Fungal spores as a source of sodium salt particles in the Amazon basin

In the Amazon basin, particles containing mixed sodium salts are routinely observed and are attributed to marine aerosols transported from the Atlantic Ocean. Using chemical imaging analysis, we show that, during the wet season, fungal spores emitted by the forest biosphere contribute at least 30% (by number) to sodium salt particles in the central Amazon basin. Hydration experiments indicate that sodium content in fungal spores governs their growth factors. Modeling results suggest that fungal spores account for ~69% (31–95%) of the total sodium mass during the wet season and that their fractional contribution increases during nighttime. Contrary to common assumptions that sodium-containing aerosols originate primarily from marine sources, our results suggest that locally-emitted fungal spores contribute substantially to the number and mass of coarse particles containing sodium. Hence, their role in cloud formation and contribution to salt cycles and the terrestrial ecosystem in the Amazon basin warrant further consideration.

The claims in the MS are significant and novel. The two closest papers known to me, one from (partly) the same research group, discuss that 1) K-rich bio-aerosols are seeds for SOA formation, but the paper does not mention Na-containing particles (Poehlker et al. 2010) and 2) that when spores rupture at high RH, fungal fragments that contain Na salts are released, but the paper does not quantify Na-content of the particles nor their contribution to the particle Na-budget over Amazonia (China et al. 2016).
The results are of interest to others in the community, since they give 1) new insight into Na-budget of the Amazon from a source that was not considered previously and 2) they add a new line of thought to the debate on sources of cloud-active particles in Amazon, which is a long-standing problem.
The results are of interest to the wider field, because this work adds a new aspect to the understanding of biogeochemical cycles in the Amazon rainforest, and how the forest itself sustains biogeochemical and hydrological recycling. Especially the role of fungal spores therein is one that has not received much attention so far, and this work is likely to start a further debate on this topic.
The MS is clearly structured and generally convincing, but I think it could profit from some further clarifications: 1.
Since I am not familiar with the instruments that are used, I cannot judge the quality of the experimental work. However, I think the MS would benefit from a short explanation of the sampling procedure: are all coarse particles identified that were sampled? How is it determined that a particle is a fungal spore and not another kind of PBA (e.g. bacterium or pollen)? Is this done based on size, morphology or otherwise? A brief description could be added, for instance in L107.

2.
The estimate of the contribution of spores to the Na-budget, depends on two factors which are very uncertain: local emissions of fungal spores and transport of sea salt aerosol into the Amazon basin. I think the fungal spores emissions are constrained reasonably well by the observations of their concentration, which the model seems to underestimate somewhat. However, if global emission estimates of sea salt span 2 orders of magnitude, how reliable is the estimate of sea salt aerosol transported into the Amazon basin, with no direct observations available for validation? Can the authors provide an estimate of the uncertainty in the sea salt contribution to the particle sodium budget? This could provide a bit more context to the number of 62% of total sodium mass that is now assigned to fungal spores.
Minor comments: -Title: Since the main finding is that fungal spores contribute significantly to sodium salt particles, why not mention this in the title, for instance: 'Fungal spores as source of sodium salt particles in the Amazon basin' -L29-30: pls add that this is the case for the wet season -In L52-53, you state that marine aerosol is considered the main source of coarse aerosol, whereas in L213 you state that fungal spores dominate the coarse mode. It would be good to clarify what is actually meant by dominate: mass, number? And to briefly include the evidence for the statement in L213.
-L107: how many particles were sampled and how was it ascertained that they are representative for the wet-season coarse aerosol? And how was a particle identified as a fungal spore? Was every collected particle in the sampling periods in Supp. Table 2 classified?   -L180: since the back trajectories and rain records are based on reanalysis data, I would not refer to them by 'observations' here, but rather by 'simulations' - L190-191: could you clarify here where the percentages of 70 and 13% for sodium richspores and spore sodium mass, respectively, are based on? -L208-209: how would nighttime emissions or a shallow nocturnal boundary layer explain that the Na fraction from spores drops only after noon, and not in the early morning at the onset of turbulence? Besides, the violin plots in figure 3 seem to show bimodal distributions for the Na fraction from fungal spores for each time of the day. What is the reason for that? -L217: 'a major fraction to the sodium budget': please add here that this is for the wet season -L533-534: which size distribution was assumed for the spores? Within the range of the model coarse mode (1-10 micron), the assumed size can make large differences for the efficiency of the wet and dry removal -Supp. Fig. 9: It seems that the same violin plot is show twice: the plot looks very similar to that in Fig. 3 and the title says 'wet season' while the figure is for the dry season. Please include the right plot for the right season.   1 We appreciate the reviewers' comments and suggestions; they provided positive and constructive feedback. The reviewers pointed out several important issues that we believe addressing will indeed strengthen the manuscript.
In the following discussion, the reviewers'comments are in black-colored font, our responses are in blue-colored font, and changes in the manuscript are noted by italics blue-colored font.
Reviewer #1 (Remarks to the Author): This paper reports that fungal spores are major sources of particle sodium, with size distributions and properties that are similar to that from sea salt particles. This is a new finding that will be of broad interest and certainly justifies publication in Nature Communications. Given the importance of understanding the role of particles in cloud formation and particles, which impacts climate, this will definitely influence thinking in the field. The paper is very well-written and documented, and this reviewer has essentially no substantiative comments or suggestions to make. A very (!!) minor suggestion is to add to the description of Supplementary Figure 9 the same description as for Figure 3b, since it is an unusual (but effective) way to present data. We appreciate the positive response of the reviewer regarding the novelty and importance of the findings reported in the manuscript and for recommending the manuscript for publication. As suggested, we added more detailed description of Supplementary Figure  Reviewer #2 (Remarks to the Author): 1. The major claim of the paper is that sodium salts present in atmospheric aerosols arise from fungal spores. Fungal spores represent a salt source that has previously not been reported. This is a fascinating source. However, the manuscript's assessment of the importance of this finding is not supported by previous literature or clear explanations. Regarding previous literature, the abstract states, "In sharp contrast with the assumption that sodium-containing aerosols arise solely from marine sources,..." This statement is simply not true. There is a wealth of papers on non-marine sources of sodium salts in aerosols. We agree that fungal spores are a fascinating source of sodium salt particles in Amazonia. Sodium content in atmospheric aerosol is primarily attributed to sea water, especially the coarse mode particles. However, as the reviewer pointed out, there are also non-marine sources of sodium salts in atmospheric aerosols. For example, Ooki et al 1 observed the existence of anthropogenic sodium in fine mode aerosol particles in urban air. Ooki et al 1 suggested potassium as a tracer for anthropogenic sodium. Mamane 2 found presence of sodium and potassium in submicron particles from waste incinerators emissions. Another study conducted in the metropolitan area of São Paulo by Bourotte et al. 3 found K/Na ratio of 1.4 and 0.9 for coarse and fine particles respectively and they attributed this particles to waste incineration and vehicles. Most of these studies show the presence of sodium in fine mode and suggest potassium as a tracer for anthropogenic sodium. Observations of our study are fundamentally different. First, our study focuses on coarse mode particles. Second, the potassium observed in our study mostly originates from biological particles 4 , especially during the wet season when biomass burning is not a dominant source in the basin.
To clarify these details, we modified the sentence and the manuscript (as suggested by Reviewer 3) to reflect that our main finding is that fungal spores contribute to sodium salt particles in the Amazon basin. "In contrast to the common assumption that sodium-containing aerosols are primarily arise from marine sources, our results suggest that locally-emitted fungal spores containing sodium contribute substantially to the number and mass of coarse particles containing Na and Cl (similar to sea salt)." We adopted the title proposed by reviewer 3. "Fungal spores as a source of sodium salt particles in the Amazon basin" 2. More problematically, while the observation of salts arises in fungal spores in interesting in its own right, the manuscript in its present form does not provide an adequate explanation or even compelling suggestions for how/when/under what circumstances or via what mechanism the salt in incorporated into some, but not all, spores. If an insightful discussion were included, I would rate this manuscript much more positively. As written, it is lacking.
In the original manuscript we provided basic information (lines: 167-172) regarding the nature for sodium content in fungal spores. To address the reviewers concerns, in the revised manuscript we provided more detailed discussion on this topic.
The following text was added in the revised manuscript (lines: [172][173][174][175][176][177][178][179][180][181][182][183][184][185][186][187][188][189][190]. "The presence of sodium is common in so called halophilic fungi which cannot grow without NaCl 5 . Active discharge, uptake, and efflux processes are likely responsible for the sodium content in the spores and sodium content may vary with different classes and genera of fungi. During active discharge, fungi forcibly eject spores into the atmosphere, together with osmotic fluid containing hexoses, mannitol, phosphate, sodium, and potassium 4,6 . Fungi require K + for electrical and osmotic equilibria of the cells and presence of K + in several fungal spores is well understood 7 . Previous studies suggested that K + can be partially replaced by Na + and Na + can enhance the growth of fungi 8 and plants 9 under K + deficiency conditions. The growth of fungi and uptake of Na + varies with their physiological conditions and uptake rates depend on the species and their genes 10 . For example, specific genes (e.g., acu1 and acu2) are responsible for high-affinity Na + uptake of Ustilago maydis. When fungi contain excess Na + , they activate Na +efflux ATPase (Adenosinetriphosphatase), which acts as a key enzyme for the biological evolution of fungi 11 . We suggest that uptake and efflux of Na + varies with different classes and genera of fungal community. For example, previous studies in Amazonia shows several genera (e.g., Agaricus; Amanita; Aspergillus; Boletus; Cladonia; Mortierella; Puccinia; Lepsita; and Rhizopus) within one class of fungi (Lecanonomycetes) 12,13 . Furthermore, the transpiration of plants 14 and nutrient uptake 15 can also influence the sodium content of spores. Further studies are needed to better comprehend the sodium content in fungal spores by linking chemical composition, molecular biology and diversity of fungal spores."

3.
A second finding is that spores which contain salt are more hygroscopic than those which do not. While I'm not aware of any previous results on precisely this topic, it is very well known that salts are deliquescent materials whereas organic macromolecules and other organic compounds found in spores are not. So, this result is very obvious. Indeed, it is expected that sodium-containing spores are more hygroscopic that those without sodium. Even if the growth factors are similar for sodium chloride particles and sodium containing spores, it is interesting to note that growth factors of NaCl and sodium-containing spores are different at relative humidity between 70 and 96%. Notably, variations in the hygroscopicity are associated with variations in the chemistry of these spores. We believe this finding is important and worth discussing in the manuscript.
A minor point -4. According to Figure 3, the Na fraction for fungal spores is estimated to be between 0.1 and 1.0 (i.e. nearly any value in the early hours, and later in the day transitions to an even wider spread in possible fractions. I may be missing something, but I don't find this result to be a significant finding. We believe Figure 3b provides important contextual information to help readers understand the temporal variability in the simulated fungal spore contribution to sodium, near the measurement site. Considering the high temporal variability of aerosol particle concentrations, including the simulated sea salt and fungal spores in this study (Supplementary Figure 11), it is not surprising that the relative contribution of fungal spores to total sodium mass spans a wide range of values. Figure 3b shows that from midnight to noon, the wet season median contribution of fungal spores to Na ranges from about 0.5 to 0.6. This implies that fungal spores are the dominant contributors of Na on the majority of days at these times. By contrast, from 3 pm to 9 pm, however, the median ranges from 0.15 to 0.3, and the dry season median fungal spore fraction ranges between only about 0.1 and 0.2 (Supplementary Figure 11). In other words, the model helps contextualize the limited number of observations, by identifying the circumstances under which the fungal spores are predicted to contribute the majority to simulated boundary-layer Na at the site: specifically, during the late night and early morning hours of the wet season.
5. In summary, the authors have one potentially interesting surprising field observation in the Amazon. However, additional consideration of how this may arise is needed before I could recommend publication. We appreciate the comments and suggestions provided by the reviewer. As outlined in the response to the comments 2, in the revised manuscript we expanded the discussion substantially regarding the presence of sodium in fungal spores. In addition, the observation of sodium in fungal spores is on samples from a range of different days.
The results are of interest to others in the community, since they give 1) new insight into Nabudget of the Amazon from a source that was not considered previously and 2) they add a new line of thought to the debate on sources of cloud-active particles in Amazon, which is a longstanding problem.
The results are of interest to the wider field, because this work adds a new aspect to the understanding of biogeochemical cycles in the Amazon rainforest, and how the forest itself sustains biogeochemical and hydrological recycling. Especially the role of fungal spores therein is one that has not received much attention so far, and this work is likely to start a further debate on this topic. We appreciate the reviewers' careful reading of our manuscript and providing us positive feedbacks.
The MS is clearly structured and generally convincing, but I think it could profit from some further clarifications: 1) Since I am not familiar with the instruments that are used, I cannot judge the quality of the experimental work. However, I think the MS would benefit from a short explanation of the sampling procedure: are all coarse particles identified that were sampled? How is it determined that a particle is a fungal spore and not another kind of PBA (e.g. bacterium or pollen)? Is this done based on size, morphology or otherwise? A brief description could be added, for instance in L107.
Initially, due to space constraints, we couldn't provide details on the methods used for identifying the fungal spores in the manuscript. Typically fungal spores are 1-6µm, while pollen grains are larger in size (5−150 µm), bacteria are smaller (<1µm) and typically elongated in shape. As we selected only impactor stages 4 and 5 (aerodynamic diameter range: 1.0-3.2 μm) for this study, the fraction of bacteria and pollen is almost negligible. We also refer to our previous publication (China et al., 2016) where we described identification of fungal spores.
To address reviewer's comment, we moved portion of text from the methods section to the main manuscript and provided additional explanations (lines: 108-112) "The fraction of fungal spores was quantified based on their unique characteristic morphologies (spherical, rod-like or spheroidal in shape), size (1-6 µm), and chemical composition (mostly carbonaceous and containing phosphorous) by electron microscopy imaging and X-ray microanalysis of over 3500 individual particles. Details of the fungal spore identification method are provided elsewhere 16  2. The estimate of the contribution of spores to the Na-budget, depends on two factors which are very uncertain: local emissions of fungal spores and transport of sea salt aerosol into the Amazon basin. I think the fungal spores emissions are constrained reasonably well by the observations of their concentration, which the model seems to underestimate somewhat. However, if global emission estimates of sea salt span 2 orders of magnitude, how reliable is the estimate of sea salt aerosol transported into the Amazon basin, with no direct observations available for validation? Can the authors provide an estimate of the uncertainty in the sea salt contribution to the particle sodium budget? This could provide a bit more context to the number of 62% of total sodium mass that is now assigned to fungal spores.
The reviewer is correct that the model emission estimates of the sea salt are uncertain.
To address this, we compared model estimates of atmospheric sea salt concentration with observed values at few coastal and remote island sites to investigate the fidelity of the model estimates of sodium from sea-salt. We have added a scatterplot comparing observed sodium mass concentration with model-simulated mass sodium mass concentration from sea-salt, together with a global map (Supplementary figure 9) of measurement sites.
We added the comparison of model-simulated and measured sodium mass (Supplementary figure  10) and added the following text. We would also like to note that the sodium mass fraction in sea salt in the original manuscript was underestimated. The sodium mass has been corrected (30% sodium in sea-salt) 20 and all values recalculated. The revised model estimate of fungal spore contribution to the total sodium mass is 69% during wet season.
Minor comments: 1. Title: Since the main finding is that fungal spores contribute significantly to sodium salt particles, why not mention this in the title, for instance: 'Fungal spores as source of sodium salt particles in the Amazon basin' Based upon the reviewer's suggestion the title is changed to "Fungal spores as a source of sodium salt particles in the Amazon basin" 8 clarify what is actually meant by dominate: mass, number? And to briefly include the evidence for the statement in L213. When we state that marine aerosol is considered the main source of coarse aerosol, we meant this was a common assumption, in contrast to our measurements. We meant that fungal spores dominate the coarse mode by number in below-canopy air, and at night, based on our microscopy observations. Fractions of fungal spores are provided in lines 113-116: "During sampling, the number fraction of fungal spores was higher below the canopy (~60%) than above the canopy (~38%) and higher during night (~52%) than day (~37%)." Figure 1b demonstrate the number fraction of Na-rich particles. We modified the sentence as follows (lines: 254-255). "Overall, fungal spores dominated the number fraction of coarse particles in the Amazonia during nighttime and in below-canopy samples, and half of the coarse particles are sodiumrich." 4. L107: how many particles were sampled and how was it ascertained that they are representative for the wet-season coarse aerosol? And how was a particle identified as a fungal spore? Was every collected particle in the sampling periods in Supp. Table 2 classified? As explained in the reply to the comment 4, we added the information of total number of particles in the main text (line: 111). "The fraction of fungal spores was quantified …..by electron microscopy imaging and X-ray microanalysis of over 3500 individual particles." Approximately 80% of the particles that were collected in the sampling period mentioned in the SI Table 2 were classified (see detailed reply in comment 1). We added the number of particles analyzed per sample in the supplementary table. We sampled particles for different durations (~2.5-8 hours) with varying duty cycles to get representative samples both during day-and nighttime. We also performed back-trajectory analysis throughout the wet season and overall transport is similar to the transport pattern of airmass that was sampled during our sampling time. In addition, to better understand the representativeness of the analyzed particles with respect to the total particle population, we performed statistical analysis (applying binomial distribution) 21 to estimate the particle sample size of the particle population with given confidence interval (99%) and margin of error. We revised supplementary table 1 to reflect that samples particles are representative of the particle population with 99% confidence interval, and added the margin of error during each of the sampling periods.
As explained in the reply to the comment 1, we have provided additional text (lines: 108-112) in the main manuscript explaining the methods used for identification of fungal spores.

L180
: since the back trajectories and rain records are based on reanalysis data, I would not refer to them by 'observations' here, but rather by 'simulations' We changed the sentence to make it clear. "Hence, these findings support the dominance of locally-emitted spores in the Amazon basin during this study." 6. L190-191: could you clarify here where the percentages of 70 and 13% for sodium rich-spores and spore sodium mass, respectively, are based on?
We added the following sentences in the revised manuscript to clarify our results (lines: 536-540) "The percentage of sodium content in fungal spores was estimated using the known mass fraction of spores 22 and carbon/sodium wt% from X-ray microanalysis. Carbon content in spores ranges from 42-66%, with an average of 51% 23 . The average carbon/sodium wt% in fungal spores is 4:1. The mass of the spores is estimated from the size distribution of microscopy analysis and assuming a density of 1 gm cm -3 ) 22,24 ." Our estimation of 70% represents an upper bound. In the revised manuscript we provide a sensitivity analysis where we use different fractions (e.g., 30% and 50%) of sodium-containing fungal spores estimated from the samples. We add a modeled time series plot of sodium mass contributed by fungal spores assuming 30% and 50% sodium-rich spores in the sample (Supplementary Figure 13). We added few lines (242-245) to present this sensitivity analysis.
"We performed a sensitivity analysis by assuming either 30% or 50% of fungal spores to be sodium-rich. When applying these different assumptions to the model simulations, fungal spores account for 58% (21-90%) or 65% (27-93%) of the total simulated sodium mass during the wet season ( Supplementary Fig. 13)." 7. L208-209: how would nighttime emissions or a shallow nocturnal boundary layer explain that the Na fraction from spores drops only after noon, and not in the early morning at the onset of turbulence? Besides, the violin plots in figure 3 seem to show bimodal distributions for the Na fraction from fungal spores for each time of the day. What is the reason for that? We suggest that there may be residual particles from the night time emissions that can contribute to the sodium fraction from fungal spores beyond the onset turbulence. It is also possible that high relative humidity throughout the day may be responsible for the higher fraction of fungal spores in the late morning. We added the following senescence in the manuscript. "This may be the combined result of diurnal cycles in emissions, boundary-layer dynamics, and removal processes (e.g., precipitation)."

Supplementary
The bimodal distribution of the fractional contribution (as seen on the violin plots) is a reflection of the strong variability in simulated sea salt mass concentrations at the site. As a consequence of this variability, most of the time, simulated values of sea salt sodium mass are either much greater or much less than fungal spore sodium mass. Since fractional contributions are presented here, the values cluster in the upper and lower portions of the range 0-1.
8. L217: 'a major fraction to the sodium budget': please add here that this is for the wet season We clarified that the statement is for the wet season. "Remarkably, our experimental and modeling results demonstrate that fungal spores emitted from the rainforest biosphere contribute a major fraction to the sodium budget during wet season…" 9. L533-534: which size distribution was assumed for the spores? Within the range of the model coarse mode (1-10 micron), the assumed size can make large differences for the efficiency of the wet and dry removal We added the following lines to clarify the size distribution used in the model. "Upon emission both fungal spore mass and coarse mode number are increased, with the massto-number ratio for emissions determined by assuming that fungal spores have 4 μm diameter upon emission. The model's coarse mode particle number concentration also includes number contributions from other aerosol species, which are treated as internally mixed within the coarse mode, and the particle diameter is calculated dynamically by the aerosol microphysics routines from the total coarse mode aerosol mass and number, and a fixed geometric width." 10. Supp. Fig. 9: It seems that the same violin plot is show twice: the plot looks very similar to that in Fig. 3 and the title says 'wet season' while the figure is for the dry season. Please include the right plot for the right season. By mistake we placed the wet season plot. We incorporated the correct dry season plot in the revised manuscript.
11. Technical corrections/suggestions:-L32: pls remove 'we' Done. Reviewer #2 (Remarks to the Author): As I pointed out in my initial review, there are many non-marine sources of sodium salts, which the manuscript failed to discuss. It is frustrating to me that the authors acknowledge this fact in their response and yet refuse to improve the text accordingly. A number of relevant references are missing. In fact, the authors discuss a number of these in their rebuttal but don't include them in the revised text. I find the manuscript seriously lacking in this regard. It is a serious problem in the discussion and introduction. It would seem that the authors are concerned that once the proper references are cited, the novelty of their manuscript will fall beneath the high standards of Nature. Additionally, failure to acknowledge and discuss Na sources is a weakness in the estimates of how much salt in the Amazon comes from marine sources. I appreciate the effort in revising and adding the modeled estimates of sources. However, that section is very simplistic and I question the merit of conclusions drawn from it.
Also, the authors state in their response that other Na sources in aerosol are "completely different" because they are small sized aerosols, whereas fungal spores are in the coarse mode. This statement is contradictory to the discussion in the manuscript text that describes the bursting of fungal spores and generation of smaller aerosol as a major reason that fungal spores have climatic relevance.
What is known in the literature about the concentration of NaCl in spores? I suspect that a lot more work has been conducted that is not cited here.
The experimental details are sparse. My concern that the highly variable present of salt may simply be a function of operator error (and the occasional contamination of some samples through contain with human skin or contaminated (or re-used) laboratory gloves. I note that another reviewer requested additional details, and the authors chose not to include them, but rather to refer readers to a previous manuscript. I appreciate the limitation of space, but this does not help to eleviate my concern that the experimental results are artifacts rather than new findings.
Reviewer #3 (Remarks to the Author): The authors have appropriately addressed the comments of me and the other two reviewers. I only have one (very minor) comment on the revised version of the MS: in the main text (lines 236-256), the authors discuss the uncertainty in the 69% contribution of fungal spores to sodium mass, due to assumptions on spore sodium content and sea salt contribution. I would like to see this uncertainty reflected in the number that is given in the abstract, so could you please indicate the uncertainty range there (line 36)?
We thank the reviewers' for their comments and suggestions. In the following discussion, the reviewers' comments are in black-colored font, our responses are in blue-colored font, and changes in the manuscript are noted by italics blue-colored font.
Reviewer #2 (Remarks to the Author): As I pointed out in my initial review, there are many non-marine sources of sodium salts, which the manuscript failed to discuss. It is frustrating to me that the authors acknowledge this fact in their response and yet refuse to improve the text accordingly. A number of relevant references are missing. In fact, the authors discuss a number of these in their rebuttal but don't include them in the revised text. I find the manuscript seriously lacking in this regard. It is a serious problem in the discussion and introduction. It would seem that the authors are concerned that once the proper references are cited, the novelty of their manuscript will fall beneath the high standards of Nature. Additionally, failure to acknowledge and discuss Na sources is a weakness in the estimates of how much salt in the Amazon comes from marine sources. I appreciate the effort in revising and adding the modeled estimates of sources. However, that section is very simplistic and I question the merit of conclusions drawn from it. Also, the authors state in their response that other Na sources in aerosol are "completely different" because they are small sized aerosols, whereas fungal spores are in the coarse mode. This statement is contradictory to the discussion in the manuscript text that describes the bursting of fungal spores and generation of smaller aerosol as a major reason that fungal spores have climatic relevance. What is known in the literature about the concentration of NaCl in spores? I suspect that a lot more work has been conducted that is not cited here.
In response to the reviewer's request, we added relevant text in the introduction as noted below (lines: 58-68). Regarding the bursting of fungal spores and generation of smaller particles, we suggest that total contribution of sodium budget from fungal spores can be even lager because we did not take into account the smaller particles in this study. This study focuses solely on the coarse mode particles and investigates samples from stages 4 and 5 (size range: 1.0-3.2 µm) where the relative abundance of biological particles is high. We added the following text in the discussion (lines: 268-273) "Previous studies in different atmospheric environment show the presence of sodium in fine mode and suggest potassium as a tracer for anthropogenic sodium [1][2][3][4][5] . Observations of our study are fundamentally different. First, our study focuses on the coarse mode particles. Second, the potassium observed in our study mostly originates from biological particles 6 confirmed by their unique and distinguishable morphology. Those particles were especially abundant during the wet season when biomass burning is not a dominant source of particles in the basin." The experimental details are sparse. My concern that the highly variable present of salt may simply be a function of operator error (and the occasional contamination of some samples through contain with human skin or contaminated (or re-used) laboratory gloves. I note that another reviewer requested additional details, and the authors chose not to include them, but rather to refer readers to a previous manuscript. I appreciate the limitation of space, but this does not help to eleviate my concern that the experimental results are artifacts rather than new findings.
As we discussed in our revised manuscript, we suggest that variable present of sodium content may depend on different classes and genera of fungi. Sample collection and experiments were conducted with cautions. In our previous revised version as requested by the reviewer 3, we provided details of the method (please see the notes of the reviewer 3 who found our arguments convincing). To elaborate more on this note, we added the following text in the method section of the revised manuscript. The Amazon rainforest is often termed the "green ocean" because its relatively pristine 48 atmospheric conditions resemble those in remote marine regions interns of the low particle count  Fig. 4). Subsequently, when exposed to high humidity these hygroscopic 161 fragments grow and become supermicron 18 .   Fig. 13). The model suggests that 257 the fractional contribution of fungal spores to sodium will increase during nighttime and decrease 258 during daytime (Fig. 3b). This may be the combined result of diurnal cycles in emissions, 259 boundary-layer dynamics, and removal processes (e.g., precipitation).   The hygroscopicity of biological particles was investigated using a temperature-controlled 584 cooling stage in the environmental SEM to estimate the area growth factor (ratio of wet-to-dry 585 diameters) and a microreactor for in situ STXM hydration experiments 27 to estimate the mass 586 growth factor (ratio of wet-to-dry oxygen mass).

588
France) was used to perform imaging of Na+ ions. NanoSIMS provides high lateral resolution.

589
Samples were coated with a thin layer of Au to minimize charging and non-equilibrium sputtering  We thank the reviewers' for their comments and suggestions. In the following discussion, the reviewers' comments are in black-colored font, our responses are in blue-colored font, and changes in the manuscript are noted by italics blue-colored font.
Reviewer #2 (Remarks to the Author): As I pointed out in my initial review, there are many non-marine sources of sodium salts, which the manuscript failed to discuss. It is frustrating to me that the authors acknowledge this fact in their response and yet refuse to improve the text accordingly. A number of relevant references are missing. In fact, the authors discuss a number of these in their rebuttal but don't include them in the revised text. I find the manuscript seriously lacking in this regard. It is a serious problem in the discussion and introduction. It would seem that the authors are concerned that once the proper references are cited, the novelty of their manuscript will fall beneath the high standards of Nature. Additionally, failure to acknowledge and discuss Na sources is a weakness in the estimates of how much salt in the Amazon comes from marine sources. I appreciate the effort in revising and adding the modeled estimates of sources. However, that section is very simplistic and I question the merit of conclusions drawn from it. Also, the authors state in their response that other Na sources in aerosol are "completely different" because they are small sized aerosols, whereas fungal spores are in the coarse mode. This statement is contradictory to the discussion in the manuscript text that describes the bursting of fungal spores and generation of smaller aerosol as a major reason that fungal spores have climatic relevance. What is known in the literature about the concentration of NaCl in spores? I suspect that a lot more work has been conducted that is not cited here.
In response to the reviewer's request, we added relevant text in the introduction as noted below (lines: 58-68). Regarding the bursting of fungal spores and generation of smaller particles, we suggest that total contribution of sodium budget from fungal spores can be even lager because we didn't take into account the smaller particles in this study. This study focuses solely on the coarse mode particles and reported results from samples collected on stages 4 and 5 (size range: 1.0-3.2 µm) where the relative abundance of biological particles is high. The experimental details are sparse. My concern that the highly variable present of salt may simply be a function of operator error (and the occasional contamination of some samples through contain with human skin or contaminated (or re-used) laboratory gloves. I note that another reviewer requested additional details, and the authors chose not to include them, but rather to refer readers to a previous manuscript. I appreciate the limitation of space, but this does not help to eleviate my concern that the experimental results are artifacts rather than new findings.
As we discussed in our revised manuscript, we suggest that variable present of sodium content may depend on different classes and genera of fungi. Sample collection and experiments were conducted with cautions. In our previous revised version as requested by the reviewer 3, we provided details of the method (please see the notes of the reviewer 3 who found our arguments convincing). In addition to that, we provide following discussions, which suggest that presence of sodium salt in our sample indeed from spores rather than contamination.

i)
We routinely investigate quality of our blank substrates from different batches (total 5 in this case). SEM images at different field of view show that blank sample are with negligible contamination (see Supplementary Figure 14). We found 1 particle per 0.0005 cm 2 area, where as field collected particles, on average, consist of ~500 individual particles on stages 4 and 5. Particle loading is higher for smaller size particles (~2100 particles per 0.0005 cm 2 area on stages 6 and 7, see Supplementary  Figure 15a, b). ii) As we mentioned earlier, this study reports samples from stages 4 and 5 (size range: 1.0-3.2 µm). However, we also investigate particles from lower stages 6 and 7, size range: 0.32-1.0 µm) for a separate study. Fractions of Na-containing particles in stages 6 and 7 are significantly lower (Na-containing particle <15%) than particles on stages 4 and 5 (~50%). Please see Supplementary Figure 15c. This result suggest that high fraction of sodium-salt particles are indeed from spores that were deposited on stages 4 and 5. iii) Na-content in spores were investigated using different techniques (electron microscopy and Nano-secondary ion mass spectrometry) with different operator, sample preparation and sample handling procedures. Irrespective of that, both methods show presence of sodium in spores.
To elaborate more on this note, we added two new Supplementary Figures (14 and 15) and the following text (lines: 555-564) in the method section of the revised manuscript.  Figure 14). Investigation of particles from stages 6 and 7 (size range: 0.32-1.0 µm) showed Na-containing particles in stages 6 and 7 are significantly lower (Na-containing particle <15%) than particles on stages 4 and (Supplementary Figure 15). This result suggest that high fraction of sodium-salt particles are indeed from spores that were deposited on stages 4 and 5. Furthermore, sodium content in spores was observed irrespective of different sample preparation and techniques (e.g., SEM/EDX, STXM and NanoSIMS)"

Supplementary Figure 14. Investigation of particle contamination. SEM images show a) low magnification and b) high magnification image of blank sample; c) low magnification and d) high magnification image of collected sample.
a b c d The authors have appropriately addressed the comments of me and the other two reviewers. I only have one (very minor) comment on the revised version of the MS: in the main text (lines 236-256), the authors discuss the uncertainty in the 69% contribution of fungal spores to sodium mass, due to assumptions on spore sodium content and sea salt contribution. I would like to see this uncertainty reflected in the number that is given in the abstract, so could you please indicate the uncertainty range there (line 36)?
My recommendation is therefore to publish the paper after this minor revision.
As suggested, we provided the uncertainty range in the abstract as well.  The Amazon rainforest is often termed the "green ocean" because its relatively pristine 48 atmospheric conditions resemble those in remote marine regions interns of the low particle count particles 26 . For example, the particle in Fig. 2g shows a stronger sodium signal around the particle 147 boundary. That in Fig. 2h displays complex morphology and heterogeneity in its sodium spatial 148 distribution, while the distribution of sodium ions in the elongated fungal spore in Fig. 2i appears   149 homogeneous. These images suggest that sodium distributions within individual fungal spores is 150 diverse, presumably depends upon spore type, and may be influenced by the particle's origin and 151 aging history.
152 153 X-ray microanalysis indicates that the average percentage of total spores containing sodium 154 (minimum detection of 3 wt %) is higher above the canopy (daytime: 60%±6%; nighttime: 155 39%±3%) than below the canopy (daytime: 46%±4%; nighttime: 22%±2%). These differences 156 may partially be influenced by changes in physio-chemical properties of spores that transform 8 ( Supplementary Fig. 4). Subsequently, when exposed to high humidity these hygroscopic 161 fragments grow and become supermicron 18 .  Fig. 7). The rain records along backward trajectories also indicate that 208 multiple precipitation events occurred during transport, which would likely wash out most of the 209 sea salt particles before their arrival at the sampling site. Hence, these findings support the  sodium mass would range between 51-84%. We performed a sensitivity analysis by assuming 254 either 30% or 50% of fungal spores to be sodium-rich. When applying these different assumptions 255 to the model simulations, fungal spores account for 58% (21-90%) or 65% (27-93%) of the total 256 simulated sodium mass during the wet season ( Supplementary Fig. 13). The model suggests that 257 the fractional contribution of fungal spores to sodium will increase during nighttime and decrease 258 during daytime (Fig. 3b). This may be the combined result of diurnal cycles in emissions, 259 boundary-layer dynamics, and removal processes (e.g., precipitation). Fig. 3c presents a regional  20 . Fungal spores were assigned a material density of 1 g cm −3 , which is 640 commonly used in literature 13,16 . Fungal spores were assigned a hygroscopicity parameter (κ) of 0.1, similar to that of pollen grains 56 . The observed overall κ value for ambient below-canopy 642 aerosol in the Amazon is 0.22±0.05 in the accumulation mode, with an overall mean value of 643 κ=0.17±0.06 57 .
1. The major claim of the paper is that sodium salts present in atmospheric aerosols arise from fungal spores. Fungal spores represent a salt source that has previously not been reported. This is a fascinating source. However, the manuscript's assessment of the importance of this finding is not supported by previous literature or clear explanations. Regarding previous literature, the abstract states, "In sharp contrast with the assumption that sodium-containing aerosols arise solely from marine sources,..." This statement is simply not true. There is a wealth of papers on non-marine sources of sodium salts in aerosols.
We agree that fungal spores are a fascinating source of sodium salt particles in Amazonia. Sodium content in atmospheric aerosol is primarily attributed to sea water, especially the coarse mode particles. However, as the reviewer pointed out, there are also non-marine sources of sodium salts in atmospheric aerosols. For example, Ooki et al 1 observed the existence of anthropogenic sodium in fine mode aerosol particles in urban air. Ooki et al 1 suggested potassium as a tracer for anthropogenic sodium. Mamane 2 found presence of sodium and potassium in submicron particles from waste incinerators emissions. Another study conducted in the metropolitan area of São Paulo by Bourotte et al. 3 found K/Na ratio of 1.4 and 0.9 for coarse and fine particles respectively and they attributed this particles to waste incineration and vehicles. Most of these studies show the presence of sodium in fine mode and suggest potassium as a tracer for anthropogenic sodium.
Observations of our study are fundamentally different. First, our study focuses on coarse mode particles. Second, the potassium observed in our study mostly originates from biological particles 4 , especially during the wet season when biomass burning is not a dominant source in the basin.
To clarify these details, we modified the sentence and the manuscript (as suggested by Reviewer 3) to reflect that our main finding is that fungal spores contribute to sodium salt particles in the Amazon basin.
"In contrast to the common assumption that sodium-containing aerosols are primarily arise from marine sources, our results suggest that locally-emitted fungal spores containing sodium contribute substantially to the number and mass of coarse particles containing Na and Cl (similar to sea salt)." We adopted the title proposed by reviewer 3.

"Fungal spores as a source of sodium salt particles in the Amazon basin"
2. More problematically, while the observation of salts arises in fungal spores in interesting in its own right, the manuscript in its present form does not provide an adequate explanation or even compelling suggestions for how/when/under what circumstances or via what mechanism the salt in incorporated into some, but not all, spores. If an insightful discussion were included, I would rate this manuscript much more positively. As written, it is lacking.
In the original manuscript we provided basic information (lines: 167-172) regarding the nature for sodium content in fungal spores. To address the reviewers concerns, in the revised manuscript we provided more detailed discussion on this topic.
The following text was added in the revised manuscript (lines: 172-190).
"The presence of sodium is common in so called halophilic fungi which cannot grow without NaCl 5 . Active discharge, uptake, and efflux processes are likely responsible for the sodium content in the spores and sodium content may vary with different classes and genera of fungi. During active discharge, fungi forcibly eject spores into the atmosphere, together with osmotic fluid containing hexoses, mannitol, phosphate, sodium, and potassium 4,6 . Fungi require K + for electrical and osmotic equilibria of the cells and presence of K + in several fungal spores is well understood 7 . Previous studies suggested that K + can be partially replaced by Na + and Na + can enhance the growth of fungi 8 and plants 9 under K + deficiency conditions. The growth of fungi and uptake of Na + varies with their physiological conditions and uptake rates depend on the species and their genes 10 . For example, specific genes (e.g., acu1 and acu2) are responsible for highaffinity Na + uptake of Ustilago maydis. When fungi contain excess Na + , they activate Na + -efflux ATPase (Adenosinetriphosphatase), which acts as a key enzyme for the biological evolution of fungi 11 . We suggest that uptake and efflux of Na + varies with different classes and genera of fungal community. For example, previous studies in Amazonia shows several genera (e.g., Agaricus; Amanita; Aspergillus; Boletus; Cladonia; Mortierella; Puccinia; Lepsita; and Rhizopus) within one class of fungi (Lecanonomycetes) 12,13 . Furthermore, the transpiration of plants 14 and nutrient uptake 15 can also influence the sodium content of spores. Further studies are needed to better comprehend the sodium content in fungal spores by linking chemical composition, molecular biology and diversity of fungal spores."

3.
A second finding is that spores which contain salt are more hygroscopic than those which do not. While I'm not aware of any previous results on precisely this topic, it is very well known that salts are deliquescent materials whereas organic macromolecules and other organic compounds found in spores are not. So, this result is very obvious.
Indeed, it is expected that sodium-containing spores are more hygroscopic that those without sodium. Even if the growth factors are similar for sodium chloride particles and sodium containing spores, it is interesting to note that growth factors of NaCl and sodium-containing spores are different at relative humidity between 70 and 96%. Notably, variations in the hygroscopicity are associated with variations in the chemistry of these spores. We believe this finding is important and worth discussing in the manuscript.
A minor point -4. According to Figure 3, the Na fraction for fungal spores is estimated to be between 0.1 and 1.0 (i.e. nearly any value in the early hours, and later in the day transitions to an even wider spread in possible fractions. I may be missing something, but I don't find this result to be a significant finding. We believe Figure 3b provides important contextual information to help readers understand the temporal variability in the simulated fungal spore contribution to sodium, near the measurement site. Considering the high temporal variability of aerosol particle concentrations, including the simulated sea salt and fungal spores in this study (Supplementary Figure 11), it is not surprising that the relative contribution of fungal spores to total sodium mass spans a wide range of values. Figure 3b shows that from midnight to noon, the wet season median contribution of fungal spores to Na ranges from about 0.5 to 0.6. This implies that fungal spores are the dominant contributors of Na on the majority of days at these times. By contrast, from 3 pm to 9 pm, however, the median ranges from 0.15 to 0.3, and the dry season median fungal spore fraction ranges between only about 0.1 and 0.2 (Supplementary Figure 11). In other words, the model helps contextualize the limited number of observations, by identifying the circumstances under which the fungal spores are predicted to contribute the majority to simulated boundary-layer Na at the site: specifically, during the late night and early morning hours of the wet season.
5. In summary, the authors have one potentially interesting surprising field observation in the Amazon. However, additional consideration of how this may arise is needed before I could recommend publication.
We appreciate the comments and suggestions provided by the reviewer. As outlined in the response to the comments 2, in the revised manuscript we expanded the discussion substantially regarding the presence of sodium in fungal spores. In addition, the observation of sodium in fungal spores is on samples from a range of different days.
Reviewer #3 (Remarks to the Author): China et al. present new multi-instrument observations of the sodium content of fungal spores in the Amazon basin, and use those in combination with backward trajectories and earth system model simulations to suggest that the fungal spores are the main source of sodium containing particles in the wet season.
The major claims of the paper are as follows: 1. Fungal spores contribute significantly to particle-sodium in the Amazon, contributing at least 30% by number of sodium salt containing particles 2. Sodium-containing fungal spores show higher hygroscopic growth than sodium-free spores 3. Fungal spores account for ~62% of sodium mass during the wet season, according to model simulations 4. Therefore, the role of fungal spores in cloud formation and salt cycles in the Amazon deserves further consideration The claims in the MS are significant and novel. The two closest papers known to me, one from (partly) the same research group, discuss that 1) K-rich bio-aerosols are seeds for SOA formation, but the paper does not mention Na-containing particles (Poehlker et al. 2010) and 2) that when spores rupture at high RH, fungal fragments that contain Na salts are released, but the paper does not quantify Na-content of the particles nor their contribution to the particle Na-budget over Amazonia (China et al. 2016).
The results are of interest to others in the community, since they give 1) new insight into Nabudget of the Amazon from a source that was not considered previously and 2) they add a new line of thought to the debate on sources of cloud-active particles in Amazon, which is a long-standing problem.
The results are of interest to the wider field, because this work adds a new aspect to the understanding of biogeochemical cycles in the Amazon rainforest, and how the forest itself sustains biogeochemical and hydrological recycling. Especially the role of fungal spores therein is one that has not received much attention so far, and this work is likely to start a further debate on this topic.
We appreciate the reviewers' careful reading of our manuscript and providing us positive feedbacks.
The MS is clearly structured and generally convincing, but I think it could profit from some further clarifications: 1) Since I am not familiar with the instruments that are used, I cannot judge the quality of the experimental work. However, I think the MS would benefit from a short explanation of the sampling procedure: are all coarse particles identified that were sampled? How is it determined that a particle is a fungal spore and not another kind of PBA (e.g. bacterium or pollen)? Is this done based on size, morphology or otherwise? A brief description could be added, for instance in L107.
Initially, due to space constraints, we couldn't provide details on the methods used for identifying the fungal spores in the manuscript. Typically fungal spores are 1-6µm, while pollen grains are larger in size (5−150 µm), bacteria are smaller (<1µm) and typically elongated in shape. As we selected only impactor stages 4 and 5 (aerodynamic diameter range: 1.0-3.2 μm) for this study, the fraction of bacteria and pollen is almost negligible. We also refer to our previous publication (China et al., 2016) where we described identification of fungal spores.
To address reviewer's comment, we moved portion of text from the methods section to the main manuscript and provided additional explanations "The fraction of fungal spores was quantified based on their unique characteristic morphologies (spherical, rod-like or spheroidal in shape), size (1-6 µm), and chemical composition (mostly carbonaceous and containing phosphorous) by electron microscopy imaging and X-ray microanalysis of over 3500 individual particles. Details of the fungal spore identification method are provided elsewhere 16 ." We added the following text in the method section "The computer-controlled CCSEM/EDX analysis automatically investigates the specified scanning area and detects particles in the specified fields of view. In this way, we can detect ~90% of the particles deposited on the grid. Particles collected on the edges (~5-10%) of the Cu mesh were excluded from the results based on their high Cu background signal. We analyzed ~80% of the particles that were collected onto TEM B-film grids placed on stages 4 and 5 of the impactor." 2. The estimate of the contribution of spores to the Na-budget, depends on two factors which are very uncertain: local emissions of fungal spores and transport of sea salt aerosol into the Amazon basin. I think the fungal spores emissions are constrained reasonably well by the observations of their concentration, which the model seems to underestimate somewhat. However, if global emission estimates of sea salt span 2 orders of magnitude, how reliable is the estimate of sea salt aerosol transported into the Amazon basin, with no direct observations available for validation? Can the authors provide an estimate of the uncertainty in the sea salt contribution to the particle sodium budget? This could provide a bit more context to the number of 62% of total sodium mass that is now assigned to fungal spores.
The reviewer is correct that the model emission estimates of the sea salt are uncertain.
To address this, we compared model estimates of atmospheric sea salt concentration with observed values at few coastal and remote island sites to investigate the fidelity of the model estimates of sodium from sea-salt. We have added a scatterplot comparing observed sodium mass concentration with model-simulated mass sodium mass concentration from sea-salt, together with a global map (Supplementary figure 9) of measurement sites.
We added the comparison of model-simulated and measured sodium mass (Supplementary figure  10) and added the following text.

Reply to the Reviewer's comments (NCOMMS-17-33311A)
Reviewer #2 (Remarks to the Author): As I pointed out in my initial review, there are many non-marine sources of sodium salts, which the manuscript failed to discuss. It is frustrating to me that the authors acknowledge this fact in their response and yet refuse to improve the text accordingly. A number of relevant references are missing. In fact, the authors discuss a number of these in their rebuttal but don't include them in the revised text. I find the manuscript seriously lacking in this regard. It is a serious problem in the discussion and introduction. It would seem that the authors are concerned that once the proper references are cited, the novelty of their manuscript will fall beneath the high standards of Nature. Additionally, failure to acknowledge and discuss Na sources is a weakness in the estimates of how much salt in the Amazon comes from marine sources. I appreciate the effort in revising and adding the modeled estimates of sources. However, that section is very simplistic and I question the merit of conclusions drawn from it. Also, the authors state in their response that other Na sources in aerosol are "completely different" because they are small sized aerosols, whereas fungal spores are in the coarse mode. This statement is contradictory to the discussion in the manuscript text that describes the bursting of fungal spores and generation of smaller aerosol as a major reason that fungal spores have climatic relevance. What is known in the literature about the concentration of NaCl in spores? I suspect that a lot more work has been conducted that is not cited here.
In response to the reviewer's request, we added relevant text in the introduction as noted below (lines: 58-68). Regarding the bursting of fungal spores and generation of smaller particles, we suggest that total contribution of sodium budget from fungal spores can be even lager because we didn't take into account the smaller particles in this study. This study focuses solely on the coarse mode particles and reported results from samples collected on stages 4 and 5 (size range: 1.0-3.2 µm) where the relative abundance of biological particles is high. The experimental details are sparse. My concern that the highly variable present of salt may simply be a function of operator error (and the occasional contamination of some samples through contain with human skin or contaminated (or re-used) laboratory gloves. I note that another reviewer requested additional details, and the authors chose not to include them, but rather to refer readers to a previous manuscript. I appreciate the limitation of space, but this does not help to eleviate my concern that the experimental results are artifacts rather than new findings.
As we discussed in our revised manuscript, we suggest that variable present of sodium content may depend on different classes and genera of fungi. Sample collection and experiments were conducted with cautions. In our previous revised version as requested by the reviewer 3, we provided details of the method (please see the notes of the reviewer 3 who found our arguments convincing). In addition to that, we provide following discussions, which suggest that presence of sodium salt in our sample indeed from spores rather than contamination.

i)
We routinely investigate quality of our blank substrates from different batches (total 5 in this case). SEM images at different field of view show that blank sample are with negligible contamination (see Supplementary Figure 14). We found 1 particle per 0.0005 cm 2 area, where as field collected particles, on average, consist of ~500 individual particles on stages 4 and 5. Particle loading is higher for smaller size particles (~2100 particles per 0.0005 cm 2 area on stages 6 and 7, see Supplementary  Figure 15a, b). ii) As we mentioned earlier, this study reports samples from stages 4 and 5 (size range: 1.0-3.2 µm). However, we also investigate particles from lower stages 6 and 7, size range: 0.32-1.0 µm) for a separate study. Fractions of Na-containing particles in stages 6 and 7 are significantly lower (Na-containing particle <15%) than particles on stages 4 and 5 (~50%). Please see Supplementary Figure 15c. This result suggest that high fraction of sodium-salt particles are indeed from spores that were deposited on stages 4 and 5. iii) Na-content in spores were investigated using different techniques (electron microscopy and Nano-secondary ion mass spectrometry) with different operator, sample preparation and sample handling procedures. Irrespective of that, both methods show presence of sodium in spores.
To elaborate more on this note, we added two new Supplementary Figures (14 and 15) and the following text (lines: 555-564) in the method section of the revised manuscript.  Figure 14). Investigation of particles from stages 6 and 7 (size range: 0.32-1.0 µm) showed Na-containing particles in stages 6 and 7 are significantly lower (Na-containing particle <15%) than particles on stages 4 and (Supplementary Figure 15). This result suggest that high fraction of sodium-salt particles are indeed from spores that were deposited on stages 4 and 5. Furthermore, sodium content in spores was observed irrespective of different sample preparation and techniques (e.g., SEM/EDX, STXM and NanoSIMS)" My recommendation is therefore to publish the paper after this minor revision.
As suggested, we provided the uncertainty range in the abstract as well.
"Modeling results suggest that fungal spores account for ~69% (31-95%) of the total sodium mass during the wet season and that their fractional contribution increases during nighttime."