The authors conducted about one month (12/2/2020-1/1/2021), 3 weeks (10/14/2020-11/4/2020), and 2 weeks (12/29/2019-1/16/2020) of TGM measurements at two island sites, JHI and HNI, and on a cruise ship. They estimated anthropogenic contributions to ambient concentrations of TGM using PMF and linear regression analysis. They also estimated sea-air exchange flux of Hg° using an air-water exchange flux equation (Wanninkhof, 1992, JGR). Over the past couple of decades since the Tekran series has been deployed globally, numerous long-term datasets of speciated, operationally defined, mercury concentrations have been available and used to study atmospheric Hg cycling, which has generated a large body of research in the literature. While the authors performed a comprehensive analysis with what they got, the short-term nature of their datasets limited their ability to provide substantial insights into atmospheric Hg budgets. The study also presents several methodological concerns.

1. PMF application and interpretation issues. There are multiple concerns regarding the PMF application and interpretation:

• The authors used PMF to identify the factors contributing to ambient TGM at DSL and JHI. DSL was a long-term (2015-2019) monitoring site over land near the coast in Shanghai, while the field campaign at HNI, an island site near DSL, took place a year later. The authors should clearly specify the time periods used in their PMF analysis for the two sites in Section 2.5.
• The study assumes that the empirical relationship between anthropogenic TGM contributions and BC concentrations (derived from 2015–2019 land-based data) is applicable to TGM data at an island site over 100 km away and one year later. This assumption is highly questionable, as empirical relationships may not necessarily hold across different locations and timeframes. At the very least, the authors should acknowledge the potential uncertainty introduced by this approach.
• The size of the dataset for JHI appears too small for PMF. It was a month-long campaign. What was the temporal resolution they had for their datasets? Fig. 5d seems to show about 20 data points. If this represented the number of data points for a single variable used for PMF analysis, then their results would be questionable. The authors should reference Zhang et al. (2009, https://doi.org/10.1016/j.atmosenv.2009.07.009), which discusses the minimum sample size required for PMF applications. That reference is just an example from a large body of literature on the topic. In fact, I am wondering, if the authors used the 2015-2019 data for DSL PMF analysis, how come only a handful of data points were shown in Fig. 5c?
• The rationale behind selecting certain tracers is unclear. Why did the authors use V for the shipping emission tracer? The tracers for the cement industry were Ca and Fe, which could very well be indicative of dust. Why was BC not used in the PMF analysis?

2. TGM/BC ratio. The authors highlighted the TGM/BC ratio in the abstract as a key finding. However, this ratio appears to be just another variable rather than a novel result that provides additional insights.

3. Sea-Air exchange flux calculations. The study recalculates sea-air exchange flux after removing anthropogenic contributions from ambient data. However, the purpose of this recalculation remains unclear. This issue also relates to the statement in the abstract (Lines 39–40), which needs further clarification.

4. Unsupported assertions. Assertions throughout the manuscript lack supporting evidence or citations. Below are a few examples:

• Line 256: The term “elucidation” is misleading, as the statement is purely speculative.
• Lines 261–263, 265–266 (JHI), 266–268, 287: Assertions require supporting evidence or references.
• Lines 281–300: This paragraph is speculative and lacks supporting evidence.
• Line 319: The claim appears overstated.
• Line 322: Requires a reference.
• Lines 426–428: Wouldn’t higher temperatures enhance the partitioning of Hg° from water to air? The authors should clarify this mechanism.
• Lines 427–428: Supporting evidence or references are needed for the statement.
• Lines 430–431: Requires citations.

5. Insufficient methodological details. For the ancillary data of ion concentrations, trace gases, and meteorological variables, the authors provided little information on the instruments used, and no information on data quality control and assurance as well as temporal resolution. Also, where were the PBL data from? They were introduced abruptly at one point in the results section.

6. Random citations. Some references seemed to be cited arbitrarily. While citing every study on a given topic is impractical, it is important to acknowledge milestone research appropriately. Here are a few examples. There have been hundreds and thousands of journal articles on PMF applications. Did Qin et al. (2020) develop the PMF approach? Was Gibson et al. (2015) the first to recognize PMF “for its efficacy in elucidating sources profiles and quantifying source contributions”? In lines 280-281, were those studies the first to establish the role of temperature in GEM evasion? In lines 321-322, were those studies the first to identify fossil fuel combustion as a major mercury source?

7. Uncertainty in sea-air exchange flux calculations using TGM as a proxy for Hg° in sea-air exchange flux calculations could introduce significant uncertainty. While this may be reasonable in a landlocked atmosphere, it can be problematic in the marine boundary layer, where halogen compounds are abundant and subsequently GOM concentrations are probably not negligible at times. For example, Castagna et al. (2018, atmos. Env.) reported GOM reaching well over 100 pg/m³, ~10% of TGM, at times. Note that in the reference cited, GOM was measured using the Tekran series, which has been in the literature suggested to be largely under-biased, primarily by Gustin et al.’s team. The actual GOM concentrations may be even higher.

Please find in the attachment.