| 28 Jul 2026
Contrasting physical and biological drivers of oxygen change in the northern California Current System
Abstract. Eastern Boundary Upwelling Systems like the California Current System are among the ocean’s most productive regions, shaped by distinct surface and subsurface currents and water masses. They are also vulnerable to deoxygenation, with several systems experiencing extreme low-oxygen events and associated mass mortality in recent decades. Determining potential drivers is complicated by seasonal dynamics, multiple water sources, and biological consumption. In the northern California Current System, oxygenated subarctic water mixes with oxygen depleted water from the south. We analyse three 20-year records, including one synthesized from local data, to provide valuable insights into subsurface oxygen variability. These long coastal time series are rare. We show that on the slope in spring, subsurface oxygen variations are physically driven by the relative contributions of these water masses, and their salinity and temperature ratios (a property called `spiciness') can be used to estimate oxygen levels. Seasonally, decreasing spiciness provides evidence of springtime oxygen renewal on the shelf, before respiration consumes oxygen over summer. Since 2003, mid-depth isopycnals show increased southern water contributions in spring and summer, leading to lower oxygen in spring, but not summer, suggesting summer-time oxygen consumption may have decreased with time. On the densest isopycnal near the shelf floor (~ 200 m), where oxygen is already lowest, no changes in spiciness or oxygen have been observed. Future research should examine the use of temperature and salinity ratios to estimate oxygen levels in the wider California Current System, and how changing source water conditions and biological activity may affect the risk of low oxygen in coastal waters.
This manuscript investigates the drivers of oxygen changes in the coastal northern Pacific using a novel approach based on the spiciness-oxygen relationship. The method and results are clearly laid-out. This contribution will be suitable for publication once further validation of some of the results is provided.
Main comments
1. In a recent publication by the same authors, the temporal evolution in spiciness is discussed in much the same way as in the present manuscript. Most of the figures in Maier et al (2025) are reproduced in the supplementary material of the present manuscript or as panels of the main figures. Since these results have already been published, I recommend that for these figures the authors refer to the published paper instead of reproducing them. Part of the results and of the discussion of trends could probably also be removed from the present manuscript.
2. The analysis is based on annually averaged observations. Two pieces of information are missing to support the robustness of the results: the spread of the values underlying each annual average (standard deviation), and the number of observations contributing to it. If the spread is much larger than the obtained trends, the trends may not be significant. Without this information, it is also not possible to assess whether there is a strong spatial or temporal variability within the considered regions and time frames, which could affect the results if there are biases in the observations (towards specific locations or dates in certain years). Therefore, I suggest that the authors:
(i) compute the standard deviations associated with the annual averages, and possible the associated uncertainty by also taking into account the number of observations;
(ii) display them as error bars on figure 2, 3, 4;
(iii) modify the statistical tests so that they account for these standard deviations. For instance, one approach is to weight each data point by the inverse of its variance.
These validations appear essential to confirm the validity of the results.
3. While most of the analysis are qualitative, the proposed method seems to offer possibilities for more a quantitative treatment.
First, the mixing ratio of the two parent water masses could be quantified from the distance along the spiciness axis of the mixing line (in Fig. 1).
Second, I believe that the vertical offset in oxygen between individual points and the mixing line would provide a quantitative estimate of departure from parental water properties, and hence of respiration.
Third, as discussed at L186-187 referring to other studies, a trend in parent water masses properties would affect the results. These trends could be quantified based on the dataset used here. I suggest adding a figure showing these trends possibly in the supplementary material, and then discussing the effect of the trends on the results. Maybe such trends could provide hints into the unexplained trends discussed at L186-187.
Then, I would encourage the authors suggest to use these quantitative analysis to provide more quantitative results in the abstract and conclusion, instead of general conclusions such as the one at L201.
4. The proposed method assumes that mixing is isopycnal. I suggest that the authors discuss the potential role of diapycnal mixing, as mixing could change the properties as waters move between the slope and the shelf. For instance, the following studies seems to indicate a significant role for diffusion and vertical mixing in shelf-sea interactions in that region: https://doi.org/10.1029/2010JC006720, https://doi.org/10.1080/07055900.1989.9649345.
5. I suggest deepening the interpretation of the results (L93-100, L147, L163) proposing a physical/climatic hypothesis to explain the observed change in water mass composition, for instance in relation to known changes in circulation, ventilation, or other mechanisms at play in the region. Hypothesis are provided for the change in respiration, but not in composition. Relatedly, multi-annual variability appears to be visible on Fig. 4 in addition to the long-term trends. It could be acknowledged, and suggests a role for natural or climatic variability.
6. I would like to clarify one part of the discussion. In Table S3, the spiciness trend is weaker in LUW than in EUW, as also mentioned at L166-167. However, Fig. 4 shows the opposite. Please clarify this inconsistency. From the figure, the difference in the spiciness trend between the two appears significant, and would reduce the O2 trend, bringing it close to the level of detectability, as observed.
Specific comments
Throughout, these appears to be a formatting problem with the density, with a missing equal sign or space. See for example L31.
L7-10 : The text from “We show” to “over summer” would benefit from clarification. It is not clear what the authors mean by “Seasonally” in the second sentence.
L25: Please clarify the northern CCS are the most productive in terms of what and compared to what.
L26: Please replace “extreme low” with a concentration.
L26: “a mix of” may be more accurate than “subject to”.
Fig. 1 inset: a colorbar is missing for spiciness
Fig. 1 caption: A colon may be needed so that the text reads “Inset: mean ...”
L61 : The reader wonders if the depths provided in parenthesis are for a specific season. Actually, I suggest adding a figure showing density and oxygen profiles in the four locations. This would help the reader interpret the water mass dynamics.
Fig. 2: I am not convinced that panels b-c, e-f, and h-i are necessary.
L132: I suggest comparing these rates with existing estimates, and commenting on whether they are realistic.
L134: Is the higher productivity related to terrestrial nutrient input ? If yes, this could be mentioned.
L136: This first sentence is strongly redundant with sentence “However...” at L137.
Fig. 3 caption: Please clarify what is meant by “dots slightly offset for legibility”. Is it difficult to distinguish offsets from genuine differences.
L160-165: These sentences are difficult to follow; I suggest rewriting them for clarity.
L161: The comma after the closing parenthesis appears unnecessary.
Fig. 4: For the shading showing the bootstrap confidence intervals, a bootstrap interval should contain the trend line. Here, the trend line lies at one edge of the shaded range or even outside it. Please check the calculation or the plotting. A further indication of a possible problem is that the shading appears to be symmetrical about the zero-slope line rather than about the fitted trend. This would normally imply that the trends are not statistically significant at the corresponding confidence level.
L192: The same result would, I believe, also be obtained if the contributions of the two water masses changed by equal amounts in opposite directions.
Fig. S2: Panels d and e are redundant, I suggest removing panel d.
Fig. S10: I am always uncomfortable when trends that are not statistically significant are displayed on figures. I would suggest removing the trends from this figure, or maybe showing them with a very faint dotted line.