a good approximation, names cannot always be used to infer the gender (e.g., Sebo, 2021; Lockhart et al., 2023; Marty et al., 2023). Therefore, the results presented below must be taken with caution and considering the aforementioned bias."

Issue: This perpetuates binary gender assumptions and doesn't acknowledge how this methodology erases non-binary identities and reinforces cisgender normativity. The authors should explicitly state this limitation.

Problematic Categorizations

Lines 150-153: "Therefore, in further analysis, we classify the Caucasian, White or European categories as 'White', while the other ethnicities are classified as 'non-white' with the exception of participants who preferred not to identify their ethnicity."

Issue: This binary centers whiteness as the default and groups all other identities into a single "other" category, which obscures important differences in experiences among diverse racial/ethnic groups. This approach reinforces white supremacy frameworks.

Minority Definition

Lines 50-54: "The word 'minority' denotes negatively stigmatized, ostracized, oppressed, and outcast individuals or groups (Blanz et al., 1995). In the context of DEI, it refers to a numerically smaller population of a group within a larger population based on characteristics such as gender, race, religion, ethnicity, disability, sexual orientation, or other attributes. In the present study, 'minority' is used to describe an underrepresented or disadvantaged group, unless stated otherwise."

Issue: While acknowledging negative connotations, the definition still focuses primarily on numerical representation rather than power dynamics and systemic oppression. Missing discussion of how majority groups maintain power regardless of numbers.

Global South Terminology

Lines 66-69: "Systemic bias, under-representation of minorities, and a lack of recognition of researchers in and from developing countries (the so-called 'Global South') are reflected in the limited scientific collaborations between research groups from developed (the so-called 'Global North') and developing countries (Gewin, 2023)."

Issue: Uses "so-called" but doesn't interrogate the problematic colonial legacy embedded in these geographic/economic binaries or how they perpetuate deficit narratives.

Parachute Science

Lines 393-408: The discussion of parachute science identifies the practice but doesn't explicitly name it as a continuation of colonial extraction practices or discuss how implicit biases about expertise and knowledge ownership enable these dynamics.

Specific concern at lines 393-397: "Our results suggest that the so-called helicopter or parachute science... is unfortunately common in the cloud physics community; however, a deeper analysis is required to confirm this."

Issue: Needs explicit statement about colonial legacies and power imbalances, not just a neutral "deeper analysis required."

Intersectionality

Lines 284-286: "A higher fraction of non-White women reported experiencing bullying and harassment, which aligns with the concept of intersectionality, where overlapping social identities and positions can increase the likelihood of discrimination (Crenshaw, 1991). However, the sample size prevents statistical verification of this pattern."

Issue: Intersectionality is mentioned but not systematically applied throughout the analysis. The entire study design should have incorporated intersectional analysis from the beginning, not as an afterthought.

Harassment Analysis

Lines 276-278: "A higher fraction of women, 44% compared to 27% of men reported having experienced harassment, a difference that was statistically significant (χ²(1) = 5.70, p = 0.017, Cramér's V = 0.18), although the size of the gender difference is modest."

Issue: Doesn't explicitly discuss how sexism and misogyny create hostile environments or how power differentials enable harassment. Also minimizes the finding by calling a 17 percentage point difference "modest."

Recommendations for Addressing These Issues:

1. Add a limitations section explicitly discussing how binary gender categorizations and racial/ethnic groupings reflect and potentially perpetuate systemic biases
2. Incorporate intersectional analysis throughout, not just in one paragraph
3. Explicitly name colonialism when discussing parachute science and Global North/South dynamics
4. Acknowledge how the research team's own positionality may have shaped survey design and interpretation
5. Discuss implicit bias training as insufficient without structural change

The dual-method approach (metadata analysis and survey) is valuable. However, in my review of procedures and statistical analyses, I've identified several significant methodological and statistical problems:

1. Gender Inference Methodology (Critical Issue)

Lines 180-186: "Each co-author's gender was evaluated employing the genderize.io web server (Demografix ApS, 2022) and included in the analysis if the certainty estimate provided by the tool was larger than 75%."

Problems:

• The 75% threshold is arbitrary with no justification provided
• No discussion of error rates or validation
• Missing data from names below 75% certainty could introduce systematic bias
• Binary gender assumption excludes non-binary individuals entirely
• Cultural name variations may have different accuracy rates, potentially introducing geographic/ethnic bias

Lines 182-186 acknowledge limitations but don't explain how this affects the validity of conclusions.

1. Survey Sampling and Representativeness Issues

Lines 132-143: "The survey was completed by 231 respondents, but the sample size used in the analysis here decreased to 198 after data quality assurance in which participants who did not identify with either the ice nucleation field and/or cloud physics field were removed."

Problems:

• 14% exclusion rate (33 participants) without detailed explanation of exclusion criteria
• No analysis of whether excluded participants differed systematically from included ones
• Potential selection bias not addressed

Lines 136-143: Claims representativeness based on conference attendance but:

• Conference attendees =/= field practitioners (excludes industry, non-conference-attending researchers)
• Three conferences from different years (2012, 2016, 2021) compared to 2021 survey—demographic shifts not accounted for
• "30-47% of the cloud physics membership" is a wide range suggesting uncertainty about denominator

1. Acknowledged Overrepresentation Not Corrected

Lines 142-144: "Although the ICCP membership is representative of the international cloud physics community, there could be some unintended biases in the survey data coming from the overrepresentation of respondents born in India, Germany, and US (above 15% each) as well the overrepresentation of respondents younger than 40 years old (64%)."

Problems:

• Authors identify bias but don't apply any weighting or stratification to correct for it
• No sensitivity analyses to test how overrepresentation affects conclusions
• Geographic bias particularly problematic for study examining Global North/South dynamics

1. Statistical Power Issues for Subgroup Analyses

Lines 284-287: "A higher fraction of non-White women reported experiencing bullying and harassment, which aligns with the concept of intersectionality, where overlapping social identities and positions can increase the likelihood of discrimination (Crenshaw, 1991). However, the sample size prevents statistical verification of this pattern."

Problems:

• Intersectional analyses attempted despite insufficient power
• No pre-registration or power calculations presented
• Multiple comparisons conducted throughout without correction for multiple testing
• Small cell sizes (e.g., 10 respondents with disabilities, line 269) make percentages misleading

1. Inconsistent Statistical Testing

Problem: The study applies statistical tests inconsistently:

• Sometimes reports chi-square tests with effect sizes (e.g., lines 244-245, 277-278)
• Sometimes reports percentages without tests (e.g., lines 230-237)
• Sometimes explicitly states differences are "not statistically significant" (e.g., line 257-258)
• Sometimes presents differences without testing (e.g., lines 238-242)

No clear decision rule for when statistical testing is applied vs. descriptive statistics alone.

1. Metadata Analysis Issues

Lines 157-171: "Although 8199 papers were pre-selected (from the documents available in Scopus on December 2021), those papers published in Astronomical and Astrophysical journals and other papers not related to cloud physics or written in a language different than English were removed leaving us with a total of 6987 papers"

Problems:

1. a) Language bias explicitly introduced (lines 166-167): "We note that the number of papers (i.e., 97) removed because of language could slightly impact the outcome of the present analysis; however, this was necessary because the authors of the present study were unable to translate them"

• This systematically excludes non-English publications, likely disproportionately affecting Global South researchers
• Authors acknowledge this but proceed anyway

1. b) Database limitations acknowledged but not addressed (lines 168-171): "the conclusions raised from the metadata analysis could be biased as Scopus, and most databases, does not cover all areas of interest leaving out important information from e.g., non-English-speaking countries or Global South countries"

• This is exactly the population the study aims to understand
• Critical methodological flaw for a DEI study

1. Tropical Country Classification

Lines 187-192: "For a country to be considered tropical in this study, at least 50% of its inhabitants need to live in the tropics, i.e., between 23.5° N and 23.5° S."

Problems:

• Arbitrary 50% threshold not justified
• Geographic location used as proxy for "Global South" but these aren't equivalent
• India (2.86% of publications, Table 1) coded as tropical, but Taiwan (0.64%) also tropical—very different development contexts
• Conflates geography with economic development and colonial history

1. Impact Factor as Outcome Variable

Lines 379-388: "The present data shows that in addition to the low number of studies published by tropical coauthors, on average, their indexed studies get published in lower-impact journals (impact factor IF=3.91, 95% confidence interval (CI): 3.72, 4.10 (n=287)) compared to their non-tropical peers (IF=4.58, 95% CI: 4.28, 4.88 (n=6638))."

Problems:

• Uses Impact Factor as measure of quality without acknowledging its limitations and biases
• Impact Factor favors established, English-language, Global North journals
• No discussion of how IF itself reflects and perpetuates the inequities being studied
• No statistical test comparing these confidence intervals
• Different sample sizes (287 vs 6638) not accounted for

1. Career Retention Analysis Issues

Lines 444-460 (Figure 6 and discussion): The analysis of "years until last registered paper" as proxy for leaving the field.

Problems:

• Assumes last publication = leaving field (but could be career pauses, database lag, non-Scopus publications)
• Right-censoring issue: Recent entrants (2015-2020) have less opportunity to leave
• No survival analysis or time-to-event methods applied despite temporal data structure
• Compares retention across time periods with different follow-up durations

Line 453-455: "However, in recent years, after 1995, we see no difference in years between the genders in how long they remain in the cloud physics field judging by publication activity."

• This conclusion unsupported—visual inspection of Figure 6 without statistical test
• Doesn't account for different entry cohort sizes

1. Missing Data and Non-Response

Lines 196-198: "Sixty seven percent of the participants were < 40 years old and 57.1% of the participants identified themselves as men, 41.4% as women, with 1.5% preferring not to respond"

Problems:

• 1.5% non-response on gender question acknowledged but not analyzed
• 5.6% non-response on sexual orientation (line 200-201)
• 6 respondents (3%) opted not to disclose ethnicity (line 207)
• No analysis of whether non-response patterns relate to other variables
• No sensitivity analyses for missing data

1. Temporal Confounding

Lines 75-78: "the inclusion of researchers from new geographical locations was not significant from 1990 to 2000 (Wagner, 2005) but it was significant between 1990 and 2013 (Wagner et al., 2017)"

Problem: The study combines 50 years of data (1970-2020) but:

• Academic publishing, databases, and geopolitics changed dramatically over this period
• No time-stratified analyses
• Cold War effects on collaboration (1970-1991) not distinguished from post-Cold War
• Internet and digitization effects not accounted for

1. Chi-Square Test Assumptions

Throughout the document, chi-square tests are used (e.g., lines 244-245, 277-278, 291-293), but:

• No verification that expected cell counts >5 (required assumption)
• With small subgroups (LGBTQIA+ = 8%, line 200), this assumption likely violated
• Should use Fisher's exact test for small samples

1. Correlation vs. Causation

Lines 349-358: Discussion of women first authors having more women co-authors.

Problem: States "This may not be the result of them being more equitable but could rather mirror a preference for working with people similar to oneself" but:

• No longitudinal data to establish temporal precedence
• No controls for confounders (field subspecialty, institution, career stage)
• Alternative explanations not tested (e.g., mentorship networks, discrimination by male-led teams)

Recommended Corrections:

1. Re-do gender analysis with validated methods and report uncertainty ranges
2. Apply survey weights to correct for known overrepresentation
3. Conduct power analysis and clearly state which analyses are exploratory
4. Use consistent statistical testing with multiple comparison corrections
5. Perform sensitivity analyses excluding English-only restriction
6. Apply survival analysis methods to career retention data
7. Stratify by time period in metadata analysis
8. Use appropriate tests for small sample sizes
9. Acknowledge Impact Factor limitations or use alternative metrics
10. Pre-register any confirmatory hypotheses for future work