Bromide reactivity in topsoil: Implications for use as a “conservative” tracer in assessing quantity and quality of water

Bromide is a frequently used conservative tracer in soil leaching studies, including studies on contaminant leaching from arable topsoils. However, bromide often does not behave conservatively. Biogeochemists have known for many years that in natural soils, bromide is converted into organic bromine in a process known as bromination. However, bromination is seldom used as an explanation of non‐conservative leaching behavior by soil hydrologists. In a controlled small‐scale lysimeter study with arable soil we demonstrate such nonconservative behavior of bromide in opposition to control columns with fine gravel/coarse sand. By combining a literature review with the lysimeter study, we demonstrate the potential importance of bromination in topsoil and that bromination cannot be ignored, when interpreting bromide tracer experiments in arable soils. We also highlight the need for further studies on the processes of bromination and remineralization, to be able to account for these when conducting bromide leaching assessments.

One of the key properties of a conservative tracer is that it is non-reactive in the environment, in which it is used as a tracer. Apart from its uptake into plants (Kung, 1990) and some adsorption to acid soils (Goldberg & Kabengi, 2010;Groh et al., 2018), bromide is typically considered non-reactive by soil hydrologists (Käss, 1998;Leibundgut & Seibert, 2011;Torrentó et al., 2018). However, it is well known by biogeochemists, that bromide is reactive in natural organic-rich topsoils like those in forests and bogs (Biester et al., 2006;Cortizas et al., 2016;Keppler et al., 2000;Leri & Ravel, 2015) and that most bromide, that enters natural soils with precipitation, is transformed into organic bromine in a process known as bromination (Albers et al., 2011;Biester et al., 2004;Cortizas et al., 2016;Zaccone et al., 2008), see Equations (1) and (2). The first part of this reaction is biological, since it requires both an enzyme such as bromoperoxidase as well as hydrogen peroxide, which in soil is typically produced by microorganisms. The second part of the reaction, the actual bromination Vadose Zone Journal of the organic matter, occurs spontaneously, once the reactive (oxidized) bromine is formed.
Br − +H 2 O 2 +H + → HOBr+H 2 O (enzymatic) (1) HOBr + org − C → org − C − Br (nonenzymatic) Actually, it was shown as early as in 1968 that humus contains organic bromine (Yamada, 1968), and the concentration of organic bromine is typically many times higher than that of bromide in topsoil (Albers et al., 2011(Albers et al., , 2017Leri & Myneni, 2012;Maw & Kempton, 1982). This discrepancy in knowledge calls for an evaluation of bromide reactivity in soils and hence whether bromide can still be regarded as a conservative tracer by soil hydrologists. Many bromide tracer studies are conducted in soils used for agriculture, which are very different from the organic forest and peat soils used in previous bromination studies, especially with regards to organic matter content and pH, which both have shown to be important parameters in soil halogenation reactions (Gustavsson et al., 2012;Öberg et al., 1996). There are many examples of bromide mass balances being incomplete in tracer studies where bromide was applied on the soil surface (e.g., Brown et al., 2000;Hagedorn et al., 2015;Lange et al., 1996;Mehrtens et al., 2020;Torrentó et al., 2018;Wessolek et al., 2000;Wettstein et al., 2016;Whitmer et al., 2000;Wilson et al., 1993). Also examples on unexpected bromide leaching at later stages of a tracer experiment are reported in the literature, where all bromide should have been flushed from the system (Bero et al., 2016;Mehrtens et al., 2020). Such delayed leaching could be due to mineralization of organic bromine formed in the beginning of the tracer experiment or decay of crop residues and coherent release of bromide taken up during growth. So far, these abnormalities have never been interpreted as a consequence of bromination reactions occurring in the soil but are often attributed to uptake through crop roots or, in case of acid soils, to sorption on positively charged metal oxide surfaces. The purpose of the present update paper is twofold: (I) to make aware that the known process of bromination can make bromide behave very differently from a conservative tracer in topsoil and (II) to illustrate this via a highly controlled small-scale lysimeter study. We hope that this will lead to further studies that qualify and quantify the issue in different environmental settings.

A LYSIMETER SCALE STUDY
To see, if biogenic bromination observed in studies on bromide in natural organic rich soils indeed could lead to a non-conservative behavior of bromide observed in a tracer experiment, we made a small outdoor lysimeter study, where

Core Ideas
• Bromide is frequently used as tracer but often does not behave conservatively in soil. • By a combined literature review and lysimeter study, we demonstrate the potential importance of bromination in soil. • It is clear that bromination cannot be ignored, when interpreting bromide tracer experiments in arable soils. • Studies are needed to account for the processes of bromination and remineralization in topsoil.
other factors such as root-uptake could be excluded. Bromide was applied as dissolved KBr in similar amounts per area as we typically use it as a tracer in field experiments. The lysimeters consisted of a 40 cm long stainless-steel tube (inner diameter 15 cm) inserted into a "column shoe" that sealed the bottom of the lysimeter and allowed leachate to be drained (Albers et al., 2020). Four replicate lysimeters with sandy arable field soil were constructed by pushing 32-36 cm of the cylinder into the soil using the weight of a drilling rig and then excavated out of the soil to ensure as little disturbance as possible. The lower 12 mm of each soil core was replaced with fine quartz sand before fitting it into the column shoe. In addition, four lysimeter were packed with a commercial gravel type (Slotsgrus®, Stenrand Gravel pit) in the laboratory. The gravel contained particles up to 11 mm with ∼40% gravel sized particles (>2 mm), ∼30% coarse sand, ∼20% fine sand, ∼10% fine particles (<0.063 mm), 3% clay, and ∼0.3% organic matter. Packed gravel lysimeters were included as controls with hydraulic properties like the Jyndevad arable soil, but with minor bromination activity expected. An empty lysimeter was also included. More details on the lysimeter system are available in Albers et al. (2020). After cutting away a few small herbaceous plants growing on the lysimeters, each lysimeter received 88 mL of a 500 mg L −1 bromide-solution directly on the bare soil (dissolved as KBr in tap water diluted three times with milliQ-water) corresponding to 5 mm of precipitation and 2.5 g m −2 bromide (44 mg per lysimeter). The bromide was applied on September 4, 2017 and all leachate was collected thereafter in stainless steel bottles through a 1 m stainless steel tube (inner diameter 2 mm) hanging from the middle of each lysimeter. Dissolved bromide in the leachate was analyzed on an ion chromatograph (Metrohm 819 IC detector with a Metrosep A 150/4.0 column) after filtration (0.45 μm polyethersulfone filter). The leaching of bromide was followed for 17 months in the agricultural topsoil, but only for 2 months in the gravel and empty lysimeters since all added bromide was recovered within that period and bromide concentrations had ceased to below limit of quantification, which was 0.05 mg L −1 bromide.
The sandy arable soil and the control lysimeters with gravel showed similar timing and width of the main tracer peak. However, while the bromide recovery across this peak was 97%-99% in the control lysimeters and 99% in the empty lysimeter, only 40% was recovered in three out of four arable lysimeters and only 25% in the fourth (Figure 1). The lysimeters with arable soil were then monitored for an additional 16 months showing a continuous release of bromide, especially during the first year. The leachate collected within that period corresponded to a net precipitation of 300-350 mm, which again corresponds to at least seven times exchange of the active pore volume in the lysimeters (Albers et al., 2020). The bromide concentrations periodically increased, (Figure 2), but towards the end of the experiment, the bromide concentrations in the leachate of all lysimeters was around or below the quantification limit (0.05 mg L −1 ). The bromide concentration in the precipitation was always below the detection limit. At the end of experiment, bromide recovery had increased to 43%-58% from the arable lysimeters.
The lysimeters each received 44 mg bromide, corresponding to ∼4 mg kg −1 soil. 60%-75% of this was retained in the soil during the first 2 months (Figure 1a), where all the tracer should have been flushed out if it behaved conservatively (130 mm net precipitation in this period corresponding to at least 2.5 times the exchange of the active pore volume). This would correspond to a bromide removal (bromination) rate of ∼0.04 mg kg −1 per day if including the whole lysimeter as active. Since the bromide would not be distributed over the whole lysimeter across the period and since some bromide leached from the lysimeters within the first week after tracer application, the actual rate must have been higher.

DISCUSSION
Our results with the lysimeters indicate that bromination is not only important in natural organic-rich soils, but also occurs in arable soil and that this could make bromide a nonconservative tracer when applied on topsoil. This may not be a big surprise to many biogeochemists but is something that is not widely considered among those who use bromide as tracer (Käss, 1998;Leibundgut & Seibert, 2011).

Bromide works as a tracer under certain conditions
But then, why has bromide been adopted as a conservative tracer? Levy and Chambers (1987) showed that bromide does not sorb to soil and are often cited for the conservative nature of bromide in soil. However, their experiment lasted 6 h and included relatively high bromide concentrations (8-8000) mg L −1 . Any biotic transformation of bromide to organic bromine would not be observed under these conditions. There are, as already mentioned in the Introduction, many examples of bromide mass balances being incomplete in tracer studies where bromide was applied on the soil surface (e.g., Brown et al., 2000;Hagedorn et al., 2015;Lange et al., 1996;Mehrtens et al., 2020;Torrentó et al., 2018;Wessolek et al., 2000;Wettstein et al., 2016;Whitmer et al., 2000;Wilson et al., 1993). However, there are also examples of high recoveries, so the question is if these examples are contradictory to the proposed non-conservative behavior due to bromination in topsoil? For example, Srinivasan and Sarmah (2014) found high bromide recovery (up to 94%) in a column study. However, they applied a very high dose and concentration (3.6 g L −1 ) and rapid elution, with the whole experiment conducted within a day and most of the tracer eluted within 10 h. Bromination reactions would obviously remove only a very small fraction of bromide under such conditions. In another study, Frey et al. (2012) found high bromide recovery (>100%) in the tile drains of a small field study. They applied 250 g m −2 of bromide, which is 100 times more than applied in our study and could explain the high recovery/low incorporation in soil, since bromination activity would not keep pace with this amount of bromide. A high bromide recovery in some studies therefore seems not to be contradictory to a bromination activity occurring in topsoil. Actually, these studies may well be in line with the proposed importance of bromination in many other studies because bromination activity would have minor relevance at very high bromide concentration or very short retention time, where enzyme activity would be rate limiting for the brominating reaction. One could therefore hypothesize that at high bromide concentration and short retention time, the bromide could be suitable as a "conservative" tracer applied on topsoil. A short retention time may be applicable in laboratory studies, but not in field or lysimeter studies with natural precipitation. Using a high concentration may also not be desirable, as high bromide concentrations are known to influence microbial processes (Bech et al., 2017) and furthermore could influence the density and hence flow of the water.

Application time and plant cover probably influences the effect of bromination
Timing of the tracer application will probably also influence the loss of tracer through bromination of organic matter due to differences in infiltration rates and hence, the time that bromide will stay in topsoil and because bromination activity similar to other biological processes will depend on temperature. For example, a spring or summer application probably will lead to a higher tracer loss through bromination because of low infiltration rate and high microbial activity during summer than if the tracer is applied in the late Autumn, where the bromide would leach faster to below the A-horizon where less bromination activity can be expected (although lower bromination activity in subsoil remains to be shown experimentally). On the other hand, our lysimeter tracer study, was started in the autumn and the first month of the study was very rainy for Danish conditions and still a large fraction of the bromide was lost in the soil. Although more experiments are needed to understand the seasonal influence of bromination on bromide tracers in soil, timing of a tracer study therefore probably cannot be used in itself to exclude the influence of bromination.
In addition to conversion of inorganic into organic bromine in the topsoil, the presence of plants may further affect a bromide tracer experiment, since plants are known to take up bromide (Kung, 1990;Schnabel et al., 1995;Shtangeeva et al., 2017;Xu et al., 2004). As an example, that bromide fate can be complicated, also when bromide is taken up by plants, Wishkerman (2006) found that some of the bromide taken up in plants was converted into organic bromine by bromination in decaying leaves, of which some of the organic bromine was then apparently lost from the system by volatilization.

3.3
Can sorption also affect bromide recovery?
Other explanations for low bromide recovery than plant uptake have been attempted in the literature. Sorption to protonated mineral surfaces was, for example, given as possible explanation for low bromide recovery in a tracer experiment with bromide in topsoil (Hagedorn et al., 2015). However, the authors provide no evidence for this explanation (theoretical or experimental) and although some minerals do seem to have the capability of adsorbing bromide at low pH (Goldberg & Kabengi, 2010) evidence in the scientific literature suggest that even in highly acidic soils rich in metal oxides, sorption will only lead to a delay of the bromide peak, not a lower recovery (Groh et al., 2018). Actually, the study by Hagedorn et al. (2015) probably is a nice example that bromination in soil is an important process governed by microbial activity. Hagedorn et al. (2015) obtained almost 100% bromide recovery in 10-year-old soils that had just emerged from a retreating glacier where microbial activity can be expected to be very low. In 70-year-old soils a little further downhill, 70% bromide recovery was obtained and in old soils further downhill between 20% and 50% was recovered. Instead of sorption processes, bromination due to microbial activity seems to be a much more plausible explanation for the recoveries observed in that study.
In our lysimeters, the few plants present were cut before adding the tracer and uptake by plant roots is therefore highly unlikely. Batch sorption experiments in the lab showed no measurable sorption to the same arable soil as used in the lysimeter experiments (K d < 0.02 L/kg) and significant influence of sorption is therefore highly unlikely as well. Sorption being insignificant in the Jyndevad arable soil fits with the fact that the soil had a pH of approximately 6, which is higher than pH-values at which a significant sorption of bromide to soil constituents has been shown to occur (Goldberg & Kabengi, 2010;Groh et al., 2018).

Delayed leaching of bromide due to remineralization
The increased concentrations and even peaks of bromide measured long after application of KBr that we ( Figure 2) and others (Bero et al., 2016;Mehrtens et al., 2020) have observed can most likely be explained by mineralization of organic bromine formed during the tracer experiment, similar to what is known for the chlorine cycle (Gustavsson et al., 2012;Svensson et al., 2021). A seasonal variation in the ratio between bromination and debromination of organic matter in soil could then be a likely explanation of, for example, the unexpected results in Bero et al. (2016), where bromide concentrations above background persisted unexpectedly long after tracer application and where bromide reappeared in the application plot wells in concentrations that were greater than the initial breakthrough concentrations. Such mechanistic details will, however, remain speculative until future studies dig deeper into soil bromination and debromination processes.

Loss of bromide due to volatilization
All in all, microbially derived bromination of soil organic matter is the most likely explanation for some or all of the removal observed in a number of studies as well as in our lysimeter test. Yet, the formation of volatile organic molecules containing bromine should be mentioned as another mechanism that could influence mass balances of bromide in topsoil. It is well known that volatile organic compounds containing bromine can be formed in soil (Albers et al., 2017;Hoekstra et al., 1998;Redeker et al., 2000;Rhew et al., 2003), especially when the soil is enriched in bromine (Albers et al., 2017) or when bromide is added to the soil (Hoekstra et al., 1998). Some of the bromide may therefore be lost by bromination followed by volatilization or by direct methylation to form the very volatile compound methyl bromide. The amount of volatile organobromine formed when using bromide as tracer is difficult to quantify, when the specific compounds are unknown, but it would be very interesting to attempt this in future studies either by targeting a range of suspect compounds or by nontarget analysis using high resolution mass spectrometry. This would reveal if some of the added tracer is not just retained in the soil organic matter by bromination reactions but is lost from the system by volatilization.

While we await further studies
More studies are surely needed to evaluate the importance of bromination on tracer studies in soil. Ultimately these studies should include both the above-discussed analysis of solid and volatile organic bromine species as well as organic bromine in soil water, of which no previous studies exist. A modification of the well-known adsorbable organic halogen (AOX) method could be useful in that regards. Until such studies are made, we strongly recommend that bromination is considered when planning and interpreting tracer studies with bromide in the vadose zone. Probably there are situations where bromide will work well as a tracer, such as: 1. At much higher bromide concentration than we have used (may have other effects, though as discussed). 2. At very short retention time using artificial percolation (not possible in field and lysimeter studies, though).
3. In situations, where mass balance is not important since a fraction of the bromide does behave conservatively. 4. In studies with subsurface soil, where bromination activity most likely is low or absent. However, this remains to be shown experimentally.

CONCLUSIONS
There is plenty of evidence in the literature that bromide is not conservative in topsoil. In organic rich natural topsoils, biogeochemists ascribe this to bromination of organic matter.
Probably this mechanism would also occur in other soil types, such as arable soils that are often the target of bromide tracer studies. This would explain many odd tracer results in the literature and is also the most plausible explanation of bromide behavior in the controlled lysimeter experiment we made as an example.
On the other hand, bromide is suitable for detecting water breakthrough since a fraction of the bromide does behave conservatively. The size of the conservative fraction will probably depend on bromide concentration, scale, soil media, and retention time. The incorporation of bromide into organic structures also means that bromide leaching from remineralized organic bromine must be expected long (years) after bromide application. The full understanding of these processes is still lacking, so future studies should focus on describing the processes of bromination and remineralization, to be able to account for these when conducting bromide leaching assessments. This knowledge will reveal the appropriateness of applying bromide as a representative tracer for the water flow and hereby to which degree bromide application can add value to leaching studies of contaminants in the aquatic environment.

AU T H O R C O N T R I B U T I O N S Christian Nyrop Albers:
Conceptualization; data curation; investigation; methodology; visualization; writing-original draft. Annette Elisabeth Rosenbom: Funding acquisition; investigation; writing-review and editing.

A C K N O W L E D G M E N T S
The authors thank Danish Pesticide Leaching Assessment Program for providing soil samples. The authors are also thankful to Ole Stig Jacobsen for valuable discussions on halogen behavior in the environment and on the construction of the lysimeter system.

C O N F L I C T O F I N T E R E S T S T A T E M E N T
The authors declare no conflicts of interest.