Isotopic offset between unconfined water and water adsorbed to 1 organic matter in equilibrium 2

Hydrophilic surfaces influence the structure of water close to them and may thus affect the isotope composition of 8 water. Such an effect should be relevant and detectable for materials with large surface areas and low water contents. The 9 relationship between the volumetric solid:water ratio and the enrichment of heavy isotopes in adsorbed water compared with 10 unconfined water was investigated for the materials silage, hay, organic soil (litter), filter paper, cotton, casein and flour. 11 Each of these materials was equilibrated via the gas phase with unconfined water of known isotopic composition to quantify 12 the isotopic difference between adsorbed water and unconfined water. Across all materials, enrichment of the adsorbed water 13 was significant and negative (on average -0.91 ‰ for O and -20.6 ‰ for H at an average solid:water ratio of 0.9). The 14 observed enrichment was not caused by solutes, volatiles or old water because the enrichment did not disappear for washed 15 or oven dried silage, the enrichment was also found in filter paper and cotton, and the enrichment was independent of the 16 isotopic composition of the unconfined water. Enrichment became linearly more negative with increasing volumetric 17 solid:water ratio and even exceeded -4 ‰ for O and -44 ‰ for H. This enrichment behavior could be modeled by 18 assuming two water layers: a thin layer that is in direct contact and influenced by the surface of the solid and a second layer 19 of varying thickness depending on the total moisture content that is in equilibrium with the surrounding vapor. When we 20 applied the model to soil water under grassland, the soil water extracted from 7 cm and 20 cm depth was significantly closer 21 to local meteoric water than without correction for the surface effect. This study has major implications for the interpretation 22 of the isotopic composition of water extracted from organic matter, especially when the volumetric solid:water ratio is larger 23 than 0.5 or for processes occurring at the solid-water interface. 24 25 Key-words: discrimination; protein; cellulose; surface effect; O-18; H-2 26


Introduction
The 18 O and 2 H isotope composition of water reflects climate and many processes within the water cycle (Bowen, 2010;Gat, 1996).Changes in the isotope composition of water can either result from the mixing of water with differing isotopic composition or from the change in isotopic composition by fractionation, especially between vapor and liquid.Such fractionation can be affected by ion hydration.In aqueous solutions, ions change the activities of the isotopologues of water (H 2 O, HDO, and H 2 18 O) due to their hydration.This, in turn, causes the isotopic fractionation between aqueous solutions and water vapor to differ from the fractionation between pure water and vapor (Kakiuchi, 2007;Stewart and Friedman, 1975).
Similar to salt, the surface of hydrophilic materials also interacts with water molecules creating a two-dimensional ice like water layer near the surface and a three dimensional liquid layer far from the surface (Asay and Kim, 2005;Miranda et al., 1998).Additionally, adsorption, may cause an energetic difference between water molecules at the surface of solids and the bulk water molecules.These structural and energetic differences may cause a difference in isotopic composition between Biogeosciences Discuss., doi:10.5194/bg-2016-71, 2016 Manuscript under review for journal Biogeosciences Published: 10 March 2016 c Author(s) 2016.CC-BY 3.0 License.
these two layers of water.If existent, such a surface effect should be strongest in materials with large specific surface area and with low water content.There are some indirect hints from studies of plant water uptake from soil, which show that mobile water differs isotopically from immobile water (Brooks et al., 2010;Evaristo et al., 2015;Tang and Feng, 2001) but to the best of our knowledge, such a surface effect has only been directly studied for clay (Oerter et al., 2014) and silica surfaces (Richard et al., 2007).It is not known how large the effect is for organic matter, which are associated with practically all mineral surfaces in the critical zone or form major constituents of other surfaces in the biosphere (Chorover et al., 2007;Nordt et al., 2012;Vazquez-Ortega et al., 2014).
A surface effect may be detected by establishing an equilibrium between water adsorbed to a material and air vapor created by unconfined water with known isotope composition in a closed chamber.If there is no surface effect, then the 18 O and 2 H isotope composition of the adsorbed water and unconfined water should be identical after equilibration.This is because the isotope composition of water under steady conditions is determined by the isotope composition of the water vapor, air humidity, equilibrium fractionation and kinetic fractionation (Helliker and Griffiths, 2007;Welhan and Fritz, 1977).All of these parameters are identical for adsorbed water and unconfined water when they both share for a sufficiently long enough time the same atmosphere, as is the case in a closed chamber.
We examined the hypothesis that the surfaces of organic materials influence the isotopic composition of adsorbed water and we choose materials of broad relevance.Silage is an important feedstuff delivering water to the animal and thus influencing the body water composition (Kohn, 1996) and animal products like milk.Hay in particular for example, has a low water content.Organic horizons at the soil surface (we call them litter thereafter) provide the interface where most vapor and water flows have to pass (Haverd and Cuntz, 2010).More pure materials like filter paper, cotton, casein powder and flour were included to identify whether the chemical identity causes or influences the effect.Finally we had to exclude that the effect resulted from artifacts like old water or volatiles and solutes interfering with the isotope measurements (Martín-Gómez et al., 2015;Schmidt et al., 2012;Schultz et al., 2011;West et al., 2011).Silage, which likely is a source of volatiles and solutes in rather large amounts (e.g., lactic acid, acetic acid, propionic acid, ethanol, and propanol; Porter and Murray, 2001), was also pretreated by washing and heating to remove potentially interfering substances.Water of contrasting isotope composition was used to identify any old water.Finally, we derived a simple prediction model for the effect and demonstrated its versatility in an application case with environmental samples.

Materials and Methods
We performed three equilibration experiments.Each equilibration experiment involved the exposure of samples to water vapor which originated from unconfined water, followed by cryogenic water extraction from samples and isotope composition measurement.We use δ 18 O and δ 2 H to describe the isotope composition of oxygen ( 18 O) and hydrogen ( 2 H) in water (with δ 18 O or δ 2 H = R sample /R standard −1, where R sample and R standard denote the ratio of the abundances of heavy and light isotopes in samples following the international SMOW standard).

Preparation of samples
The materials comprised fresh silage, oven dried silage, washed silage, hay, fibric litter (slightly decomposed organic material; for definition see Schoeneberger et al., 2012), hemic litter (decomposed organic material of low fiber content), filter paper, cotton, casein and wheat flour.The silage and hay were obtained from a farm near Freising and were cut in pieces (4 cm to 8 cm).The silage was stored in a -18 °C deep freezer while the hay was kept in a dark and dry place.The

Unconfined water
Five isotopically distinct, unconfined waters were used.We term them very heavy, heavy, tap, light and very light waters according to their relative ranking of δ 18 O and δ 2 H.These waters were produced from deionized water (δ

Set-up of the equilibration procedure
The different materials were placed in closed chambers (glass exsiccator vessels with a volume of approximate 20 L with drying agent removed) to equilibrate with unconfined water (Fig. 1).In a preliminary experiment, the effectiveness of the chambers' air seal was verified by flushing the containers with N 2, followed by monitoring the concentration of CO 2 and water vapour inside the vessels.The concentrations after closing the chamber remained constant, which indicated that leaks were negligible.In another preliminary experiment we assessed the development of humidity in the chamber.The humidity reached 100 % within 20 min (half-life 1.8 min) after we put 200 mL of water at bottom of the chamber (Fig. 1), closed it and started the recycling pump (Laboport, Germany).All equilibration experiments lasted for 100 h.Sun et al. (2014) have shown that even for moist samples equilibration is relatively fast (half-life 20 h).
In each experiment, 200 mL of (unconfined) water was placed in a glass bowl (15 cm in diameter) on the bottom of the chamber and dishes containing the material samples under focus (about 3 g fresh matter per dish) were placed on a perforated sill in the chamber.We flushed the chamber with nitrogen gas to remove the air vapor and the oxygen to prevent the decay of the samples.After that we immediately closed the chamber and started the recycling pump to ensure homogeneity within the airspace of the chamber.After 100 h of equilibration, samples were quickly removed from the chamber, placed in 12 mL glass vials sealed with a rubber stopper and wrapped with parafilm.The samples were then stored in a -18 °C freezer until water extraction by cryogenic vacuum distillation, as described by Sun et al. (2014).In addition, the weight of samples was recorded before and after extraction.Unconfined water (1 mL) was also sampled at the end of equilibration, underwent cryogenic vacuum distillation and was stored in a refrigerator.
The extracted water was analyzed with a Cavity Ring Down (CRD) Spectrometer using a L2120 -i Analyzer (Picarro, USA).Measurements were repeated until values became stable around a mean.Mean analytical uncertainties quantified as SD of different replicate measurements for each sample were ±0.06 ‰ for δ 18 O, and ±0.27 ‰ for δ 2 H. Post-processing correction was made by running the ChemCorrect TM v1.2.0 (Picarro Inc.) to exclude the influence of volatiles according to Martín-Gómez et al. (2015).

Experiment A: Influence of materials
This experiment focused on the enrichment between water in different materials and unconfined water after equilibration.
Dishes containing oven dried silage, hay, oven dried and fresh hemic litter, oven dried and fresh fibric litter, filter paper, bleached medical cotton, casein powder, or flour were all placed in different chambers for equilibration with unconfined water to avoid interference of volatiles in different materials.Eight samples for each material that differed in solid:water ratio were put in one chamber.Some materials (i.e., litter, filter paper, silage) were replicated in different experiments.The maximum number of samples for one material (silage) was 72.Flour and casein were powders and prone to form dust during vacuum water extraction.To prohibit this, the opening of vials containing flour and casein powder was covered by parafilm with tiny holes.

Experiment B: Influence of isotopic composition in unconfined water
This experiment aimed to find evidence that the enrichment was independent of the isotopic composition of the unconfined water.This independence will also prove that the enrichment cannot be caused by old water within the materials due to insufficient equilibration.Eight samples of oven dried silage in each case were placed into chambers to equilibrate with five different unconfined waters.

Experiment C: Pretreatment of silage
This experiment investigated the influence of volatiles on the isotope measurement and it assessed the effect of silage solutes on isotopic fractionation between silage water and vapor.
Fresh silage was divided into three groups (8 samples each): The first group did not undergo any pretreatment.For the second group, about 20 g of silage was immersed in 7 L of deionized water for about 2 min, stirred during immersion, then taken out using a colander and flushed with distilled water.After that we squeezed the silage by hand until no water drained off.This washing process was repeated three times.Finally, we reduced the water content of the washed silage by drying at 80 °C for 40 min.For the third group, silage was oven dried for 16 h at 100 °C to remove water and organic volatiles.These three groups (we call them fresh silage, washed silage and oven dried silage, respectively, thereafter) were placed in individual chambers and equilibrated with tap water for 100 h.

Statistics
For statistical evaluation we report two-sided 95% limits of confidence (abbreviated CL) to separate between treatments and OLS regression to describe relations between two variables.Measured values were fitted to expected relations by minimizing the root mean squared error (RMSE).Statistical requirements (normal distribution) were met in all cases.

Modelling
Conceptually, we assumed water to be part of one of two pools, which are arranged in a shell-like structure around the solid: an inner shell (or layer) which is in immediate contact or close to the surface of the solid and an outer layer that differs in thickness depending on the moisture content or solid:water ratio of the sample.Assuming that the outer layer has the same isotopic composition as the unconfined water once equilibrium was attained and that the inner layer has an isotopic composition that is influenced by the solid, the isotope composition of total adsorbed water (δ T ) was defined as: where f O is the fraction of water in the outer layer isotopically identical to the unconfined water, δ U and δ S are the isotope compositions of unconfined water and water influenced by the surface.
We defined enrichment (ε S ) between δ S and δ U as Combining eq. ( 1) and (2) leads to: From this it follows that the apparent enrichment (ε a ) between the total water in the material and unconfined water is given as: The fraction constituted by the inner layer f I in eq. ( 4) can be replaced by the ratio between R I , the volumetric ratio of water:solid associated with the layer that is influenced by the surface, and R T , the volumetric water:solid ratio of total adsorbed water: Assuming that the size of the inner layer R I as well as ε S are constant for a certain material, ε a should be related linearly to the inverse of R T , which is the volumetric solid:water ratio for the total adsorbed water.

Application case
Soil at 7 cm and 20 cm depths and rain water were sampled at the grassland in Grünschwaige Experimental Further, the winter data, effects of soil evaporation from the vegetation covered soil, can be excluded.We verify if the offset can be corrected by accounting for the volumetric solid:water ratio of the soil according to our model.To this end, the sand content of the soil was not considered in the calculation of the solid:water ratio given that the contribution of sand to water storage is marginal (Walczak et al., 2002) and sand grains usually are coated by clay, sequioxides, organic matter and biofilms and do not directly interact with water (Bisdom et al., 1993;Bolster et al., 2001).

Experiment A: Influence of materials
The apparent enrichment (sensu eq.4) of δ 18 O and δ 2 H was negative and significant for all materials, except for 18 O with filter paper and cotton and for 2 H in a few samples of cotton.The volumetric solid:water ratios differed between materials but also between different samples within the materials providing a wide range.δ 18 O and δ 2 H apparent enrichments decreased significantly with volumetric solid:water ratio over the range of materials.The decrease was also significant for the different samples within each material (Fig. 2).

Experiment B: Influence of isotopic composition in unconfined water
The isotope composition of absorbed water correlated closely with the unconfined water due to the wide range compared to the measurement errors (R² = 0.9990 and 0.9989 for O and H, respectively; Table 1).However, the regressions showed that the intercept differed significantly from zero and the slope from one, which indicated that the isotope composition of adsorbed water was significantly different from that of unconfined water.
Equation (3) predicted a linear relation between δ T and δ U similar to the linear regressions shown in Table 1.Different to a regression, however, the slope and the intercept of eq. ( 3) are not independent but depend on ε S × f O .To account for this dependency, the slope and the intercept of the linear equations were estimated by adjusting ε S × f O in eq. ( 3) to minimize RMSE, while fitting the measured δ T and δ U values.The optimal fits lead to: The R 2 between the predictions resulting from the two-layer model and the measurement were similar to that of the linear regression (R² = 0.9990 for 18 O and 0.9989 for 2 H), although the model has one degree of freedom less than the regression.
The resulting optimal ε S × f O values were -1.23 ‰ for 18 O and -22.6 ‰ for 2 H meaning that the effect was 18 times stronger for 2 H than for 18 O.
Equation ( 5) predicted that the apparent enrichment changes linearly with the solid:water ratio.This relation was highly significant also in the case when waters with very differently isotopic composition were used (R 2 : 0.7589 and 0.8599 for O and H, respectively; Fig. 3).These relations were identical for very heavy, heavy, tap, light and very light water.

Experiment C: Pretreatment of silage
There was no significant difference between mean gravimetric water contents (based on dry matter) of washed silage (153 % ± 33 %) and fresh silage (128 % ± 10 %) after 100 h equilibration.The water content of oven dried silage did not reach again the same water content as fresh silage but was significantly lower (81 % ± 13 %).The apparent enrichment of washed silage, oven dried silage and fresh silage all decreased with the solid:water ratio (Fig. 4), as already noted in the experiment with different materials (Fig. 2) or in investigations with unconfined waters of different isotopic composition (Fig. 3).Washing and oven drying should have removed most solutes and volatiles respectively and thus have created a large variation in the amount of solutes and volatiles among the treatments.Still, the relationship between enrichment of three types of silage and solid:water ratio followed the same line and the areas overlapped each other for the three types of silage (Fig. 4).This implied that neither the volatiles, which possibly could have adulterated the measurements, nor the solutes, which possibly could have influenced water activity in the silage, were the reason of enrichment.The different treatments, however, separated along the common line due to their differences in water content, which again corroborated the prediction that the apparent enrichment should linearly change with solid:water ratio.

Combining experiments A, B and C
When combining all experiments with different materials, different pretreatments and different unconfined waters, apparent enrichments covered a wide range of about 5 ‰ for 18 O and 46 ‰ for 2 H (Fig. 5).Even within the same materials, the range was up to 2.5 ‰ for 18 O and 25 ‰ for 2 H. Apparent enrichments within materials linearly decreased with the volumetric solid:water ratio.
In order to exclude that the enrichment was caused by incomplete extraction, we compared the observed enrichments with predictions based on Rayleigh fractionation resulting from incomplete extraction (Fig. 5).The enrichments predicted for Rayleigh fractionation fell far apart the observed enrichments.The average deviation between the expected and the observed 2 H enrichment was about 15 ‰.Furthermore, the slope of the relation between the enrichment of 2 H and 18 O was significantly steeper for the observed enrichment than the slope predicted for a Rayleigh process.Additionally, an unrealistically small fraction of water would have to be extracted (far below 0.8) to cause the same enrichment of 2 H as observed for most of the samples.

Application
For the growing season, soil water at 20 cm depth and 7 cm depth showed a distinct deviation from the local meteoric water line (mean deviation for 2 H: -8.1 ‰) with a slope almost identical to that of the meteoric water line (regression lines in Fig.

6, top panel
).An identical mismatch was detected for the winter season (markers in Fig. 6 The deviation between the winter season data and the local meteoric water line correlated significantly (p < 0.001) with the solid:water ratio for 7 cm depth but not for 20 cm depth.For both depths, the data moved closer to the local meteoric water line when the influence of confined water was removed by applying the general regression with solid:water ratio from Fig. 2 (Fig. 6, bottom panel).The mean deviation for 2 H changed from -8.1 ‰ to 1.0 ‰ due to this correction.

Discussion
The extraction of water from solid-water mixtures can be biased by incomplete extraction (Araguás-Araguás et al., 1995) or by the exchange of hydrogen or oxygen from the soil material with water molecules (Meißner et al., 2014).Here we add another confounding effect, which is the inhomogeneous isotopic composition of water above a solid surface.In the following we will discuss (1) whether the observed effect can be due to measuring errors or other reasons than the proposed surface effect, ( 2) what could be possible reasons of the surface effect, (3) which fields of application will this surface effect likely be important, and (4) which further work related to the surface effect may follow.

Excluding other mechanisms than the proposed surface effect
The study provided clear evidence that the water adsorbed by organic surfaces differed from what would be expected from the isotopic composition of the source water and it showed that this deviation became larger with decreasing water content.
Alternative mechanisms leading to an isotopic offset other than the proposed surface effect could be (A) volatiles (A) The surface effect was largest for flour and casein that do not produce volatiles.Also the filter paper and cotton, which contain no volatiles, had the decreasing trend between apparent enrichment and solid:water ratio (Fig. 2).Even for silage the influence of volatiles was not evident because washed or oven-dried silage, which should have lost all their volatiles, behaved identical to fresh silage.Also the error in water content caused by not accounting for volatile losses was negligible.
Using the correction function by Porter and Murray (2001) to calculate the true water content from the loss of weight, moves the respective data points of silage in Fig. 5 only invisibly (about 0.03 L/L towards right side).
(B) Solutes in water can influence the isotopic fractionation between water and vapor because the energy stage of water molecules bound in the primary hydration sphere of cations and anions differs from that of the remaining bulk water molecules (Kakiuchi, 2007).This effect has been shown for many salts (e.g., KCl, NaCl, Na 2 SO 4 and ZnSO 4 ).The strength of this effect varies between different ions and may be small (Kakiuchi, 2007;Sofer and Gat, 1975;Stewart and Friedman, 1975).NaCl even does not have an measurable effect on 18 O (O' Neil and Truesdel, 1991).Most of the solutes in our materials were organics for which the effect is unknown.However, this effect must have been small as the washed silage did not show a different pattern in enrichment compared to fresh silage (Fig. 4).Also the filter paper of analytical grade and bleached cotton that both should not carry any solutes did not show a different pattern.
(C) Insufficient time for equilibration may especially be relevant for silage and litter, which had the highest initial water content.For silage we could show that the apparent enrichment was independent of the isotopic composition in the unconfined water (Experiment B) despite the wide range of differently labelled unconfined waters (range for 18 O: 32 ‰; range for 2 H: 285 ‰).However, any old water would have led to a separation in the apparent enrichment.In contrast, our results were in accordance with the general rule that isotopic enrichment is independent of the isotope composition of the source, which is also underlying eq.( 4) and ( 5).Furthermore, all our experiments used deionized water prepared from tap water, except for the experiment with labelled waters for which we can exclude the existence of old water.Our deionized water was similar in isotopic composition to silage water and soil water.The mean δ 18 O of our water was -10 ‰ while the water ratio and enrichment that we have observed (Fig. 6).However, the predicted enrichment by incomplete extraction based on a Rayleigh fractionation fell far apart from the observed enrichment (Fig. 5).In addition, no significant weight difference before and after oven drying of the samples was observed after vacuum extraction.Incomplete extraction is thus an unlikely explanation.
(E) Kreuzer-Martin et al. (2005) found that 10 % of the total water extracted from Escherichia coli cells during the logphase of growth was generated by metabolism from atmospheric oxygen.Thus, intracellular water was distinguishable from extracellular water in δ 18 O.We flushed the chambers with nitrogen gas before equilibration to reduce availability of atmospheric oxygen and minimize microbial growth.For materials like silage dried at 100 °C or filter paper, any significant microbial growth is unlikely.Furthermore, isotopic adulteration caused by microorganisms should have caused 18 O and 2 H deviations in the opposite direction for the very heavy and the very light labeled experiments akin to the experiments by Kreuzer-Martin et al. (2005).In contrast to this 18 O and 2 H were always depleted in our experiments regardless of the isotope composition of unconfined water.
(F) Hydrogen bound to oxygen and nitrogen in many organic materials like bitumen, cellulose, chitin, collagen, keratin or wood may exchange isotopically with ambient water hydrogen (Bowen et al., 2005;Schimmelmann, 1991).At room temperature, this isotopic exchange occurs rapidly in water and an exchange with vapor is even several orders of magnitude faster (Bowen et al., 2005;Schimmelmann et al., 1993).Thus the exchange and the subsequent equilibration with the unconfined water will happen within 100 h.Furthermore, an exchange of hydrogen would not explain the observed offset in 18 O.

Possible reason for the surface effect
The enrichments became more negative with increasing solid:water ratio and they followed the predictions of eq. 5.This implied that similar enrichments existed in different materials and that the simple two-layer model sufficiently described the experimental values.Abundant evidence exists that the properties of water change close to a surface (Anderson and Low, 1957;Goldsmith and Muir, 1960;Miranda et al., 1998).A hydrogen-bonded ice like network of water grows up as the relative humidity increases.Above 60% relative humidity, the liquid water configuration grows on top of the ice like layer (Asay and Kim, 2005).This transition from a two-dimensional ice-like water to a three-dimensional water-like layer has been already been shown in several cases (Kendall and Martin, 2005).As we used 100% relative humidity in our chamber, both layers should have been present.
The anomalies of water close to a surface appear not to be particularly affected by the detailed chemical nature of the solid substrates with which the water is in contact.This is referred to as the "paradoxical effect" and is tentatively interpreted in terms of an energy delocalization phenomenon (Drost-Hansen, 1978).This agrees with our observation that the difference between materials was small compared to the large variation of the effect caused by a varying solid-water ratio.The small differences between materials that appear in Fig. 2 may hence only be an effect due to differences between the different materials in their specific surface area per volume of solid but not due to their chemical nature.
In accordance with our study, Richard et al. (2007) found that water adsorbed in porous silica tubes was depleted in 2 H compared to unconfined water and depletion increased with decreasing water quantity as a result of the interplay of molecular vibrational frequencies and intermolecular H-bonding.This mostly depends on the difference in zero-point energy between the 16/18 O-1/2 H bonds, which is compressed at the transition between the bulk liquid and the confined liquid influenced by the surface (Richard et al., 2007).Our data show, that the effect is much larger for 2 H than for 18 O and it practically disappears for 18 O when the solid:water ratio decreases below 0.5 (Fig. 6).This may explain why the effect has been previously described for 2 H but not for 18 O.

Fields of application
The isotopic composition of water in porous samples is usually determined by extracting all water in order to avoid any shift caused by Rayleigh fractionation.Hence, the inner layer close to the surface and the outer layer will be mixed.For many processes, especially in the transport of liquid water (e.g., groundwater recharge, stream flow discharge, water uptake by plants) only the outer, mobile layer will be relevant.The extraction of total water will then give a biased estimate of the mobile water.In accordance with our hypothesis, Brooks et al. (2010) even suggested two different soil water worlds to explain their data (mobile water and tightly bound water), which were not identical in terms of isotope composition.Tang and Feng (2001) also found isotopic differences between mobile and immobile water in soil and explained this by incomplete replacement of soil water by rainwater.Our laboratory experiments aimed to exclude such an effect.In our application case we also found a consistent offset between rain water and soil water that cannot result from incomplete replacement of old rain water in soil with new rain water because soil water had an offset from the meteoric water line.Such an offset has been shown for many locations around the world (Brooks et al., 2010;Evaristo et al., 2015), which challenges the assumption in land surface models that plants and streams derive their water from a single, well mixed subsurface water reservoir.
In other cases, which focus on the liquid-solid interface, only the water of the inner layer, which is influenced by the surface effect, will be relevant.For example, in studies of cell wall formation or degradation, the total water should be a biased estimate of the isotopic composition near the cell wall.Due to the change in apparent enrichment with water content, the total cell water will change just by a variation in vacuole volume even if the isotopic composition near the cell wall and in the vacuole remain unchanged.

Further work
Solid:water ratio is clearly not the best parameter to describe the two-layer model.The relation should be influenced by specific surface area and by wettability.Hence, the water volume per wetted surface area would likely be a better parameter.
For instance, when we wet the filter paper inhomogeneously, we got random results because the average solid:water ratio neither reflected the situation of the wet spots nor that of the dry spots.Also the increasing scatter for solid:water ratios >1.5 (Fig. 6) likely resulted from an inhomogeneous water distribution in these rather dry samples that may have left some parts of the sample completely dry and thus underestimated the water content of other parts.Still, our model was easy to apply and it worked sufficiently for the wide variety of materials examined.More materials varying in hygroscopic/hydrophobic behavior and in surface area should be included to better understand the rule behind the variation of enrichment and to expand the model.

Conclusions
There was an abundance of evidence to suggest that the surface effect influenced the enrichment between water adsorbed by organic matter and unconfined water.Many hypothetical reasons for an erroneous enrichment could be excluded.The variation of apparent enrichment with water content was well described by a simple, easy to apply two-layer model.This enrichment should not be neglected when the surface area is huge and the water content is low.The surface effect will become especially relevant for processes happening at the liquid-surface interface like the growth or degradation of the organic materials.
hemic and fibric horizons were gathered from a conifer forest near Freising (Germany) from a Haplic Podzol (according to IUSS Working Group WRB, 2014) area and stored in air tight bags in a refrigerator until use.In order to create a relative big range of water content, half of the litter samples were oven dried (16 h for 100 °C) before the equilibration experiment.Filter Biogeosciences Discuss., doi:10.5194/bg-2016-71,2016 Manuscript under review for journal Biogeosciences Published: 10 March 2016 c Author(s) 2016.CC-BY 3.0 License.paper(Rotilabo®-round filters, type 11A, Germany), made of 100 % cellulose, and bleached medical cotton (Hartmann, Germany) were prewetted by spraying because the initially dry filter paper and cotton hardly adsorbed any humidity from air.Both materials were then slightly oven dried for different times (ranging from 0 to 60 min) at 50°C before the equilibration experiment.The casein powder contained 90 % natural casein and a small amount of carbohydrate.The wheat flour contained 70.9 % carbohydrates, most of which was starch.Casein and flour were oven dried for 6 h at 50 °C before use.
, top panel) for which confounding effects of evaporation are minimal (on average potential evaporation 0.65 mm/d; actual evaporation 0.54 mm/d; precipitation 1.86 mm/d) and because the soil was completely covered by grass.Biogeosciences Discuss., doi:10.5194/bg-2016-71,2016 Manuscript under review for journal Biogeosciences Published: 10 March 2016 c Author(s) 2016.CC-BY 3.0 License.
adulterating the measurements; (B) solutes influencing the isotopic composition of adsorbed water; (C) insufficient equilibration time; (D) incomplete extraction of water; (E) metabolically produced water from microorganisms adhering to the materials; (F) exchange of hydrogen and oxygen between the organic matter and the adsorbed water.
Biogeosciences Discuss., doi:10.5194/bg-2016-71,2016 Manuscript under review for journal Biogeosciences Published: 10 March 2016 c Author(s) 2016.CC-BY 3.0 License.mean for 52 fresh silage samples analyzed by Sun et al. (2014) was -11 ‰ (SD 3 ‰).A small fraction of old water thus cannot cause the large observed effects.(D) An incomplete extraction should cause a large error at low moisture content, similar to the general relation between solid: Biogeosciences Discuss., doi:10.5194/bg-2016-71,2016 Manuscript under review for journal Biogeosciences Published: 10 March 2016 c Author(s) 2016.CC-BY 3.0 License.