Reviewers Comments Are in Italic (And Bold Sometimes) and My Responses Are in Regular Font

Reviewers’ comments are in Italic (and bold sometimes) and my responses are in regular font.

Reviewer #2: This paper studies mobilization of arsenic, lead and mercury under the conditions of seawater intrusion by conducting laboratory experiments and analysis of the some literature data. The experimental method is suitable for studying these phenomena, although the system and the results should be better described. The manuscript is written with a solid level of English; however, it could be better organized in terms of idea and especially the relationship between the studies conducted in this paper and the previous studies. Although this contribution is a useful experimental contribution to the important subject of water contamination by heavy metals, I have a number of points that need to be addressed before it is recommended for publication:

We reorganized some of the text. Now it starts with the experiment set-up, data results, then coastal salt water data, stream and soil solution data results from our field study, and the paper ends with the discussion. Hopefully with these changes the paper should be more readable. that helps.

1) The main novelty in this paper is the experiment in laboratory columns. I suggest this to be emphasized from the beginning, rather than starting with examples from data analyses from other contributions.

We adapted the suggestion. See changes made on p.3-p.7.

2) p2, 5-7 While I agree that this study may shed more light on the mechanisms of mobilizsation of heavy metals in coastal aquifers the authors should refrain from a priori making conclusions as a matter of certainty when it comes to exposing their views on this manuscript.

We adopted took partly of the suggestions and deleted three short paragraphs regarding the chemical properties of As, Pb, and Hg in the introduction section. Some sentences in other paragraphs in the introduction were reworded as well.

See changes in marked text on p.2-3.

3) p4, 54-55 The original data by Grassi and Netti (2000) plots Hg versus Cl-, this should be explained in the text to avoid ambiguity.

The original sentences were changed to the underneath sentences. See p.10 , line 219- 220 on clear text). Notice the high-lighted words used “collected“ and “their data”. It should be clear now.

1 “Here we also analyzed the geochemical data collected by Grassi and Netti (2000) for Ansedonia and Castiglione in the Pescaia region in Italy. Their data showed a clear relationship between concentrations of Hg and Cl (Figure 6)”.

4) sections 3.1 and 3.2 should have a short explanation on the purpose of the column experiments and why the particular methodology and experimental sequence is used.

Two short paragraphs were added, as suggested. Here are the added text on p.3, line 73-75, right after the bold section 2 title.

“The salt injection batch experiment with regular and arsenopyrite and galena enhanced soil columns under controlled condition can help to better understand the relations between concentrations of As, Pb, and Hg and varied salt environments”.

5) Porosity and permeability of the columns used for measurements should be reported. Can you clarify flow rates used in the experiments? Porosity, permeability, and flow rates are added in the text on p.3, line 82-83, and line 80-85 in the clear version of the text.

“Four columns with a diameter of 5.08 cm (2 inches) and height of 30.5 cm each were filled with about 480 grams of loamy soil collected from A and upper B horizons at the Centennial Lake Watershed (CLW), New Jersey (Figure 1). The soils used in the four columns have an average porosity of 0.66, an average hydraulic conductivity of 1.57 m/day, average pore volume of 298 cm3, and an average organic matter content of 4.2% based on Loss On Ignition method.”.

6) p5, 30-33 "After the injection, low-ionic-strength water was added once or twice daily (an average of about 300 milliliter daily) with no overflow allowed."

I assume the reviewer did not like the term “low-ionic-strength water”. We used distilled water, which has a low ionic strength, as the background solution. “Distilled water” replaced “low-ionic-strength water” throughout the text now. See line 89 on p.4.

7) This experimental sequence should be given a reasoning.

I assume the reviewer here is referring to Section 2.2, “CaCl2 and NaCl mixed salt experiment” on p.5 here. See line 155-159 on p.5 for the added text which is given underneath as well.

2 “CaCl2 is a common sea salt in a salty coastal aquifer and the second most common deicing salt in a salted watershed. Because mixed CaCl2 and NaCl salts have been promoted as a more environmental friendly deicing combination and have gained popularity in recent years (Breault and Smith, 2010), their effect on the mobilization of the As, Pb and Hg should be examined separately from that of NaCl salt alone”.

8) Figures 3 to 7- it would be more readable if you have x-axis in days since the start of the experiment / start of sampling.

The x-axis label in diagrams was changed to days as suggested.

9) Arsenopyrite and galena are used in mixed soil columns to mimic the elevated sources of As and Pb. How exactly were these minerals positioned in the column?

One sentence was added in to address this at line 88-89 on p. 3-4 in the clear version. “The mineral powder layers were placed at about 1/5 from the bottom of the columns (at ~6 cm in a 30.5 cm height column)”.

10) The results from Figure 4 imply late elevated concentrations of Hg , As and Pb. In the data interpretation the explanations offered include increased dispersion of organic matter, complexation with Cl, and an increased reducing environment after prolonged water flushing. To support your explanations can you: a) provide chemical formulae of arsenopyrite and galena “Arsenopyrite (FeAsS) and galena (PbS) minerals “ were added in on line 63, p.3 in the clear text. b) discuss the dispersion in terms of some estimated dispersion characteristics in the soil colummns (e.g. dispersivity).

Actually, the dispersion here is not the transport dispersion we commonly use in the transport modeling. The highlighted lines are added to the new text, line 123-137 on p. 5, to make it clear.

“Flushing of the soil columns with distilled water, following the injection of NaCl, leached soluble salts from the soil column and created a sodic soil condition in which soil aggregates and soil organic matter breaks up. This break-up of soil particles under sodic condition is called the soil dispersion (defluocculation) (Brady and Weil 2008). A sodic soil is a soil condition when

3 sodium ions are disproportionately high and sodium adsorption ratios (SAR) is larger than 13 (Brady and Weil 2008). Swarms of loosely absorbed hydrated Na+ surround the outer-sphere complexes of soil colloids and the poorly balanced electronegative colloids repel each other under a sodic soil condition. These processes cause the soil particles to disperse, increase the reaction surface area, and accelerate the ion desorption and adsorption in soil and aquifer (Brady and Weil 2008). The SAR values in the leachate were higher than 70 after the first day of the injection and were higher than 13 throughout the experiment time until the last day (Figure 3). Therefore, the soil sodic condition exists through most of this NaCl injection experiment”.

Brady and Weil’s text also has a nice picture on the sodic soil dispersion on their p.423.

The sodium adsorption ratio is calculated as: Na molar concentration/sqrt (0.5 Ca +0.5Mg molar concentration) (P.418, Brady and Weil, the Nature and Properties of Soil, others put sodic condition at SAR =15).

In our runoff collection from highway (15 feet zone), the peak SAR was about 237. They were above 14 most of the winter/spring time (Table 2).

On p.9, line 236 to 244 in the clear version, regarding the dispersion in the soil along the highway from runoff water, these lines were added into the text.

“Overall, the peaked Na-concentration was more than 266 times (6678 ppm) those of the pre- salting Na-concentration (25.1 ppm) (Table 2). The SAR was as high as 237 on 2/13/2013 in transect 1 and was larger than 13 in all 18 samples collected in February 2013. This disproportional high sodium concentration created the field sodic condition and caused soil dispersion. Based on Sun et al. (2009, their Table 5), soil organic matter contents in top soil at the studied I-95 transect ranged from 2.11% to 4.08% from the loss on ignition method. Therefore, dispersion of soil organic matter exists as well. The large surface area of dispersed soil and soil organic colloids increases the reaction kinetics and accelerate the adsorption and desorption processes.”

We also added a SAR plot of the three injections in Figure 3 and a SAR column in Table 2 for runoff off I-95.

Hope that helps.

11) One of the main conclusions is that concentrations of both Pb and Hg were positively correlated with the measured concentrations of Na and Cl - this was interpreted to be most likely due to chloride complexation of Pb and Hg. Can this be experimentally proved? Good points.

4 Two pe-pH figures (Figure 12,14) and pe-pH data were added in to show the stability ranges of the As, Pb, and Hg. The solid boundary shows where Pb2+ could be stable. It exists in water either as PbCln or Pb(OH)2 (Figure 12). In a Pb-Cl-H2O system, those are the most possible species. With increased Cl species, equilibrium reaction will tend to go the PbCl+ complexation.

For Hg, from Figure 14, either various species of HgnCln or Hg(OH)2 in the Hg-Cl-S-H2O system are its most possible species. The pe-pH data shows they are probably mixed. PhreeqC database has more of these reactions listed.

I added a few lines regarding the stability of the As species in the text based on the pe-pH diagram of Figure 12 on p.11, line 310-316.

“Arsenic in natural water occurs mainly as oxyanions in the form of trivalent arsenite (As(III)) and pentavalent arsenate (As(V)) with As(V) dominating (Smedley and Kinniburgh 2002) based on its pH-pe stability diagram (Figure 12)”.

- 2- H2AsO4 and HAsO4 both have As(V).

Figure 12. Simplified pe-pH diagram for the system of As-S-O-H2O (dash boundary lines) and Pb-Cl-OH (Solid boundary lines) at 25oC and one atmospheric pressure for the estimated pe-pH data from the stream water of DRT gage station, the creek water and soil solution in CLW, and groundwater data from Bangladesh by Reza et al (2010b).

There are a few modified sentences in the text as well, p.15, line 406-412.

“Hg exists in water as Hgo, or Hg2+, and 2+ Hg2 and has a strong tendency to form complexes with Cl-, OH-, S2-, and S- containing functional groups of organic ligands, particularly, in a surface water body or shallow ground water as seen in the pe-pH diagram of Hg-Cl-S-H2O (Figure 14, right)”.

5 Minor points:

P2, 18 "As (Arsenic), Pb (lead), and Hg (Mercury) " should be reversed with As, Pb and Hg in the brackets. Changed P2, 18 should read as "toxic metals". Changed.

6 Reviewer #3: General Comments Reference mentioned are listed in the reference section of the text. Texts in italic are reviewer’s comments. Texts in regular font are my response. Quoted texts are copied from the current paper.

The authors examined the aqueous geochemical conditions that mobilize As, Pb, and Hg in one coastal area and two salted watershed; presented the results of a column study that released As, Pb, and Hg from soil by salt injection; and evaluated As, Pb, and Hg chemistry in a watershed with large deicing salt application. I have major concerns about the writing, experimental design, data presentation, statistical analysis, and result discussion.

1. Overall, this manuscript is very hard to read. The manuscript did not follow the article structure of the Journal of Contaminant Hydrology, i.e. Introduction, Material and methods, and Results and discussion. The manuscript was organized by subsection and the methods were either missing or inserted into each subsection.

We reorganized the texts. Now it starts with the introduction, experiment set- up/results, then coastal salt water data/stats, stream water and soil solution (Field method/results) from our field study, and ends with the discussion on species stability. See the marked text. Hope this change cleared this concern.

The methods were poor explained and not acceptable for scientific publication. For example, the source or methods for the field study in New Jersey, US were not provided. The basic parameters of the columns (porosity, pore volume, length of the column etc.) were not provided. It is unclear how the authors conducted the column experiments. Was the high alkaline salt solution continuous or pulse injected into the columns? What is the flow rate?

Good suggestions. But some details may not be all necessary, like pore volume? One only needs pore volume for Phreeqc transport simulations. We added them in any way. The underneath quoted texts (in “space”) are the changes we added (or edited) based on this reviewer’s suggestion.

Refined descriptions for experiment setup are added on p.3, line 81-85.

“2.1.1 Setup

Four columns with a diameter of 5.08 cm (two-inch) and height of 30.5 cm each were filled with about 480 grams of loamy soil collected from A and upper B horizons at the Centennial Lake Watershed (CLW), New Jersey (Figure 1). The soils used in the four columns have an average porosity of 0.66, an average hydraulic conductivity of 1.57 m/day, average

7 pore volume of 298 cm3, and an average organic matter content of 4.2% based on Loss On Ignition method.”

Description for runoff water sample collection setup along highway I-95 was added on p. 8, line 235-239.

“4.1 Runoff water samples off interstate highway

“A total of nine runoff water sampling cups were buried right below the surface along three transects perpendicularly to Interstate highway 95 in the Centennial Lake Watershed (CLW), New Jersey, US, (Figure 1). The three transects along I-95 are 0.9, 1.8, and 2.7 meters behind the guardrail of the highway in a grassy area and are approximately 5 meters apart. Five groups of samples were collected before and after salting periods between 2012 and 2013 (Table 2)”.

Description for stream water sample collections was edited on p.10, line 26-267

“4.2. Stream water

Water geochemistry data for the Delaware River at Trenton (Abbreviated as DRT from here on, Figure 1), New Jersey were obtained from USGS database between 2002 and 2015 when the As or Pb data were available. The water samples for the stream in the CLW, New Jersey were collected from two sample sites with one upstream and one downstream between 2011 and 2013”.

Description for set-up of the lysimeters and soil water collections was added on p.10, line 283- 286.

“4.3. Soil solutions from lysimeters

Three suction soil lysimeters were buried at approximately 110 cm depths from the surface along the banks of upstream sample sites, with two (#1 and #3) on one side of the bank and one (#2) on the other side of the stream bank (Figure 1). Lysimters #1 and #2 are near the bank, and lysimeter #3 is approximately 7 meters away from the bank.”

The initial text has this paragraph on p.4, line 91-96. Now the pulsed injection was added in.

“The columns used the falling-head apparatus set-up. 0.125 and 0.25 moles of sodium chloride solutions (1100 ml in volume) were injected into each column, respectively. After the injection (pulsed injection), distilled water was added once or twice daily (an average of about 300 milliliter daily) with no overflow allowed. The flow velocity was roughly 0.52 m/day when

8 the solution/water was added initially in all four columns. The velocity decreases as the hydraulic head drops”.

The methods for statistical analysis of the field data in section 2 and 5 were missing. How did the authors generate the dendrograms (Figure 8-10)? And so on. The manuscript need to be rewritten.

The current section 3 is the old section 2 that this reviewer referred. The statistical method used in this section for the coastal data was the simple linear correlation for concentrations of elements (Zar, Biostatistical Analysis, 5ed, p.380). Correlation values, significance based on t test, and sample sizes are given in Table 1. The “linear” (correlation) was added on P.7, line 200 during the first use.

Regarding the dendrogram, it is called the “concentration correlation dendrogram” during its 1st use on p.10, line 305 in this text and in Figure Captions (Figures 9-11). Also, on the figure y-axis label (Figure 9). The y-axis was labeled as Correlation Similarity Measure. I calculated the correlation of each group of the data (the concentrations of each element in this situation) to obtain a correlation matrix, then transferred them over to a stats package to plot the dendrogam (weighted average). I did this just to see the individual correlations. One can just transfer the original concentration data to a stats package, obtain the same cluster correlation in STATA program without obtaining the correlation matrix first.

Other stats program like SPSS (or R) can do the same dendrogram as STATA does. Most geo- bio stats books talk about the similarity dendrogram. A couple of lines were added in the current text on p.10, line 305-306.

“…concentration correlation dendrogram plotted with STATA program (Figure 8). Correlation dendrogram shows how closely one cluster/group is correlated with another cluster/group”.

2. Based on the experimental description, I don't feel that the experiments were well designed, carefully performed, and appropriately recorded. For example, the flow was occasionally stopped accordingly to section 3.2.1: "The injections were followed by a daily flushing with about 300 milliliter of low ionic strength water for three weeks, except weekends". However, the stop of the flow was not indicted in the figures. Usually, stop flow will cause re-equilibrium of chemicals and have major impact on their concentrations in the effluent.

The experiment probably can be refined. But the results probably would not differ. We first run the results using the constant head apparatus back in 2010 (Sun et al., 2010), but the overflow was an issue due to loss of the solution (there is a picture of constant head setup on Sun et al., 2010a, http://users.rider.edu/~hsun/sodium.pdf).

9 The current work used falling head set-up to eliminate the overflow issue. Four columns have almost identical conditions, except varied salt concentrations. They were flushed at the about the same time each day with the same amount of distilled water and have the same hydraulic condition. The set-up has been ing refined multiple times so that and the results are pretty reliable (Sun et al., 2010, 2012, 2014). The absolute concentration probably means less than the concentration change (or relative concentration) in this study.

However, the stop of the flow was not indicted in the figures.

The data points are discrete data points with date labeled in x-axis (Figures 2,4,5, x-axis label is now changed to days based on reviewer one’s suggestion). One can tell how long the gaps are between collection. For pulsed injection/falling head set-up, there are times when the flow slows down. Falling head test starts with high velocity and slows down as the head drops.

The following texts are edited on p.4, line 93-98

3. Powdered arsenopyrite and galena were added to two of the columns, of which As and Pb simultaneously eluted with Na+. This observation was expected because of the dissolution of powdered arsenopyrite and galena. However, it doesn't mean that As and Pb source in the environments, which are unlikely be powder and have been long equilibrium with the environments, will be released by high alkaline solution. I don't see the merit of adding powdered minerals in the column experiments.

Powdered arsenopyrite and galena were added just to increase the reaction area and therefore, the reaction kinetics. They are at a progressive equilibrium in a sense in nature. As the weathering continues, new particles are formed in soil and new organic matters/compounds are added to the soil. For example, in the studied CLW watershed, some spots of the soil samples have arsenic level more than 25mg/g based from my 10% phosphoric acid leaching result (Sun et al., 2015). CLW used to be an orchard with arsenic pesticide applied before 1960. Also, lead in the top soil near highway I-95 in the CLW where the runoff sample collected were 5- 10 times higher than background from one of my students’ study (Sarwar et al., 2015—reference is listed in the reference section). These metal stocks take time to be depleted.

10 Sodic soil (disproportionally high sodium with a sodium adsorption ratio >13 by definition) increase the dispersion of soil (i.e., breaking it down into smaller colloids, and preventing it from aggregating), breaks down soil particles, and forms new particles. This dispersion increases the kinetics of the reaction due to increased surface area, therefore, the release of arsenic contained particles. Similarly, powered minerals will increase the reaction/desportion.

Here are the changed texts on p.3-4, line 86-88 to emphasize this.

“Two of the four soil columns contained powdered arsenopyrite (~35 grams) and galena (~70 grams) mixture to initiate the potential elevated sources of As and Pb and to increase the reaction kinetics.”

Data and results are missing. In section 2, the field data from the Gangers, Chapai-Nawabgan and Meghna rivers were not described or analyzed at all before the statistic results were given in Table 1

Old section 2 is the new section 3. The raw geochemical data from Gangers, Chapai- Nawabgan and Meghna rivers were all published in their respective articles (referenced). The current text is only interested in the concentration correlation of As and Cl/Na of their data. We did not do any other stats on those data.

A short data description was added on p.7, line 194-199 is copied here.

“Arsenic (As) related geochemistry data from three areas: Gangers, Chapai-Nawabgan, and Meghna rivers across the country of Bangladesh were obtained from Reza et al. (2010a,b), Halim et al. (2010) and Shamsudduha et al. (2008). This delta area of Bangladesh where the data were collected is infamous for its high As concentration in the groundwater and As related health problems. The concentrations of As in some wells of Chapai-Nawabganj District was as high as 462 ppb (Reza et al., 2010b).

Linear correlations were analyzed between the concentrations of As and other elements for those data…..”

In section 3 and 4, some of the experimental results were not provided, such as the elute of Ca, Mg, K, Mn, Fe, P, S, Al, and Si. Yet the authors intensively discussed those data.

The data plots (Figures 2b, 4b) were added as suggested, and Figure 5 was edited to add Fe, Mn, Al concentrations.

11 In section 5, the authors indicated that "K is related to both cation exchange and dispersion of organic matter", but there was no data of organic matter at all. I don't think a scientific conclusion should be made without any support.

I deleted "K is related to both cation exchange and dispersion of organic matter" as suggested. However, K is somewhat different from the Ca, Mg and Na (the common cations). Organic matter is almost like the second source of the K+. It can be leached from fresh leaves, dead organic matter. I will discuss more on K in a later section.

Regarding the organic matter contents but there was no data of organic matter at all. on the study site next I-95 along the three transects, here is the added line ( p.8, line 245-247).

”Based on Sun et al. (2009, their Table 5), soil organic matter contents in top soil at the studied I-95 transect ranged from 2.11% to 4.08% from the loss on ignition method. Therefore, dispersion of soil organic matter exists as well. The large surface area of dispersed soil and soil organic colloids increases the reaction kinetics and accelerate the adsorption and desorption processes.”

5. By analyzing the field data in section 2, the authors concluded that high alkaline salt water had unfavorable influence on the release of As, while high alkaline salt water facilitated release of Pb and Hg.

There is negative correlation between concentrations of As and Cl among all the data we analyzed based on the stats. This relation is also seen in the coastal data as well.

In the laboratory column study, the results showed no significant release of As, Pb, and Hg with NaCl peaks (Figure 3) and synchronized release of As and Pb with Ca2+ and Na+ peaks (Figure 5).

In Figure 2 (old figure 3) for NaCl only injection, there were only delayed peaks of As, Hg and Pb after NaCl injection. This is showing up as negative correlation between the concentrations of these three metals with Na, Cl concentrations, though practically, injection of of NaCl enhanced the release of As, though delayed. But, in the CaCl2 and NaCl mixed salting, the concentration peaks are synchronized due to the strong cation exchange from Ca2+. The delayed peaks in NaCl only injection was almost certainly related to the sodic soil dispersion (particle breaking down, repel) condition. The sodium adsorption ratio (SAR) was above 70 for the samples we collected in the second day and stayed high until almost two weeks later in the NaCl only injection experiment.

However, the field samples are not collected as continuously (or discretely in a short time period as done in the lab experiment). Also, the salt level stays higher for a much longer

12 time, a month or two after the salt application as compared to days in the lab run. The delayed effect seeing in the lab seemed to be blended in in the field.

In the field observation in New Jersey, US (Figure 6), Hg and Na were negatively correlated at 10/18/2012 and positively correlated at 5/6/2013. Why the results are not consistent in the field observations and column experiments?

Misreading? Visually, they look different. But, Hg and Na are NOT negatively correlated for the field data on 10/18/201. The correlation coefficient is positive at 0.17 for the 2012 fall season (10/18/2012) and 0.43 for the whole data set. Concentrations of Hg and Na are high in the winter-spring (2013 season). Maybe seasonal differences in the fall and winter makes stats different. But, they are all positively correlated in the field data (p.15, line 427-429).

The stats are consistent.

6. In terms of discussion in section 5, the authors generally speculate beyond their results. The authors discussed mechanisms of As, Pb, and Hg release based on the results of column experiments and dendrograms. However, most of the conclusions cannot be made from their results.

The only thing that we felt might be on the speculative side is the species of As, Pb and Hg we assume existing. We added As, Pb and Hg’s metal-Cl-H2O (S) pe-pH pourbaix diagrams and new set of data from a large watershed (Figures 12,13,14). We had Pb (solid boundary) and As (dash boundary) in Figure 12 and Hg in Figure 14. The positions of pe-pH plots in the pe-pH diagrams show the stability of the most likely species for these three elements in water. This should substantiate our claim on the type of species existing.

The other points we made in the discussion/conclusion are: 1). dispersion contribution to the As, Pb and Hg release due to sodic soil condition, 2) cation exchange release, 3)…possible substitution release for As..

Soil/organic particles disperse, and breakup under sodic condition. Dispersion increases the surface area and brings the metals into the solution and re-equilibrates them under new condition. Maybe we did not explain the dispersion (deflocculation, repel) clearly initially. The sodic soil dispersion is well known in most soil textbooks (Brady and Weil, 2008). The SAR values in the leachate were higher than 70 after the first day of the injection and were higher than 13 throughout the experiment time until the last day (See SAR plot in Figure 3). The SAR in the I-95 runoff data were above 80 and between 40 and 130 in March 2013 when the samples were collected. This SAR value indicates the sodic soil condition for the poorly balanced electronegative colloids to repel.

13 For the cation exchange release, this is basically the Gaines and Thomas' (1953) ratio law, which states that the concentrations of one element in solution and solid is proportional to the concentration of another elements in solution and solids. When one has a swarm of Na or Ca, it bonds to exchange and release a proportional amount of other elements as the Gaines and Thomas’ law has to hold.

We modified the text to easy the guessing as much as possible throughout the text.

To me, the conclusions are more like a literature review. For example, the elements in Figure 8 were grouped into ion (Cl is not cation) exchange elements and dispersion-weathering related elements (what does dispersion-weathering mean?). K, an ion exchange element, was grouped into the dispersion-weathering related elements. The authors argued that K is also related to dispersion of organic matter. However, the dispersion of organic matter was not shown or discussed in previous sections.

From the questions the reviewer#3 raised before about the correlation stats we had for the coast data and the correlation dendrogram, it seems there is a misunderstanding. Figure 8 was grouped into Ion exchange group (Na, Mg, Ca as the cation, Cl and S as the anions) and the dispersion-weathering related group.

1). Ion exchange group: Na and Cl were injected at the same molar concentrations (1 to 1 molar ratio for Na,Cl), therefore, they should show a strong correlation in the dendrogram. Cation exchange dominates the soil reaction when the NaCl was injected because most soil colloids are negatively charged due to the isomorphic substitution, organic matter.. etc . Cl, as a anion, for anion exchange, is not as prevalent unless it is tropic soil which tends to have more positive charge. Enough chloride will replace SO42- or NO3- anions in the soil colloids as most textbook would show (the adsorption sequence are: PO43- >SO4-->Cl->NO3- ,Brady and Weil, 2008, p.345). There is nothing wrong with chloride in the ion exchange group.

2). Dispersion-weathering related group. Again, here the dispersion means soil dispersion or breaking down of soil particles as being used in the soil textbook by Brady and Weil (2008). Other soil book talks about this too (Hausenbuiller, 1985. Soil Science). We added Figure 3 of SAR plot and SAR value in Table 2 to make a point. The dispersion here is not the transport modeling dispersion. We explained the dispersion better in the new text since it seems reviewer #1 has the same issue with this term too.

Cation exchanges are instantaneously (in theory), and dispersion related weathering dissolution has a delay-time effect as almost all the exchange experiment will show.

Regarding the K element classified into the dispersion-weathering related group. K is kind of in its own category. It is easily exchangeable, the reviewer is right in that sense (Cation

14 selectivity: Al+++>H+>Ca++>Mg++>K+=NH+>Na+ Weak). However, its source is not the same as other elements such as Ca, Mg, Na in water. For example, the total elemental concentration of other elements in a stream can be divided into three categories = baseflow+ runoff water+ precipitation. This category works for Ca, Mg, Na and many other elements. But does not work for K because K can be easily leached not only from minerals (biotite.. feldspar), but also organic particles, like live leaves and dead branches. Ca, Mg, and Na can’t be leached that easily from the organics. Other studies had the same opinion too.

Cite from Sun et al. (2014), “Sources of potassium are attributed to the dissolution and hydrolysis of K-feldspar and biotite (Bowser and Jones, 2002; Brady and Weil, 2008). Potassium also has a high solubility and is readily leached from organic matter, hence, its concentration in stream water is affected more by decaying of organic matter than other cations (Markewitz and Richter, 2000)”. End citation.

In a sodic soil dispersion situation, colloid organic particles will disperse (as discussed before), increase its surface area, and increase the reaction kinetics, therefore, release more potassium. The dendrogram reflect exactly what the data is.

Another examples, As and P were close to each other in the dendrogram, the authors indicated that "The positive correlation between the concentrations of As and P implies the possible substitution of PO43- and AsO43- anionic groups in the minerals, because of their chemical similarity." It may be so, but I don't think the conclusion can be made from the dendrogram.

As and P positive relation has been documented in many studies. Some studies even reported 10-15% phosphoric acid can remove 99% of the arsenic(Tokunaga,S. Hakuta, T., 2002, Acid washing and stabilization of an artificial arsenic-contaminated soil. Chemosphere, Vol. 46, No. 1, p. 31-38). My own study also indicated 10% of phosphoric acid is more effective than 75% of concentrated nitric acid (Sun et al., 2015, GSA abstract report). The phosphate accelerated release was reported widely by other studies (Smedley P. L., and Kinniburg, D. G., 2002).

This text indicates “the possible substitution of PO43- and AsO43- anionic groups in the minerals, because of their chemical similarity “. Because there is a high correlation in the dendrogram between concentration of As and P. Also, since the dendrogram is based on a hierarchical relationship between elements, correlation between As and P can’t be always shown when As has a higher correlation with another element which weighted more.

We had some modification in the text. The current text reads as (p.11, line 310-315):

“The positive correlation between the concentrations of As and P implies a possible substitution of PO43− and AsO43− anionic groups in the minerals, because of their chemical

15 similarity (Cornell and Schwertmann 2003; Tessier et al. 1996). This positive relationship between concentrations of P and As was also reported in other studies (Smedley and Kinniburg, 2002) and a recent study for the DRT data by Sun et al. (2015).”

Similar discussions were made in this section. Since this manuscript has to be rewritten; I feel there is no need to list all of them at the present stage.

Many sections were re-written. Added five graphs (Figure 2b,4b, Figures 11-13) and data collected by USGS from another watershed was added as well to substantiate the points made in the discussion.