The study by Zhang et al. measured atmospheric GEM concentrations and isotopic compositions at long-term monitoring urban site and multiple short-term monitoring urban, suburban, rural sites in China and Pakistan. Based on the comprehensive observations, the authors show notable declines in GEM concentrations and clear changes in the isotopic compositions in the urban atmosphere. By using a Hg isotope mixing model, they quantify the relative contributions of anthropogenic emissions and evidence that the GEM declines are mainly driven by the control of anthropogenic emissions. This research is well designated. I broadly agree with the interpretations throughout the manuscript. I would suggest a minor revision of this manuscript (the manuscript is well written currently). Some of the minor suggestions are presented below.

The source apportionment: in this study, the authors use a δ202Hg and Δ199Hg mixing model to quantify the source contributions. I would suggest to use the Δ199Hg and Δ200Hg mixing model as previous studies. This is because the GEM δ202Hg in the boundary layer apt to modified by vegetation uptake process (particularly at the suburban and rural sites in this study), while the Hg-MIF signals are relative stable. As seen from Fig. 3a, some of the δ202Hg and Δ199Hg assemble especially in the background areas are not fully encompassed by δ202Hg and Δ199Hg of the three source endmembers. The authors argue that the Δ200Hg of anthropogenic emissions overlap surface re-emissions. However, the Δ199Hg of anthropogenic emissions and surface re-emissions are distinguishable, and this could enable the source quantification of these two sources. In addition, the Tianjin sampling site is close to seas, and the seawater re-emissions should be also considered. A recent study reported mean Δ199Hg and Δ200Hg values of -0.13‱ and 0.02‱, respectively, for ocean emissions (Fu et al., 2026, NSR). This oceanic signature is identical to soil re-emissions. Therefore, using the Δ199Hg and Δ200Hg mixing model would help to understand the contributions from the two most important natural emissions (e.g., soil + seawater reemission).

Line 13: please add ‘mean’ before GEM concentration.

Fig.1: better to show the location of Karachi in this figure.

Line 127: please specify whether the sampling flow rate is operated under the standard temperature and air pressure.

Section 2.2: please provide the sampling interval or duration for each sample at the sampling sites, including the pump-trap and passive samples.

Section 2.5: please add the method for the calculation of GEM concentrations using the pump-trap. Are all concentrations reported for the STP conditions?

Line 221: add ‘mean±1sd’ after the values.

Line 223: please note ‘mean±1sd’ for the values reported in Beijing and Shijiazhuang.

Line 235: add ‘mean,’ before n =35.

Line 238: add ‘mean,’ before n =35.

Line 240: better to show the exact p value when it is higher than 0.05.

Line 245: add ‘mean’ before δ202Hg.

Line 246: change ‘directly’ to ‘mainly’?

Line 247-248: add ‘mean±1sd’ before n=16. Same in line 249, 251 and 255.

Line 257: show the real value instead of >0.05.

Line 262: add mean before GEM.

Line 267-268: not note these are mean values.

Line 277: add ‘mean’ before values.

Line 303: please add ‘mean±1sd’ after the D200Hg values.

Line 309: add ‘mean’ before GEM.

Line 324: please add ‘mean’ before δ202Hg and D199Hg. please also check other place if the values are referred to mean values.

Line 343: please add ‘on average’ before contributing.

This paper characterizes the decreasing trends of GEM and shifts in Hg isotopes in Chinese and Pakistani atmospheres. The authors conclude that decreased anthropogenic emissions are the main driver of this decline. Although the study is clearly presented, the accuracy of the source apportionment remains questionable due to significant gaps in the sampling periods, which may bias the seasonal representation. Please find my specific comments below;

Major comments: 

- 2.1 The samplers were installed on the rooftop of Tianjin University. Has the influence of local exhaust systems (e.g., laboratory vents or heating units) been evaluated? It is necessary to confirm that the collected GEM represents the ambient urban atmosphere rather than localized emissions.

- 2.6 There are significant temporal gaps in the dataset across Phases I to III. Notably, Phase II lacks data from the winter heating season (most of December to March).

The current source apportionment may represent an optimized solution for the measured periods, but it likely carries a substantial bias as an annual average for 2021-22 due to the missing winter data.

Please clarify which datasets (active only, or a combination of active and passive) were used for the mixing model. While passive samplers were deployed during the periods when active sampling was unavailable, these provide only 30-50 day integrated averages, making them difficult to compare directly with high-resolution active data.

In Phase II, δ202Hg values range from -0.49‰ to 1.38‰. Since 1.38‰ exceeds the defined background end-member (≈ 0.50‰), how was the contribution calculated for such samples?

Please provide a quantitative estimate of the errors/uncertainties associated with the source contribution results.

- The authors attribute the decline in GEM concentrations to mercury emission regulations. Are there any auxiliary atmospheric composition data (e.g., SOx, NOx) available to support this claim and verify the reduction in combustion-related emissions?

Minor comments: 

- L164 The expression (***Hg/198Hg)NIST3133 should be revised to "NIST SRM 3133" in accordance with Equation (1).

- For all data presented in the text, please specify whether the indicated errors represent 1SD or 2SD, and consistently include the number of samples (n).

- Figure 1 The text in Figure 1 is difficult to read, and the scale of the enlarged map is unclear. Additionally, the sampling point for Karachi should be explicitly marked on the map.

- A systematic offset in δ202Hg exists between active and passive sampling. While the authors performed baseline experiments (Figure S2) and applied corrections (Table S4), it is recommended to use labels such as "corrected-δ202Hg” in both the main text and supplementary figures to avoid confusion.

- Figure 6b There is a symbol resembling a stop button. Is this symbol necessary? If not, please remove it to maintain graphical clarity.

- The use of the minus sign (hyphen vs. en-dash) is inconsistent throughout the manuscript. Please unify the mathematical symbols.

This paper presents a long term GEM concentration - isotope record spanning pre-control, lockdown, and post-pandemic phases (2018–2025) in E China and S Pakistan. Certainly, it documents a fundamental shift in urban atmospheric mercury quality in Tianjin, China. The study is timely, policy-relevant (Minamata Convention), and methodologically advanced. However, I have a number of comments:



Major comments:

1. Representativeness of sampling sites
   
   Please discuss the possibility of the sampling site being a receptor of localized emissions rather than urban ambient atmosphere.
   
   

2. Small sample sizes in Phase I (2018) – consists of only 3 active samples collected in a single month (November). Please acknowledge this as a limitation, but also note that Phase I isotopic values are consistent with a larger compilation of Chinese urban GEM (Zhang and Sun, Earth-Science Reviews, 2006).

3. Statistical treatment of trends The paper states that GEM concentrations and isotopic compositions differ significantly between phases (p<0.05), but no trend analysis (e.g., Mann-Kendall, Theil-Sen) is applied to the continuous time series within Phase II and III.
   
   Add trend analysis for Phase II–III; report statistical tests in figure captions.
4. Diurnal interpretation
   
   The diurnal isotopic shift (lower δ²⁰²Hg, higher Δ¹⁹⁹Hg during daytime) is attributed to surface re-emissions, but atmospheric photochemical oxidation of GEM to Hg(II) can also fractionate isotopes in ways that might appear similar. The paper does not fully rule out oxidation-driven MIF changes.
   
   Please add a brief discussion.

5. Lack of direct flux measurements – surface re-emissions inferred but not quantified.

6. Passive sampler MDF correction
   
   The MDF correction factor for MerPAS samplers is applied uniformly, but its uncertainty (±0.33‰) is large relative to some of the δ²⁰²Hg differences reported (e.g., urban-suburban difference ~0.3‰). The potential impact on the mixing model results is not fully assessed.
   
   Perform a sensitivity analysis showing how varying the correction factor affects source apportionment, or at least discuss this uncertainty explicitly.

Minor comments:

• Equation (2) appears incorrectly written – likely a typo.
• Figure 2 caption: The caption states "no statistically significant differences" between active and passive sampling but doesn’t specify the test used or p-value.
• Please add information if GEM concentrations are normalized to STP.