Concentration and spectroscopic characteristics of

Science of the Total Environment 622–623 (2018) 385–393

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Concentration and spectroscopic characteristics of DOM in surface runoff and fracture flow in a cropland plot of a loamy soil Qingsong Xian a,b, Penghui Li c, Chen Liu a,⁎, Junfang Cui a, Zhuo Guan a, Xiangyu Tang a a b c

Key Laboratory of Mountain Surface Processes and Ecological Regulation, Institute of Mountain Hazards and Environment, Chinese Academy of Sciences, Chengdu 610041, China University of Chinese Academy of Sciences, Beijing 100049, China Department of Ocean Science and Engineering, Southern University of Science and Technology, Shenzhen 518055, China

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• Surface runoff has higher DOC and fluorescent DOM contents than fracture flow. • Soil DOM export responds strongly to rainfall and is dominated by underground path. • Concentrations of DOM components (C1, C2) have consistent changing trends with DOC. • C1/C2 distribution decreases with soil depth and reflects water source differences. • DOM optical properties potentially act as water tracer for hydrological processes.

a r t i c l e

i n f o

Article history: Received 26 July 2017 Received in revised form 30 November 2017 Accepted 2 December 2017 Available online xxxx Editor: Yolanda Picó Keywords: Soil Dissolved organic matter Hydrological process EEM fluorescence PARAFAC analysis

⁎ Corresponding author. E-mail address: [email protected] (C. Liu).

https://doi.org/10.1016/j.scitotenv.2017.12.010 0048-9697/© 2017 Elsevier B.V. All rights reserved.

a b s t r a c t Being crucial for predicting the impact of source inputs on a watershed in rainfall events, an understanding of the dynamics and characteristics of dissolved organic matter (DOM) export from the soil under particular land use types, particularly those associated with underground flows is still largely lacking. A field study was carried out using a 1500 m2 slope farmland plot in the hilly area of Sichuan Basin, Southwest China. The discharge of surface runoff and fracture flow was recorded and samples were collected in four representative rainfall events. For DOM characterization, concentration of dissolved organic carbon (DOC) and absorbance/excitation-emission matrix (EEM) fluorescence were analyzed. Soil water potential was also determined using tensiometers for understanding the runoff generation mechanisms. The DOC values for both surface and fracture flow showed significant responses to rainfall, with hydrological path being the primary factor in determining DOM dynamics. EEMPARAFAC analyses indicated that the soil DOM mainly consisted of two terrestrial humic-like components with peaks located at Ex/Em 270(380)/480 nm (C1) and 250(320)/410 nm (C2), respectively. Concentrations of these components also responded strongly to rainfall, fluctuating in good agreement with the corresponding DOCs. Although there was no change in the presence of the components themselves, their relative distributions varied during precipitation, with the C1/C2 ratio increasing with the proportion of soil pre-event water. As the dynamic changes of soil DOM characteristics can be successfully captured using spectroscopic techniques, they may serve as a tracer for understanding hydrological paths based on their potential correlations with water source differences during rains. © 2017 Elsevier B.V. All rights reserved.

386

Q. Xian et al. / Science of the Total Environment 622–623 (2018) 385–393

1. Introduction Due to intensive soil management and activities designed for crop yield increase, agricultural systems have been recognized globally for their role in impacting organic carbon cycling in natural watersheds. As one of the greatest cycled reservoirs of organic matter on earth, dissolved organic matter (DOM) is an important form of soil C migration (Mladenov et al., 2010). The rainfall runoff carries high concentration of biolabile and terrestrial DOM into rivers, accounting for a significant portion of annual export from watersheds. In this light, an understanding of hydrological pathways through which water moves to streams is important because different pathways provide different opportunities for water contact with soil organic matter. During rainfall events, DOM can move quickly via overland flow and also in association with lateral subsurface flow. DOM can also move vertically to ground aquifers by percolation which is sometimes via preferential subsurface channels such as soil macropores and mudstone fractures (Zhang et al., 2016; Ward et al., 2017). Thus, surface and subsurface runoff often occur synchronously and together lead to a large combined flux of DOM export from soil (Hood et al., 2006). In comparison to the large number of studies focusing mainly on rainfall response of DOM in streams and spatial variations at watershed scale (Buffam, 2001; Raymond and Saiers, 2010; Jeong et al., 2012; Singh et al., 2015), very little work has been conducted to look into the soil hydrological process that regulates the terrestrial-aquatic linkage in DOM migration. At present, we still lack information on DOM release dynamics and paths from a soil under a specific land use type, which is a resolution of uncertainties in concise prediction of the total carbon flux added up by soils of different land uses and evaluation of land cover effects in a catchment. In particular, as many studies have found that storm runoff is largely supplied by pre-event “old” water that moves via subsurface routes to a stream channel (Gremillion et al., 2000; McDonnell, 2003), it is important to understand when and how much the underground path contributes to DOM export during rainfalls, especially in soils with abundant occurrence of subsurface and ground runoff. The residence time of water is one aspect of DOM transport. The time the water and associated DOM spend in each path will ultimately determine DOM's characters and fate (Ward et al., 2017). The hydrological dynamics of underground DOM is so far understudied in comparison to overland flow due to relatively long time lag in response to rainfalls. Obviously, rainfall patterns will determine the mechanism of runoff generation and in turn affect the migration path of DOM in the soil. In some situations, the DOM characteristics of rainwater itself need to be considered seriously as they may alter DOM concentration and properties via aboveground processes such as throughfall and stemflow (Santos et al., 2012; Neu et al., 2016). Knowledge of how the flow paths are linked to the rainfall regime is warranted for addressing our questions: (1) what is the dominant route for DOM migration in a particular soil, and (2) how will the amount and characteristics of DOM change during the transformation of rainwater to runoff? On the basis of existing knowledge on the generation of runoff, we intended to examine the dynamic response of DOM in flow processes so as to evaluate the impact of rainfall events on soil carbon delivery. Answering the second question above requires insights into DOM's chemical compositions. There are many analytical techniques readily available including resin adsorption, size exclusion chromatography (SEC), UV absorbance and fluorescence spectroscopy, nuclear magnetic resonance (NMR) spectroscopy for examination of different aspects of DOM characteristics (Matilainen et al., 2011); moreover, the Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS) is a very recent high-tech with capability of capturing DOM precise molecular signatures (Stubbins et al., 2017). Additionally, biomarkers such as lignin phenol have been used to trace DOM mobilization dynamics from soils during storm events (Ward et al., 2012). Among these, the use of excitation-emission-matrix (EEM) fluorescence is of particular interest in the past decade due to high efficiency and low cost. Taking

advantages of spectroscopic techniques that have made successful characterization of structural features and changes of chromophoric DOM associated with natural and engineered processes (Stedmon et al., 2003), studies on the spatial variations of DOM in aquatic environments from streams to oceans, the source tracing and the evaluation on land cover effects in watersheds have been largely carried out in recent years (D'amore et al., 2010; Mcelmurry et al., 2014; Larsen et al., 2015; Nimptsch et al., 2015; Gonçalves-Araujo et al., 2016; Singh et al., 2017). Apparent impacts of storm events on the characteristics of stream DOM, such as, changes in aromaticity, hydrophilicity and average molecular weight have been observed previously (Hood et al., 2006; Inamdar et al., 2011; Yang et al., 2013). At the plot scale, on the other hand, we assume that DOM in a given soil that consists of relatively fixed classes of components tends to be stable during a rainfall event; whereas the component distributions may change as a result of variation in flow paths. In this research, four rainfall events of representative precipitation amounts, durations and intensities were monitored, using a 1500 m2 sloping cropland of an entisol. According to USDA Taxonomy (Gong, 1999), entisols are the second most abundant soil order, occupying about 16% of the global ice-free land area. The soil stands for the major arable land resource in the upper reaches of the Yangtze River, China. As soil hydrological processes in the region have been accounted previously (Tang et al., 2012; Zhao et al., 2013), this study seeks to provide insights into the concentration and compositional dynamics of soil DOM associated with migration pathways in rainfall events. Results of this study, as well as the demonstrated applicability in the use of spectroscopic techniques, can offer wide implications for other eco-regions and land use settings.

2. Materials and methods 2.1. Field site and experimental plot setup The study site was located in a small agricultural headwater catchment (0.35 km2) in the hilly area of central Sichuan, Southwest China (31°16′N, 105°28′E). Crop farmland accounts for 55% of the catchment area. The region has a moderate subtropical monsoon climate with an annual average temperature and rainfall of 17.3 °C and 826 mm (during 1981–2006), respectively, and 65–85% of precipitation occurs in summer and autumn. The loamy soil of the study site is an entisol with an average texture of 27% sand, 52% silt and 21% clay. The soil represents the major arable land resource in the upper reaches of the Yangtze River, China. As shown in Fig. 1, the experimental plot (slope 6°, 50 × 30 m) consists of three layers: An upper layer of readily erodible soil (with depths ranging from 30 cm at the upslope to 60 cm at the downslope), a middle layer of fractured mudrock (from 2.1 m to 4.8 m) and a lower layer of impermeable sandstone. According to previous studies in the region, the saturated hydraulic conductivity (Ks) was 37–43 mm h− 1 for the depth of 10–15 cm (Wang, 2013), 17.6 mm h−1 for the depth of 25–30 cm and 12.9 mm h−1 for the fractured mudrock (Zhang et al., 2016), respectively. The soil has relatively abundant structural pores (N 0.21 cm3 cm− 3) for water movement, among which pores of N250 μm may account for above 90% of Ks. The plot was hydrologically isolated from the surrounding geologic formations with four cement walls (20 cm above the ground and 20 cm into the impermeable sandstone layer) to make sure rainfall was the only external water recharge for the plot. It was also designed to avoid water stagnation in the vicinity of the concrete borders. Three conflux grooves were built at the soil surface, soil-mudrock interface and the interface of the mudrock and impermeable sandstone for the collection of surface runoff, interflow and fracture flow samples, respectively. The plot was ploughed in May and October every year with maize-wheat rotations without irrigation. Maize was planted in the field during the experiment. Wheat straw was left on soil surface


Fig. 1. Schematic of the 1500 m2 field plot for runoff monitoring (average soil layer thickness and slope are 40 cm and 6°, respectively).

around one month before monitoring of the first rain event (23rd June) in this study. 2.2. Sampling and data collection Surface runoff (SR) and fracture flow (FF) were collected. No sample of interflow was obtained in all observed rainfall events. Upon sampling, the average discharge was recorded with 15-min intervals by selfrecording event data logger (HOBO UA-003-64) connected to the tipping buckets, based on which the rising, peak and receding limb of the hydrography were captured. At the outlet of conflux grooves, samples of SR and FF were manually collected with time intervals varying from 5 to 60 min during rains. Pre-event fracture flow was also collected as the background flow. Flow samples were filtered using a 0.45-μm hydrophilic polypropylene filter immediately after collection, stored at 4 °C and analyzed within 24 h. As soil water conductivity changes with depth, a total of ten tensionmeters (T4e, UMS, München, Germany) were used for the measurement of soil water potential. They were installed at two depths of 15 and 25 cm of the upslope (U15/25), and at four depths of 15, 25, 35 and 45 cm of mid- and downslope (M/D 15/25/35/45), respectively, being representative of the whole soil profile across the 50 m-long plot. The tensionmeters were connected to a data logger (CR1000, Campbell Scientific, Inc., Logan, UT, USA) and soil water potential was recorded every 15 min. All electronic data were downloaded prior to and after each rainfall event. When water penetrates the soil, soil moisture increases till the soil becomes saturated. By that time, soil water potential increases to the value of zero. Afterwards, the soil may be in an oversaturated state that water moves under a positive pressure. 2.3. Analyses 2.3.1. DOC and spectroscopy measurement DOC concentration was measured using a continuous flow analyzer after acidification (pH = 2) of the sample (Auto Analyzer 3, SEAL Analytical, Norderstedt, Germany). Simultaneous measurement of UV– visible absorbance and EEM fluorescence was made by an Aqualog fluorescence spectrometer (Horiba JY, Japan). Milli-Q water-Raman-peak signal-to-noise and emission calibration validation were used to examine the wavelength calibration of the CCD detector in order to make sure the spectrometer was stable. Fluorescence intensity was measured across emission wavelengths 250–800 nm at excitation wavelength from 240 to 600 nm, with 1 nm increments and an integration time of 0.5 s. UV absorbance spectra from the wavelengths of 200 to 600 nm

387

was obtained and used for the correction of fluorescence EEMs. A Milli-Q EEM was used as blank to remove the instrument-specific spectral biases. Inner-filter effects, Raman bands and Rayleigh scatters were corrected by the built-in software. Typical EEM spectra of runoff samples as well as rainwater collected in this work were shown in Fig. S1 of the attached Supplementary information (SI). Rainwater had an average DOC range of 0.12–1.95 mg L−1 and by rough estimate, the DOM it contained accounted for b4–9% of the total runoff flux in the four rainfalls. On the basis of the corrected absorbance and EEM data, three optical indices were calculated to further describe the compositional characteristics of soil DOM: (1) SUVA254, the ratio of UV absorbance at 254 nm (m−1) and DOC concentration (mg-C L−1), is used to describe aromaticity of DOM (Weishaar et al., 2003); (2) Fluorescence index (FI), the ratio of the emission intensity at 470 nm and 520 nm at an excitation of 370 nm, reflects the relative contribution of microbial- (N1.9) and plant- (b 1.4) derived organic matter to the DOM pool (Cory and McKnight, 2005); and (3) Humification index (HIX) is the ratio of the peak integrated area under the emission spectra 435–480 nm and under the sum of the emission spectra 300–345 nm and 435–480 nm at an excitation of 254 nm (Ohno, 2002). 2.3.2. Parallel factor analysis of EEM data Parallel factor (PARAFAC) modeling was conducted using a commercial platform SOLO (Eigenvector Research Inc.). The concentration and spectra of DOM components were constrained to non-negative values, and unimodality constraint was applied to the emission spectra (Stedmon and Bro, 2008; Murphy et al., 2013). Primary Rayleigh scatter was set to missing values, Raman bands and secondary Rayleigh scatter were replaced by interpolated values (Andersen and Bro, 2003). Residuals, spectral loadings, core consistencies and split-half analysis were used to validate the identified components (Bro, 1997; Murphy et al., 2013). According to the shape and location of spectra peaks, fluorescent DOM components were identified after referring to previous studies. Fluorescence intensity at the maximum (Fmax) for each component was calculated by multiplying the maximum excitation loading and emission loading by the corresponding score (Murphy et al., 2013). 3. Results and discussion 3.1. Hydrological process Four rainfall events (I to IV) of different characteristics (rainfall amount, intensity and duration, as depicted in Table 1) were monitored during the rainy season in the summer of 2015. In general, when it begun to rain, soil moisture started to increase as indicated by the increase of soil water potential that represents the extent of soil saturation. When all soil pores were filled with water (soil water potential increased to zero), the soil became saturated with time lags varying from minutes to hours among different rains. During each hydrological process of the observed rainfalls, surface runoff and/or fracture flow occurred under certain runoff mechanisms. Their dynamics in correspondence to characteristics of rainfalls are shown in Fig. 2. In the event on June 23 (I, Fig. 2a), as soon as the rain started, soil water potential increased quickly at depths of 15 and 25 cm at all slope positions of the plot, indicating the rapid penetration of rainwater down through the soil plough layer. At the depth of 25 cm, soil water potential was lower in the upslope and increased more slowly as compared to mid- and downslope. In addition, soil in the deep layer of mid- and downslope (M35, M45 and D45) had relatively higher water potential (about −100 cm) when the rain started, and did not change significantly in the beginning of the first rainfall stage but increased quickly as the top soil layer reached saturation. Occurrence of surface runoff was observed at the maximum rainfall intensity (74.4 mm h−1) under the mechanism of infiltration excess runoff. At this point, the rainfall had exceeded the saturated soil hydraulic conductivity while

388


Table 1 Characteristics of four monitored rain events and corresponding flow depths of surface runoff and fracture flow. Rain events

PPa (mm)

Duration (h)

Rain amount (mm)

Average rainfall intensity (mm h−1)

I (Jun. 23) II (Aug. 7) III (Aug. 16) IV (Sept. 9)

9.6 – – 9.4

12.5 6 26.5 8

95.2 39.4 110.8 49.0

7.6 6.6 4.0 6.0

a b

Intensity peaks (mm h−1)

Flow depthb (mm)

1st

2nd

Surface runoff

Fracture flow

74.4 47.2 19.2 20.0

29.6 12.8 16.8 10.4

0.03 b0.01 1.28 1.08

14.96 0.20 34.56 9.81

PP: Preceded precipitation amount (3 days). Flow depth: The difference of cumulative discharge minus base flow divided by area of field plot.

Fig. 2. Dynamics of soil water potential, average discharge, DOC concentration and maximum fluorescence intensity (Fmax) of PARAFAC-identified components (C1 and C2) for surface runoff (SR) and fracture flow (FF) during (a–d) four observed rain events. (Soil water potential on three slope positions: U-upslope, M-midslope, and D-downslope; the following numbers represent the corresponding soil depths).


389

Fig. 2 (continued).

the soil was still unsaturated as indicated by the negative value of soil water potential. As the rain continued, on the other hand, a few surface runoff was observed again in correspondence to the second rainfall intensity peak (29.6 mm h−1) but under saturation excess mechanism as the plot was already saturated. As for the fracture flow that occurred in the well-developed subsurface fracture channels of mudrock underlying soil layer, the discharge showed two peaks in correspondence to the two rainfall intensity peaks but with hours of time lag. Once the first stage of rainfall ceased, fracture flow declined immediately. In the second rainfall stage, the discharge started to increase again and reached to a higher peak even though the rain had a less rainfall amount and lower intensity as compared to that in the first stage. This observation indicated that, once the soil became saturated, the fracture flow responded strongly to rainfalls. In that case, the antecedent soil moisture (indicative of the extent of soil saturation) decided the time lag between the flow peak and the

rainfall intensity peak, and the rainfall amount decided peak discharge of the fracture flow. After rain event I, there were long dry periods of 14 and 7 days prior to the commencement of rainfalls on August 7 (II, Fig. 2b) and 16 (III, Fig. 2c), respectively, which led to extremely low antecedent soil moisture (with corresponding water potential being −800 and −500 cm, respectively). The rain event II can be classified as a heavy storm due to short duration (6 h) and high rainfall peak intensity (47.2 mm h−1). Among the observed rain events, this is the only case when the soil did not reach saturation through duration of the rain due to the extremely low antecedent soil moisture. As shown in Fig. 2b, soil water potential started to increase with time lags of 30 min for the downslope (D15) and 90 min for the mid- and upslope topsoil (M15 and U15), respectively, from maximum rainfall intensity. Only three samples of surface runoff were obtained at the point of maximum rainfall intensity under infiltration excess mechanism. In

390


contrast, the discharge of fracture flow increased from 13.32 to a peak of 73.26 L h− 1, and maintained at this level even the rain was getting smaller thereafter. It can be therefore inferred that, under the condition of unsaturation, the fracture flow was dominated by the preferential flow that refers to the uneven and rapid movement of rainwater through soil macropores and fracture channels. Comparatively, rain event on August 16 (III, Fig. 2c) was characterized by long duration (26.5 h), high rainfall amount (110.8 mm) but low peak intensity (19.2 mm h− 1). Such rainfall intensity was far lower than the soil saturated hydraulic conductivity, which led to no occurrence of surface runoff for a long time (18 h) after the rain begun. Instead, only a sustained low discharge of fracture flow was observed in this stage. Soil water potential showed a ladder increase with each rainfall stage. Once the soil turned saturated, the surface runoff occurred under saturation excess mechanism. Thereafter, both surface runoff and fracture flow started to increase with similar trends in correspondence with the rainfall. The discharge was dominated by fracture flow that had a peak ca. seven-fold the maximum of surface runoff. Different from the preferential flow observed in rain event II, the rainwater displaced soil pore water and infiltrated through the soil layer in the form of piston flow. As the rain continued, the discharge increased quickly and reached the maximum of 7406 L h−1. This rain event made a fracture flow depth of 34.56 mm which was the highest among the monitored rainfalls. In the last rain event on September 9 (IV, Fig. 2d), both surface runoff and fracture flow had three respective flow peaks corresponding to the three rainfall peak intensities but occurred with certain time lags. At the point of first rainfall peak intensity (20 mm h−1), only downslope of the plot was nearly saturated, indicating that water contained in the downslope soil contributed primarily to form the first surface runoff and fracture flow peaks. Subsequently, mid-slope turned saturated at the second rainfall peak intensity (10.4 mm h−1), and resulted in the second peaks of flow discharge. As the rain reached the last peak intensity (9.6 mm h− 1), upslope (U25) finally became saturated and the discharge of surface runoff and fracture flow reached their last peaks of 972 and 3200 L h−1, respectively. This rain had demonstrated existence of subsurface lateral flow that infiltrated from upslope to downslope within the plot, leading to the gradual saturation from downslope to upslope during the rain. Overall, the soil followed a regular pattern that it became saturated from downslope to upslope, and from deep soil to topsoil layer, despite different characteristics of the monitored rainfalls. This bottom-up saturation was attributed to the decrease of soil saturated hydraulic conductivity with depth (i.e. plough layer N undisturbed layer N fractured mudrock) that is caused by pore structure differences along the soil profile. Depending on antecedent soil moisture, runoff generation mechanism varied among the rainfalls and even changed over time in the course of one rainfall event. Generally, the discharge was found to respond strongly to rainfalls, with rainfall intensity being the primary factor in determining dynamics of the surface runoff, and rainfall amount in determining that of the fracture flow, respectively. 3.2. Response of runoff DOM concentration to rainfalls As quantified by DOC measurement, soil dissolved organic matter of the collected runoff samples reached maximum concentrations of ca. 84, 16, 7 and 13 mg L−1 in four rain events, respectively, with varying dynamics observed. The significantly higher concentrations of rain event I were likely due to soil carbon accumulation in the long dry period before the rainy season. This phenomenon also occurs elsewhere as for instance some studies found that DOC levels increased the most during the initial storm events of the wet season in forest watersheds (Ward et al., 2012). Extra carbon source might also come from decomposition of wheat straw that was mixed into the surface soil through tillage in May. After a month-long period of microbial activities under high temperature and dry-wet cycles in the field, wheat straw decomposed

fast to release active carbon component that made up part of soil labile DOM. Rainwater DOC in the first rainfall after dry period was also a likely contributor due to the contained carbon-rich particulates washed from the atmosphere (Neu et al., 2016), but to a limited extent given the low DOC loads and weak fluorescent intensities of the rainwater in this work. As evidenced by sharp DOC decreases in the subsequent rains, however, these impacts were strongly dampened after June. It is also notable that DOC in the surface runoff was higher than that of the fracture flow. As overland flow represents a rapid pathway for C transport, it often contains the highest DOCs as compared to ground flows, and this is a common phenomenon in regions with highly saturated soils (Johnson et al., 2006). In fact, DOC was strongly correlated with the flow discharge (P b 0.01, n = 65). This is because the equilibrated dissolution of soil organic matter in mobile water contained in soil pores before rain event (pre-event water) is the major source for the detected DOM. Thus, in the process of rainwater infiltrating the soil to form runoff, the export of DOM occurs as a result of displacement of soil pore water with rainwater (event water). In all observed rains, peaks of rainfall intensity led to corresponding DOC peaks with almost no time lag. Upon reaching the peak concentration, DOC started to decrease with the rising of flow discharge. In several rain events, extra DOC peaks were observed after the rain ceased. In the course of a rainfall event, DOM export from the surface runoff usually stopped quickly once the rain ceased, while for the fracture flow, as the discharge decreased relatively slower, the pre-event water that contains DOM drained gradually from soil pores with pore size order from macropores to mesopores resulting in late DOC peaks. According to Zhang et al. (2015), there may be three potential DOC peaks for the fracture flow. The first one occurs during the ascending limb of hydrography and it is derived from the pre-event water in soil mesopores. The second and third peaks are possibly formed by fast drainage of soil macropores and slow drainage of mesopores during the recession (or falling) limb of hydrography, respectively. Besides, because of the time difference of soil saturation of up-, mid- and downslope of the plot, the proportion of pre-event water in the fracture flow varied at the plot scale, making dynamics of DOC even more complicated. As exemplified by rain event I, the highest DOC concentration of 83.6 mg L−1 in the second rainfall stage was mainly contributed by pre-event water from mid-slope soil layers that had reached saturation as indicated by soil water potential (Fig. 2a). In rain event II, the very low DOC peak of the fracture flow was due to the fact that the flow was dominated by rainwater as preferential flow was the major hydrological path during the rain (Fig. 2b). In rain event III, under the mechanism of piston flow, DOC concentration of the fracture flow increased quickly after the soil was saturated, while the surface runoff showed opposite dynamics due to dilution effect of rainwater under a mechanism of saturation excess runoff (Fig. 2c). Similar trends were observed in the first runoff stage of the rain event IV, while DOC of the fracture flow started to decrease at the falling limb of the hydrograph (Fig. 2d). In these rainfall events, soil DOM primarily migrated via the fracture flow path. This is different from findings that surface runoff mainly contributed to enhance DOM export during rainfalls in a large variety of ecosystems, in which case flow paths shifted from base flow to shallow depths via high discharge of surface runoff flushing out topsoil organic matter (Hu et al., 2016). Our results indicated that, even though DOCs were mostly higher in the surface runoff, the absolute high discharge of fracture flow made underground path the major way for DOM export from the studied soil. This was determined by soil features including its thin soil layer, abundant distribution of macropores and easily-formed fractures in the parent mudstone that together facilitated the generation of subsurface flow (Zhu et al., 2009). The results demonstrated the significance of understanding the underground contribution for DOM migration in the studied system as well as regions where stream flow is largely supplied by subsurface and base flow (Ward et al., 2017). In addition, as DOM originates from pre-event soil pore water, the above results have indicated that, for both surface runoff and


fracture flow, DOC dynamics is indeed the result of mixing process of soil pre-event water and rainwater. Rainwater itself is supposed to be a carbon source via certain processes such as solubilization of atmospheric particles, throughfall and stemflow in forestland (Stubbins et al., 2017), but was not counted due to its low loads relative to the runoff flux in the agricultural settings of this study. The concentration of DOC then increases with the proportion of pre-event water that is determined by hydrological paths.

3.3. DOM's spectroscopic characteristics and dynamic changes during rainfalls As shown in Fig. 3, the results of EEM-PARAFAC analyses indicated that the DOM consists of two major components with peaks located at the excitation/emission wavelengths (Ex/Em) of 270(380)/480 (component 1, C1) and 250(320)/410 nm (component 2, C2), respectively. It should be noted that even though intensities in the region of protein-like components can be observed in the EEMs (SI Fig. S1) but they are too weak to be distinguished independently in the model. The components have been previously identified in the literature. For instance, C1 was similar with C2 reported by Stedmon and Markager (2005) and C3 by Kothawala et al. (2014), which is a typical UVA humic-like component consisting of fulvic acid fluorophores with

Fig. 3. The split-half validated results of excitation (Ex) and emission (Em) loadings for two PARAFAC-identified DOM components. Corresponding contour plots of the same components are shown in the insets.

391

terrestrial/autochthonous origin, primarily in soil-derived DOM and ubiquitously existing in wide range of freshwater environments. C2 identified in this study was similar with C6 reported by Stedmon and Markager (2005), C2 by Williams et al. (2010), C2 by Hiriart-Baer et al. (2013) and C3 by Harun et al. (2016), which also belongs to UVA humic-like species consisting of humic fluorophores with anthropogenic origin and commonly presents in agriculturally dominated watersheds. Concentrations of the two components in runoff samples and their response to rainfalls as evaluated in terms of the maximum fluorescence intensities (Fmax) are shown in Fig. 2. Overall, in all observed rains, contents of both components in the surface runoff were higher than that in the fracture flow. This is also the case of rain event I, thus indicating that the surface runoff always contains a higher proportion of fluorescent DOM (FDOM) species even when the sample has a significantly lower DOC concentration than the fracture flow. It confirms that the chromophoric DOM is mainly originated from topsoil organic matter other than the deeper layers and the underlying mudrock. The decrease of the FOM/DOC ratio with soil depth has been reported previously. The preferential retaining of complex and aromatic FDOM on soil mineral horizons is a feasible mechanism; whereas more structurally simple and hydrophilic components can migrate towards deeper soil horizons (Kalbitz, 2001; Corvasce et al., 2006). Statistical analysis showed significant positive correlations of the concentrations of C1 and C2 with DOC values of the fracture flow in four rain events (R2 = 0.64, P b 0.01, Table S1 of the SI). Particularly high correlations were observed for the fracture flow collected in rain event IV (R2 = 0.91, P b 0.001, Fig. S2 of the SI). As for the surface runoff, on the other hand, such positive correlation was only found in rain events III and IV (R2 = 0.69 and 0.93, P b 0.01, respectively). We assumed that loss of the correlation was likely due to the very limited amount of samples collected in the first two rain events. Interference in DOC from the vigorous decomposition of wheat straw on soil surface may also be a potential cause. Additionally, a series of optical indicators including SUVA254, FI and HIX were calculated based on absorbance and fluorescence measurement. The results are shown in Figs. 4 and S3–S4. The average ranges of 1.52–1.61 and 0.87–0.91 for the FI and HIX values, respectively, have indicated the presence of a mixture of both microbial- and plantderived DOM sources with high humification degree from the soil organic matter pool. Among the observed rainfalls, these two indices did not show significant changes (P N 0.05, Table S1 and Fig. S3 of the SI). In contrast, changes in SUVA254 were considerable, suggesting high sensitivity of the index in capturing variations of DOM's aromaticity during rainfalls. SUVA254 values were generally higher for the surface runoff than the fracture flow (Fig. 4b), indicating more aromatic and possibly more DOM species with high molecular weight that absorb at 254 nm exporting the soil with the surface runoff (Weishaar et al., 2003; Jaffé et al., 2008). Thus, changes of the index reflected proportional variation of this fraction of DOM in accordance with changes of water sources under given hydrological processes. Additionally, for both surface runoff and fracture flow, SUVA254 showed an increasing trend though the observed rainfalls, suggesting an increase of DOM aromaticity with sustained rainfalls during the summer rainy season. In order to further examine DOM's compositional variations during rains, the ratio of Fmax values of C1 and C2 (referred to as “C1/C2”) was calculated to show their relative distributions in the collected runoff samples. Fig. 5 shows the ratio values as a function of duration of each observed rain event. The surface runoff was generally higher than the fracture flow. In fact, as depicted in the SI Fig. S5, the C1/C2 ratio decreased gradually with soil depth from the average range of 0.84–0.91 in the topsoil layer (5 cm) to the range of 0.78–0.84 at the bottom of the soil profile (40 cm). In the deeper layer of mudrock underlying, the ratios for the fracture flow had decreased to the range of 0.44–0.89, indicating that the relative abundance of C1 in the chromophoric DOM matrix was the largest in the surface runoff and the lowest in the fracture flow, with soil pore water being in-between. Additionally,

392


significant increases in the C1/C2 ratio during precipitation were observed for both the surface runoff and fracture flow. This was found to occur at the moment the soil reached saturation. On the one hand, the ratio for the initial surface runoff samples of rain events mostly captured “signals” of rainwater because the flow was dominated by rainwater under the infiltration excess mechanism. The ratio then increased as the flow contained increasing proportion of pre-event water that had relatively higher C1/C2 ratios than rainwater. The subsequent decrease of the ratio was attributed to increased rainwater distributions as the rain continued after soil saturation. As for the fracture flow, on the other hand, the increase of the ratio was due to contribution of both soil water and rainwater from upper layers via matrix flow after soil saturation. The above results have indicated that, variation of the relative distribution of DOM components during the course of rain events was indeed a result of changes in the water sources in the runoff formation. Water stable isotopes were measured in this study (data not shown). According to statistics, the C1/C2 ratio was significantly correlated with the δ18O values (R2 = 0.37, P b 0.001 for surface runoff and R2 = 0.18, P b 0.001 for fracture flow, respectively, Fig. S6 of the SI). Similar results were also found by use of the δ2H values. Correlations between these isotopic values and other calculated optical parameters were also established in this study (Table S1 of the SI). Such correlations have been reported previously in a catchment scale study, implying that DOM could act as a tracer for groundwater-surface water exchanges based on differences of DOM fluorescence characteristics between the two pools (Hu et al., 2016). As such, it can be inferred that information on the optical properties of DOM components can be used to decipher the water pathways during rains. The fast and easily-operated spectroscopic measurement can therefore be a cost-effective alternative to the expensive and time-consuming measurement of chemical tracers such as isotopic species; while its theoretical feasibility and wide applicability require further verification in future studies. 4. Conclusions

Fig. 4. (a) An example of dynamics of specific UV absorbance index (SUVA254) for surface runoff (SR) and fracture flow (FF) responding to a rainfall event and (b) average values of the index in four observed rain events. Data of the index's dynamics for other rain events are shown in Fig. S3 of the SI section.

Based on hydrological process analyses and measurement of spectroscopic metrics, this study provided new insights into the dynamics and patterns of DOM constituents migrating from an agricultural soil during rainfalls. Although DOC concentrations were mostly higher in the surface runoff, underground migration was the primary path for DOM export as the discharge was dominated by the fracture flow in all observed rain events. The DOM in the runoff responded strongly to rains, with peaks of rainfall intensity leading to corresponding DOC peaks with almost no time lags. Significant positive correlations were established between the concentrations of the PARAFAC-identified components (C1 and C2) and the DOCs for the fracture flow and also for the surface runoff that occurred under saturation excess mechanism. It was also found that variation of the relative distribution of DOM components (as exemplified by the C1/C2 ratio in this study) reflected changes of water sources. Based on their potential correlations, the dynamics of DOM's characteristics can be indicative of the water pathways during rainfalls. Findings of this study demonstrated the complex hydrological variations in the soil among rainfalls and the importance of understanding underground dynamics of DOM migration; the spectroscopic techniques may perform well for accomplishing the task, not only for the studied agricultural soil but for other land uses and soil systems. Acknowledgements

Fig. 5. Variation of relative distribution of PARAFAC-identified components of DOM as expressed by the ratio of Fmax of C1 and C2 for surface runoff (solid dots) and fracture flow (hollow dots) collected during four observed rain events (I to IV).

Support for this work is provided by the National Key Research and Development Program of China (Grant Nos. 2017YFD0800101 and 2016YFD0800203), the National Natural Science Foundation of China (Grant Nos. 41471268 and 41771521) and the 135 Strategic Program of the Institute of Mountain Hazards and Environment, CAS (Grant No. SDS-135-1702).


Appendix A. Supplementary data Supplementary information (SI) to this article can be found online at https://doi.org/10.1016/j.scitotenv.2017.12.010. References Andersen, C.M., Bro, R., 2003. Practical aspects of PARAFAC modeling of fluorescence excitation-emission data. J. Chemom. 17, 200–215. Bro, R., 1997. PARAFAC. Tutorial and applications. Chemom. Intell. Lab. Syst. 38, 149–171. Buffam, I., 2001. A stormflow/baseflow comparison of dissolved organic matter concentrations and bioavailability in an appalachian stream. Biogeochemistry 53, 269–306. Corvasce, M., Zsolnay, A., D'Orazio, V., Lopez, R., Miano, T.M., 2006. Characterization of water extractable organic matter in a deep soil profile. Chemosphere 62, 1583–1590. Cory, R.M., McKnight, D.M., 2005. Fluorescence spectroscopy reveals ubiquitous presence of oxidized and reduced quinones in dissolved organic matter. Environ. Sci. Technol. 39, 8142–8149. D'amore, D.V., Fellman, J.B., Edwards, R.T., Hood, E., 2010. Controls on dissolved organic matter concentrations in soils and streams from a forested wetland and sloping bog in southeast Alaska. Ecohydrology 3, 249–261. Gonçalves-Araujo, R., Granskog, M.A., Bracher, A., Azetsu-Scott, K., Dodd, P.A., Stedmon, C.A., 2016. Using fluorescent dissolved organic matter to trace and distinguish the origin of Arctic surface waters. Sci. Rep. 6, 33978. Gong, Z.T., 1999. Chinese Soil Taxonomy. Sci. Press, Beijing. Gremillion, P., Gonyeau, A., Wanielista, M., 2000. Application of alternative hydrograph separation models to detect changes in flow paths in a watershed undergoing urban development. Hydrol. Process. 14, 1485–1501. Harun, S., Baker, A., Bradley, C., Pinay, G., 2016. Spatial and seasonal variations in the composition of dissolved organic matter in a tropical catchment: the Lower Kinabatangan River, Sabah, Malaysia. Environ. Sci.: Processes Impacts 18, 137–150. Hiriart-Baer, V., Binding, C., Howell, T.E., 2013. Dissolved organic matter quantity and quality in Lake Simcoe compared to two other large lakes in southern Ontario. Inland Waters 3, 139–152. Hood, E., Gooseff, M.N., Johnson, S.L., 2006. Changes in the character of stream water dissolved organic carbon during flushing in three small watersheds, Oregon. J. Geophys. Res. 111. Hu, Y., Lu, Y.H., Edmonds, J.W., et al., 2016. Hydrological and land use control of watershed exports of dissolved organic matter in a large arid river basin in northwestern China. J. Geophys. Res. Biogeosci. 121, 466–478. Inamdar, S., Singh, S., Dutta, S., Levia, D., Mitchell, M., Scott, D., et al., 2011. Fluorescence characteristics and sources of dissolved organic matter for stream water during storm events in a forested mid-Atlantic watershed. J. Geophys. Res. 116. Jaffé, R., McKnight, D., Maie, N., Cory, R., McDowell, W.H., Campbell, J.L., 2008. Spatial and temporal variations in DOM composition in ecosystems: the importance of long-term monitoring of optical properties. J. Geophys. Res. 113. Jeong, J.J., Bartsch, S., Fleckenstein, J.H., Matzner, E., Tenhunen, J.D., Lee, S.D., Park, S.P., Park, J.H., 2012. Differential storm responses of dissolved and particulate organic carbon in a mountainous headwater stream, investigated by high frequency, in situ optical measurements. J. Geophys. Res. 117, G03013. Johnson, M.S., Lehmann, J., Selva, E.C., Abdo, M., Riha, S., Couto, E.G., 2006. Organic carbon fluxes within and streamwater exports from headwater catchments in the southern Amazon. Hydrol. Process. 20, 2599–2614. Kalbitz, K., 2001. Properties of organic matter in soil solution in a german fen area as dependent on land use and depth. Geoderma 104, 203–214. Kothawala, D.N., Stedmon, C.A., Muller, R.A., Weyhenmeyer, G.A., Kohler, S.J., Tranvik, L.J., 2014. Controls of dissolved organic matter quality: evidence from a large-scale boreal lake survey. Glob. Chang. Biol. 20, 1101–1114. Larsen, L., Harvey, J., Skalak, K., Goodman, M., 2015. Fluorescence-based source tracking of organic sediment in restored and unrestored urban streams. Limnol. Oceanogr. 60, 1439–1461. Matilainen, A., Gjessing, E.T., Lahtinen, T., Hed, L., Bhatnagar, A., Sillanpää, M., 2011. An overview of the methods used in the characterisation of natural organic matter (NOM) in relation to drinking water treatment. Chemosphere 83, 1431–1442. McDonnell, J.J., 2003. Where does water go when it rains? Moving beyond the variable source area concept of rainfall-runoff response. Hydrol. Process. 17, 1869–1875. Mcelmurry, S.P., Long, D.T., Voice, T.C., 2014. Stormwater dissolved organic matter: influence of land cover and environmental factors. Environ. Sci. Technol. 48, 45–53.

393

Mladenov, N., Zheng, Y., Miller, M.P., Nemergut, D.R., Legg, T., Simone, B., et al., 2010. Dissolved organic matter sources and consequences for iron and arsenic mobilization in Bangladesh aquifers. Environ. Sci. Technol. 44, 123–128. Murphy, K.R., Stedmon, C.A., Graeber, D., Bro, R., 2013. Fluorescence spectroscopy and multi-way techniques. PARAFAC. Anal. Methods 5, 6557–6566. Neu, V., Ward, N.D., Krusche, A.V., Neill, C., 2016. Dissolved organic and inorganic carbon flow paths in an Amazonian transitional forest. Front. Mar. Sci. 3, 114. Nimptsch, J., Woelfl, S., Osorio, S., Valenzuela, J., Ebersbach, P., von Tuempling, W., et al., 2015. Tracing dissolved organic matter (DOM) from land-based aquaculture systems in North Patagonian streams. Sci. Total Environ. 537, 129–138. Ohno, T., 2002. Fluorescence inner-filtering correction for determining the humification index of dissolved organic matter. Environ. Sci. Technol. 36, 742–746. Raymond, P.A., Saiers, J.E., 2010. Event controlled DOC export from forested watersheds. Biogeochemistry 100, 197–209. Santos, P.S., Santos, E.B., Duarte, A.C., 2012. First spectroscopic study on the structural features of dissolved organic matter isolated from rainwater in different seasons. Sci. Total Environ. 426, 172–179. Singh, S., Inamdar, S., Mitchell, M., 2015. Changes in dissolved organic matter (DOM) amount and composition along nested headwater stream locations during baseflow and stormflow. Hydrol. Process. 29, 1505–1520. Singh, S., Dash, P., Silwal, S., Feng, G., Adeli, A., Moorhead, R.J., 2017. Influence of land use and land cover on the spatial variability of dissolved organic matter in multiple aquatic environments. Environ. Sci. Pollut. Res. 1–18. Stedmon, C.A., Bro, R., 2008. Characterizing dissolved organic matter fluorescence with parallel factor analysis: a tutorial. Limnol. Oceanogr. Methods 6, 572–579. Stedmon, C.A., Markager, S., 2005. Resolving the variability in dissolved organic matter fluorescence in a temperate estuary and its catchment using PARAFAC analysis. Limnol. Oceanogr. 50, 686–697. Stedmon, C.A., Markager, S., Bro, R., 2003. Tracing dissolved organic matter in aquatic environments using a new approach to fluorescence spectroscopy. Mar. Chem. 82, 239–254. Stubbins, A., Silva, L.M., Dittmar, T., Van Stan, J.T.I., 2017. Molecular and optical properties of tree-derived dissolved organic matter in throughfall and stemflow from live oaks and eastern red cedar. Front. Earth Sci. 5, 22. Tang, X., Zhu, B., Katou, H., 2012. A review of rapid transport of pesticides from sloping farmland to surface waters: processes and mitigation strategies. J. Environ. Sci. 24, 351–361. Wang, H.L., 2013. Study on Soil Hydraulic Parameters of Forest Land and Sloping Farmland in the Hilly Area of central Sichuan Basin. Northwest A & F University, Yangling, China (M. Sc. Thesis). Ward, N.D., Keil, R.G., Richey, J.E., 2012. Temporal variation in river nutrient and dissolved lignin phenol concentrations and the impact of storm events on nutrient loading to Hood Canal, Washington, USA. Biogeochemistry 111, 629–645. Ward, N.D., Bianchi, T.S., Medeiros, P.M., Seidel, M., Richey, J.E., Keil, R.G., Sawakuchi, H.O., 2017. Where carbon goes when water flows: carbon cycling across the aquatic continuum. Front. Mar. Sci. 4, 7. Weishaar, J.L., Aiken, G.R., Bergamaschi, B.A., Fram, M.S., Fujii, R., Mopper, K., 2003. Evaluation of specific ultraviolet absorbance as an indicator of the chemical composition and reactivity of dissolved organic carbon. Environ. Sci. Technol. 37, 4702–4708. Williams, C.J., Yamashita, Y., Wilson, H.F., Jaffé, R., Xenopoulos, M.A., 2010. Unraveling the role of land use and microbial activity in shaping dissolved organic matter characteristics in stream ecosystems. Limnol. Oceanogr. 55, 1159–1171. Yang, L., Guo, W., Chen, N., Hong, H., Huang, J., Xu, J., et al., 2013. Influence of a summer storm event on the flux and composition of dissolved organic matter in a subtropical river, China. Appl. Geochem. 28, 164–171. Zhang, W., Tang, X.Y., Weisbrod, N., Zhao, P., Reid, B.J., 2015. A coupled field study of subsurface fracture flow and colloid transport. J. Hydrol. 524, 476–488. Zhang, W., Tang, X.Y., Xian, Q.S., Weisbrod, N., Yang, J.E., Wang, H.L., 2016. A field study of colloid transport in surface and subsurface flows. J. Hydrol. 542, 101–114. Zhao, P., Tang, X.Y., Zhao, P., Wang, C., Tang, J.L., 2013. Tracing water flow from sloping farmland to streams using oxygen-18 isotope to study a small agricultural catchment in southwest China. Soil Tillage Res. 134, 180–194. Zhu, B., Wang, T., Kuang, F.H., Luo, Z.X., Tang, J.L., Xu, T.P., 2009. Measurements of nitrate leaching from a hillslope cropland in the Central Sichuan Basin, China. Soil Sci. Soc. Am. J. 73, 1419–1426.