Response to comments from Anonymous Referee #1

Overview of Anonymous Referee #1:

This paper describes a unique dataset of water levels, chemistry, and isotopic composition of streamwater across a tropical catchment. The sampling design was such that it included intermittent and perennial streams. The results highlight the important role of geology and thereby subsurface storage capacity in determining which streams are intermittent and perennial. The differences in flow pathways in turn affect the stream chemistry.

The dataset is unique and the paper provides clear evidence for the role of geology in determining the variation in runoff responses and chemistry across the catchment, improving not only our knowledge of intermittent streams but of catchment functioning in general. The figures are all very clear and the paper is logically structured. At some places the writing can be a bit clearer or shorter (suggestions are given in the annotated pdf) but overall the text is very clear.

In my opinion the paper puts a bit too much emphasis (in its writing) on understanding intermittent streams as only some of the streams are intermittent and there is knowledge gained for the whole catchment. In other words, it is not wrong to focus so much on the intermittent streams but the paper actually provides more information than just for intermittent streams. This can however easily be solved with some rewording. Having said that, I would prefer to see an extra table regarding the statistical differences in the chemistry for the intermittent and perennial streams for the different sampling dates. Now only the correlations with land cover and geology are given.

Reply: We appreciate the reviewer’s general perspective on the value of our work and thank him/her for the valuable feedback. In the revised manuscript we will address each of his/her observations/suggestions to improve quality of the manuscript, including a table or figure of the statistical differences of the chemical data. We will also implement the suggested changes in the annotated pdf file to improve the readability of the manuscript.

 Specific comments:

L98: I think that it would be useful to add a specific hypothesis here (or some research questions). If land use has the largest effect on streamflow generation and land degradation leads to more overland flow, one would expect that flow paths are shallower, leading to less recharge and faster responses and recessions and less perennial flow for the lower parts of the catchment. If geology is the dominant factor, then soil depth and storage would determine which streams respond first and have faster recessions – which is indeed what the results show. The intro could highlight the effects of land degradation and soil depth/storage capacity on the runoff response a bit more by contrasting them and referring to some studies that have specifically looked at these factors.

Reply: We agree. We will add research questions or hypotheses regarding the effects of land degradation and geology on streamflow generation. We will also highlight the effect of land degradation and soil depth/storage capacity on the runoff response in the introduction.

Section 2: Some important information about the catchment is missing, such as the annual precipitation, temperature (and potential ET). Also, it would be good to already mention here the differences in soil depth and permeability across the catchment

Reply: Unfortunately, no climatological information was available during the study period in the study area. We will add information on the hydroclimatological conditions of the region using historical data from one monitoring station of the Ecuadorian National Agency of Hydrology and Climate (INAMHI) located at approximately 20 km from the study site to provide a general description of the climatology of the region. Also, no quantitative information about soil depth and permeability across the whole catchment is available. The study area is remote with limited access due to the steep topography limiting the chance to characterize these properties across the whole catchment. We will thus report further details of the soil information available from the Ecuadorian Ministry of Agriculture and Livestock that is reported in Figure 1c.

Section 3.4: The analysis method for COD needs to be described here as well.

Reply: We will add information on the COD analysis method.

L277: This sounds like the perennial streams also went dry. Is that correct? Then they are not perennial. This is a bit confusing.

Reply: Perennial streams did not dry. The sentence will be rephrased accordingly.

L283: Provide some information on the size and intensity of the selected events. Are they similar for the wet and dry period example?

Reply: Unfortunately, this is not possible as there were no rainfall data available during the study period. We report the dynamic of streamflow response during the wet and dry periods when such response was simultaneously detected at both intermittent and perennial streams. We will clarify this in the methods section.

Section 4.1: I would make it much clearer that the results from WETSPRO are based on graphical hydrograph analyses. That way it clearly distinguishes them from the isotope or tracer based inferences of flow pathways.

Reply: We agree. This will be highlighted in the revised manuscript.

L319-321: Is this difference statistically significant?

Reply: This will be statistically tested and reported.

L379: I fully understand the need to focus on a few solutes but it is not clear to me why you used P as an indicator of agriculture. Are there high additions of P in these agricultural systems? Provide a bit more background information on this. Why not base the selection of the solutes on the test for which solutes the differences between catchments with different land uses are statistically significant or the correlation between % agriculture and concentrations (i.e., Table 3)?

Reply: P is an indicator of the type of fertilizers used for agricultural practices in the study region, and this is the reason why we used it as an indicator of anthropogenic activities in the catchments. We will provide a detailed explanation about the type of fertilizers used in the area to justify the use of this solute in our analysis. We did not base the selection of the solutes on statistical significance between the solutes’ concentrations and the catchments’ features as no statistically significant correlations were found for most solutes. The statistical significance of the correlation analysis is rather used to assess whether there is a relation between the solutes concentrations and the factors influencing the catchments’ hydrological behavior. 

Figure 6: Perhaps it is useful to use two different colors to indicate which sites are intermittent and perennial. Overall, I would like to see a bit more information on the (statistical) differences in chemistry between the intermittent and perennial streams as this is a unique part of the dataset. Perhaps add information on the significance of the differences in concentrations for intermittent and perennial streams in Table 3 or add another table where you show this for each sampling date.

Reply: We agree it is useful to indicate which sites are intermittent and perennial in the figure and will do so in the revised manuscript. We will also carry out a statistical test to assess the differences of the water chemistry between intermittent and perennial catchments statistically.

Figure 7: I don't think that you need the 3D surface plot, you can just color-code the points in b and c according to Viche formation and elevation, respectively. If you want to keep the 3D surface plot, then at least show the points as well. Also, why not use a multiple linear model to look at the combined effect of the geology and elevation (as they are somewhat correlated). And why not show the same results/graphs for the other two chosen solutes a well?

Reply: We prefer to keep the surface plot in figure 7 as we believe it clearly depicts the relation between the Ca concentration, geology, and elevation of the catchments. However, we will plot the intermittent and perennial sites for clarity. We did not use a multiple linear model as we found a high statistically significant correlation between geology and elevation (section 4.4) that could cause a model overfitting issue (Lin et al., 2011). Also, we do not show the scatter plot of the other chosen solutes as correlation between them and the features of the landscape are low and non-statistically significant as shown in Table 3.

Section 4.4. The effects of land use and soil depth on the concentrations are not so well described and discussed. Also, it would be very useful to mention the correlation between geology and mean catchment elevation.

Reply: We will mention that the correlation between the solutes’ concentrations and land use and soil types were low and non-statistically significant in the revised manuscript for completeness. The correlation between geology and catchment elevation was reported in lines 398 and 401 of the manuscript, and the correlation (with its statistical significance) among the different features of the catchments will be reported as supplementary material.

L434-443: This is a bit of a selective comparison. I don’t think that it was the ET itself that was the cause for the differences. This study (figure 2) and the study by Zimmer and McGlynn both show that antecedent conditions are important. In the study by Zimmer and McGlynn the differences in antecedent conditions are due to seasonal changes in ET, whereas in your catchments they are likely mainly due to seasonal differences in precipitation. Also, the fact that there is no evaporative fractionation for the water in the stream does not mean that there is no effect of ET on the antecedent conditions and the relative importance of different runoff processes. Therefore, this section requires some rewording. Also, L441-453 are interesting but this is certainly not the main and most important outcome of the study. I am not sure that you need this text and think that it distracts from the main results (especially when it is the first thing in the discussion).

Reply: We thank the reviewer for pointing this out. We agree that antecedent conditions are important in our study as in the referred paper (Zimmer & McGlynn, 2017) but because of different factors. We will update this in the revised manuscript. Also, we agree that L441-453 are not relevant for the study and will remove it from the discussion as suggested.

L465-470: You may want to highlight that in addition to contact time, there could also be a difference in the reactivity of the material of the two formations.

Reply: We will add the potential influence of the reactivity of the material.

L533: Mention this already in section 5.3.

Reply: We will mention this in section 5.3 as suggested.

REFERENCES WE CITED

Lin, D., Foster, D. P., & Ungar, L. H. (2011). VIF Regression: A Fast Regression Algorithm for Large Data on JSTOR. J. Am. Stat. Assoc., 106(493), 232–247. https://doi.org/https://doi.org/10.1198/ jasa.2011.tm10113

Zimmer, M. A., & McGlynn, B. L. (2017). Ephemeral and intermittent runoff generation processes in a low relief, highly weathered catchment. Water Resources Research, 53(8), 7055–7077. https://doi.org/10.1002/2016WR019742

Response to comments from Anonymous Referee #2

Overview of Anonymous Referee #2:

This is a well written and timely manuscript looking at the various controls of streamflow generation in a tropical catchment. The work on the intermittent streams is particularly important as this is an area of need, especially in the tropics where work is limited. The authors have put in good work and I only have some minor suggestions. 

Reply: We appreciate the reviewer’s general impression on the value of our manuscript. We have updated the manuscript considering his/her valuable feedback. Below we addressed all the reviewer’s comments and suggestions.

L113-114: It will be good to include the long term average rainfall totals for each period. 

Reply: Unfortunately, there is no rainfall data for the study period. We will use historical data from one monitoring station of the Ecuadorian National Agency of Hydrology and Climate (INAMHI) located at approximately 20 km from the study site to provide a general description of the climatology of the region.

L125: Please include a reference for the soil types. 

Reply: We will include a reference as suggested.

L284-285: Please include a table showing the response characteristics. This can be included as a supplementary file. 

Reply: The information was presented in the boxes of figure 2 in the manuscript of the original submission. We consider it is useful to keep it there for completeness and will specify this in the figure caption.

L300: This section is well written. The authors do a good job of showing the isotopic differences based on seasonal changes. However, I think that it will benefit the reader if we could see this isotopic evolution for a perennial vs intermittent stream. It will be good to see if there are major differences particularly in the transition periods. 

Reply: We will show the evolution of the isotopic composition of perennial and intermittent streams to assess the temporal variability over the monitoring period.

Figure 6: Can you indicate the perennial vs intermittent streams on this? I think it will also be good to remind readers that the numbers go from headwaters to the main outlet. Why was the Ca not plot as ppm like in all other figures. 

Reply: We will show the perennial and intermittent sites in the figure and specify the order of the catchments in the figure’s caption. We will update the figure units to ppm.

Table 3: blue are positive correlations and red are negative. Please indicate this so that it is easier to recognize from the table.

Reply: We will specify this in the revised manuscript.

L454-455: In this section is it a bit unclear how the subsurface features affect the streamflow dynamics given that some of the gauges were located in different altitude and land use covers. This is especially true for the Viche formation which only had one intermittent stream which makes it difficult to compare to the other intermittent streams of the other geological formation.

Our analysis is based on a nested monitoring system approach given that our objective is to identify general patterns of the hydrological response of intermittent and perennial catchments, rather than using a comparison of paired catchments with very specific conditions that could obscure such patterns. Therefore, we consider that including catchments with variable size, mean elevation, and landscape features is an advantage of our approach, rather than a weakness. Here, we would also like to highlight that the use of the presented multimethod approach has the advantage of avoiding subjectivity when inferring the hydrological behavior of the catchments as opposed to relying on a single source of information (i.e., stable isotopes vs. geochemical data vs. hydrometric data) that could lead to high uncertainties in the identification of the mechanistic understanding of streamflow generation. Because of this, we consider that our study should remain focused on the presented spatially distributed measurements across spatial scales and landscape characteristics, rather than assessing specific differences among the catchments.

L491: Were there any perennial streams in forested areas? If so what was the geochemistry like when compared to the intermittent streams? 

Reply: We will assess and report potential differences in water chemistry in forested intermittent versus perennial streams, if any. 

Overall I would like to see a bit more of a comparison of the one intermittent stream located in the Viche geology when compared to the other locations. 

Reply: We will carry out this assessment, and report and discussed relevant findings; although the analysis will still be focused on the identification of general patterns because of the reasons specified above.