Stable isotopes of nitrate reveal different nitrogen processing mechanisms in streams across a land use gradient during wet and dry periods

Understanding the relationship between land use and the dynamics of nitrate (NO3) is the key to constrain sources 10 of NO3 export in order to aid effective management of waterways. In this study, isotopic compositions of NO3 (δN-NO3 and δO-NO3) were used to elucidate the effects of land use (agriculture in particular) and rainfall on the major sources and sinks of NO3 within the Westernport catchment, Victoria, Australia. This study is one of the very few studies carried out in temperate regions with highly stochastic rainfall patterns; enabling a more comprehensive understanding of the applications of NO3 isotopes in catchment ecosystems with different climatic conditions. Longitudinal samples were collected from five 15 streams with different agriculture land use intensities on five occasions – three during dry periods and two during wet periods. At the catchment scale, we observed significant positive relationships between NO3 concentrations, δN-NO3 and percentage agriculture reflecting the dominance of anthropogenic nitrogen inputs within the catchment. Different rainfall conditions appeared to be major controls on the predominance of the sources and transformation processes of NO3 in our study sites. Artificial fertiliser was the dominant source of NO3 during the wet periods while nitrified organic matter in sediment and 20 nitrified manure were more apparent sources of NO3 to the surface water during the dry periods. Denitrification was prevalent during the wet periods while uptake of NO3 by plants or algae was only observed during the dry periods in two streams. The outcome of this study suggests that effective reduction of NO3 load to the streams can only be achieved by prioritising management strategies based on different rainfall conditions.


Introduction
Anthropogenic sources of NO − 3 from catchments can pose substantial risk to the quality of freshwater ecosystems (Vitousek et al., 1997;Galloway et al., 2004).Over-enrichment of NO − 3 in freshwater systems is a major factor in development of algal blooms which often promote bottom water hypoxia and anoxia.Such anoxia intensifies nutrient recycling and can lead to disruption of ecosystem functioning and ultimately loss of biodiversity (Galloway et al., 2004;Carmago and Alonso, 2006).Freshwater streams are often sites for enhanced denitrification (Peterson et al., 2001;Barnes and Raymond, 2010).However, when NO − 3 loading from the catchment exceeds the removal and retention capacity of the streams, NO − 3 is transported to downstream receiving waters including estuaries and coastal embayments, which are often nitrogen-limited, further compounding the problem of eutrophication.
Understanding the sources, transport and sinks of NO − thus different fractionation effects to the residual NO − 3 pool (Chien et al., 1977;Billy et al., 2010).However, the lack of NO − 3 isotope studies in the Southern Hemisphere (Ohte, 2013) impedes a more thorough understanding of NO − 3 dynamics within catchment ecosystems.
Most previous studies investigating the relationship between land use and NO − 3 export using δ 15 N-NO − 3 and δ 18 O-NO − 3 have either focused on the seasonal or spatial variations in one stream, or used multiple streams with one site per stream (i.e.Mayer et al., 2002;Yevenes et al., 2016).Far fewer studies have incorporated longitudinal sampling of multiple streams over multiple seasons.Nitrate concentrations and concomitant isotopic signatures can change substantially, not only spatially but temporally.Changes in hydrological and physicochemical (notably temperature) conditions of a river can affect the relative contribution of different sources of NO − 3 and the seasonal predominance of a specific source (Kaushal et al., 2011;Panno et al., 2008).In some studies (e.g.Riha et al., 2014;Kaushal et al., 2011), denitrification and assimilation by plants and algae have been reported to be more prominent during the dry seasons compared to the wet seasons, but in other studies (e.g.Murdiyarso et al., 2010;Enanga et al., 2016) denitrification appeared to be more prevalent during the wet seasons as precipitation induces saturation of soils, resulting in oxygen depletion and thereby low redox potentials that favour denitrification.As such, if spatial and temporal variations of δ 15 N-NO − 3 and δ 18 O-NO − 3 are not considered thoroughly in a sampling regime, it can lead to misinterpretation of the origin and fate of NO − 3 .Proper consideration of the temporal variability of NO − 3 isotope signatures and transformation are particularly pertinent in catchments with highly stochastic rainfall patterns, such as Australia.
In this study, we examine both spatial and temporal variations of NO − 3 concentrations and isotopic compositions within and between five streams in five catchments spanning an agricultural land use gradient, enabling us to evaluate (1) the effects of agriculture land use on the sources and transformation processes of NO − 3 and (2) the effects of rainfall on the predominance of the sources and fate of NO − 3 in the catchments.

Study area
This study was undertaken using five major streams (Bass River, Lang Lang River, Bunyip River, Watsons Creek and Toomuc Creek) draining into Western Port (Fig. 1) which lies approximately 75 km south-east of Melbourne, Australia.Western Port is a nitrogen-limited coastal embayment recognised as a Ramsar site for migratory birds.The catchments in Western Port contain three marine national parks, highlighting its environmental and ecological significance.The The catchment overlies a multi-layered combined aquifer system.The main aquifer consists of Quaternary alluvial and dune deposit (average thickness of < 7 m) as well as Baxter, Sherwood and Yallock formations (average thickness between 20 and 175 m).These aquifers are generally unconfined with radial groundwater flow direction from the basin edge towards Western Port Bay.The hydrogeology of Western Port can be found in Carillo-Rivera (1975).
Five longitudinal surveys were carried out between April 2014 and May 2015, two during wet periods (14 April 2014; 15 May 2015 -the total rainfall for 5 days before sampling was between 45 and 65 mm) and three during dry periods (8 April 2014; 22 May 2014; 21 March 2015 -the total rainfall for 5 to 10 days before sampling was < 5 mm).A total of 21 sampling sites indicated in Fig. 1 were selected across a gradient of catchment land use intensity.The five streams were selected based on the extent and distribution of land use types between and within each stream sub-catchment (see Fig. S1 in the Supplement), thus enabling comparisons within and between the streams.
In this study, catchment-intensive agriculture was used as a predictor of land use intensity in the catchment.These data were obtained from the National Environmental Stream Attributes database v1.1 (Stein et al., 2014), Bureau of Rural Sciences' 2005/06 Land Use of Australia V4 maps (http:// www.agriculture.gov.au/abares/aclump, last access: 30 April 2016) and Victorian Resources Online (VRO).In the context of this study, the catchment-intensive agriculture variable is termed "percentage agriculture".This term represents the percentage of the catchment subject to intensive animal production, intensive plant production (horticulture and irrigated cropping) and grazing of modified pastures.This variable also reflects the integrated diffuse sources of nutrients derived from intense agriculture including animal manure and inorganic fertilisers.The percentage agriculture for the sampling sites ranged between 2 and 96 % with the Bass River (94 ± 2 %) > Lang Lang (79 ± 5 %) > Watsons (76 ± 4 %) > Toomuc (71 ± 16 %) > Bunyip (upper Bunyip: 12 ± 9 %; lower Bunyip: 54 ± 10 %; Fig. 2).For the purpose of this study, Bunyip is divided into two sectors (upper and lower Bunyip) based on the proximity of the sampling sites (Fig. 1) and the percentage of land use.All the sampling sites in the upper Bunyip are situated in areas with > 30 % forestation (see Fig. S1).In general, the percentage agriculture decreases with increased distance from Western Port Bay (WPB) for all the streams except Bass River.There is an increase of about 2 % in percentage agriculture for Bass River with increased distance from WPB. Watsons Creek has the largest percentage of market gardens (∼ 91 %).

Sample collection and preservation
Water quality parameters (pH, electrical conductivity, turbidity, dissolved oxygen (DO) concentration and water temperature) were measured using a calibrated Horiba U-10 multimeter.Stream samples were collected for the analyses of dissolved inorganic nutrients (DIN) (ammonium, NH + 4 ; NO − 3 and nitrite, NO − 2 ), dissolved organic carbon (DOC) and NO − 3 isotopes (δ 15 N-NO − 3 and δ 18 O-NO − 3 ).These samples were filtered on site using 0.2 µm Pall Supor ® membrane disc filters.Filtered DOC samples were acidified to pH < 2 with concentrated hydrochloric acid.Samples for δ 18 O-H 2 O were collected directly from the streams without filtering.Sediment samples were collected from the bottom of the rivers and were kept in zip-lock bags.All samples were stored and transported on ice until they were refrigerated (nutrient samples were frozen) in the laboratory.In addition to stream water and sediment, we also collected four samples of artificial/inorganic fertiliser (from the fertiliser distributor in the area) and five of cow manure (from local farmers).

DIN and DOC concentration measurements
All chemical analyses were performed within 1-2 weeks of sample collection except for isotope analyses (within 2 months).The concentrations of NO − 3 , NO − 2 , and NH + 4 were determined spectrophotometrically using a Lachat QuikChem 8000 Flow Injection Analyzer (FIA) following standard procedures (APHA 2005).DOC concentrations were determined using a Shimadzu TOC-5000 Total Organic Carbon analyser.Analysis of standard reference materials indicated the accuracy of the spectrophotometric analyses and the TOC analyser was always within 2 % relative error.

Isotopic analyses
The samples for δ 15 N-NO − 3 and δ 18 O-NO − 3 were analysed using the chemical azide method based on the procedure outlined in McIlvin et al. (2005).In brief, NO x (NO − 3 + NO − 2 ) was quantitatively converted to NO − 2 using cadmium reduction and then to N 2 O using sodium azide.The initial NO − 2 concentrations were insignificant, typically < 1 % relative to NO − 3 .Hence, the influence of δ 15 N-NO − 2 was negligible and the measured δ 15 N-N 2 O represents the signature of δ 15 N-NO − 3 .The resultant N 2 O was then analysed on a Hydra 20-22 continuous flow isotope ratio mass spectrometer (CF-IRMS; Sercon Ltd., UK) interfaced to a cryoprep system (Sercon Ltd., UK).Nitrogen and oxygen isotope ratios are reported in per mil (‰) relative to atmospheric air (AIR) and Vienna Standard Mean Ocean Water (VSMOW), respectively.The external reproducibility of the isotopic analyses lies within ±0.5 ‰ for δ 15 N and ±0.3 ‰ for δ 18 O.The international reference materials used were USGS32, USGS 34, USGS 35 and IAEA-NO − 3 .Lab-internal standards (KNO − 3 and NaNO − 2 ) with pre-determined isotopic values were also processed the same way as the samples to check on the efficiency of the analytical method.The δ 18 O-H 2 O values were measured via equilibration with He-CO 2 at 32 • C for 24 to 48 h in a Finnigan MAT Gas Bench and then analysed using CF-IRMS.The δ 18 O-H 2 O values were referenced to internal laboratory standards, which were calibrated using VSMOW and Standard Light Antarctic Precipitation.Measurement of two sets of triplicate samples in every run showed a precision of 0.2 ‰ for δ 18 O-H 2 O. Sediment samples for the analysis of δ 15 N of total nitrogen were dried at 60 • C before being analysed on the 20-22 CF-IRMS coupled to an elemental analyser (Sercon Ltd., UK).The precision of the elemental analysis and δ 15 N was 0.5 µg and ±0.2 ‰ (n = 5), respectively.

Data analysis
The relationships between percentage agriculture and surface water NO − 3 concentrations were assessed using linear regression.Percentage agriculture was the predictor variable.NO − 3 concentration and δ 15 N-NO − 3 were response variables.Relationships between δ 15 N-NO − 3 and NO − 3 concentration as well as δ 18 O-NO − 3 and δ 15 N-NO − 3 were assessed using Pearson's correlation.The NO − 3 isotopes' response variables were assessed at two spatial scales -individual stream and catchment scale.The catchment scale integrates data from all five studied streams.Any graphical patterns or relationships derived from using these scales represent processes that occur somewhere in the catchment either in the streams or prior to entering the streams with data from the individual stream likely to represent more localised processes to that particular stream.

Results
The streams were oxic throughout the course of our study period with % DO saturation between 60 and 110 % (see Fig. S2 in the Supplement).There was no apparent spatial and temporal variation in DO; however, % DO saturation was slightly lower during the dry periods (average of 73 ± 20 %) compared to the wet periods (average of 82 ± 12 %).Temperature was also relatively consistent, with an average of 13 ± 2 • C. Ammonium concentration was generally low (< 4 µM) except during the wet periods in Bunyip (∼ 7 µM), Lang Lang (∼ 21 µM) and Bass (∼ 29 µM).DOC concentrations were typically 0.8 ± 0.4 mM.Nitrite concentrations were also low in all the streams, ranging between 0.1 and 0.4 µmol L −1 .
The spatial and temporal variations of NO − 3 concentration, δ 15 N and δ 18 O across the sites are shown in Fig. 3. NO − 3 concentrations ranged between 7 and 790 µM with averages of 21 ± 15, 50 ± 130, 64 ± 43, 71 ± 43 and 190 ± 280 µM for Toomuc, Bunyip, Bass, Lang Lang and Watsons, respectively.The lowest NO − 3 concentration was observed in the lower Bunyip (4 µM), while the highest NO − 3 concentration was observed in Watsons Creek (790 µM) at the most downstream site.Nitrate concentrations were generally higher during the wet periods compared to the dry periods in all streams +4 to +5 -Stream water -−5.5 to −4.9 (Fig. 3).During the wet periods, NO − 3 concentrations typically followed an increasing trend heading downstream except for the Bass River which exhibited the opposite NO − 3 trend with lower concentrations at downstream sites.During the dry periods, only the Bunyip and Bass rivers showed apparent longitudinal patterns in NO − 3 concentrations, with decreasing concentrations moving downstream in both.Sites with high-percentage agriculture generally also exhibited high NO − 3 concentrations (Fig. 4), particularly during the wet periods.

Potential sources of NO −
3 There are three major potential sources of NO − 3 in the catchments -artificial fertiliser, cow manure/organic fertiliser and soil organic matter (SOM) -see Table 1 for the δ 15 N-TN values.The average δ 15 N-TN value of soils is used to directly represent the soil organic portion as most of the nitrogen in soils is generally bound in organic forms.Nitrogen isotope of the NO − 3 produced from the potential end members usually retains the signature of the δ 15 N-TN as a result of tight coupling between mineralisation (production of am-monium from organic matter) and nitrification (oxidation of ammonium to NO − 3 ).The δ 18 O of NO − 3 generated by nitrification of these sources (δ 18 O-NO − 3 final ) is, however, decoupled from δ 15 N-NO − 3 .As shown in Eq. ( 1), which is adapted from Buchwald et al. (2012), δ 18 O-NO − 3 final relies on the oxygen isotope of water (δ 18 O-H 2 O), oxygen isotope of dissolved oxygen (δ 18 O-O 2 ), kinetic isotope fractionation associated with incorporation of oxygen during ammonia oxidation ( 18 ε k -O 2 ), kinetic isotope fractionation associated with incorporation of oxygen from water during ammonia oxidation ( 18 ε k -H 2 O, 1 ) and nitrite oxidation ( 18 ε k -H 2 O, 2 ), equilibrium isotope effect associated with oxygen isotope exchange between nitrite and water ( 18 ε eq ) as well as the fraction of nitrite oxygen atoms exchanged with H 2 O during ammonia oxidation (x AO ) (Casciotti et al., 2010;Buchwald et al., 2012).To date, 18 ε k -O 2 and 18 ε k -H 2 O cannot be separated.Previous culture studies have reported the overall 18 ε k -O 2 + 18 ε k -H 2 O, 1 to range between 17.9 and 37.6 ‰ (Casciotti et al., 2010), while 18 ε k -H 2 O, 2 ranged from 12.8 to 18.2 ‰ (Buchwald and Casciotti, 2010).These values together with a 18 ε eq value of 14 ‰, average δ 18 O-H 2 O of −5.3 ‰ and δ 18 O-O 2 of 23.5 ‰ were used to calculate the maximum and minimum estimates of the δ 18 O of newly produced NO − 3 from nitrification.The minimum estimate of δ 18 O-NO − 3 final was calculated using the lower range of 18 ε k -O 2 + 18 ε k -H 2 O, 1 (17.9 ‰) and 18 ε k -H 2 O, 2 (12.8 ‰), while the maximum estimate was calculated using the upper range of 18 ε k -O 2 + 18 ε k -H 2 O, 1 (37.6 ‰) and 18 ε k -H 2 O, 2 (18.2 ‰).Based on the assumptions that ammonia was fully oxidised to NO − 3 (as no accumulation of NO − 2 was observed during our study period) and there was complete exchange of oxygen isotope between nitrite and H 2 O during ammonia oxidation (x AO = 1), which likely characterises most freshwater systems (Casciotti et al., 2007;Snider et al., 2010;Buchwald and Casciotti, 2013); we calculated the δ 18 O of produced NO − 3 from nitrification to be between −2.03 and −0.23 ‰.
The δ 15 N-TN of cow manure (+6 to +13 ‰) was most variable compared to other end members.This variation reflects the extent of volatilisation, a highly fractionating process.Volatilisation can cause a fractionation effect of up to 25 ‰ in the residual NH + 4 (Hübner, 1986).As such, the lower value of +6 ‰ indicates a relatively fresh manure sample and is assumed to represent the initial δ 15 N of the cow manure before undergoing any extensive fractionation.
Atmospheric deposition did not appear to be an important source of NO − ing the wet periods, and from +1.5 to +13 ‰ during the dry periods) of the riverine samples.The δ 18 O-NO − 3 of atmospheric deposition were reported to range from +60 to +95 ‰ in the literature (Kendall, 2007;Elliott et al., 2007;Pardo et al., 2004).Similarly, groundwater was not considered as an important source of NO − 3 to the streams based on the low NO − 3 concentrations (∼ 0.7 to 7.0 µM) reported in previous studies (Water Information System Online; http: //data.water.vic.gov.au/monitoring.htm,last access: 29 April 2016).

General characteristics of NO − 3 in the streams
Agricultural land use (i.e. market gardens and cattle rearing) appeared to influence NO − 3 concentrations in our study sites.As shown in Fig. 4a, during the wet periods, high NO − 3 concentrations (> 40 µM) were particularly observed at sites with more than 70 % agricultural land use.During the dry periods, although NO − 3 concentrations were generally lower than 36 µM, the outliers were observed at sites with more than 70 % agricultural land use.Similarly, enriched δ 15 N-NO − 3 in the streams were mainly found at sites with highpercentage agricultural land use (between 75 and 85 %) for both dry and wet periods, suggesting that enriched δ 15 N-

Figure 1 .
Figure 1.Map of Western Port Bay (WPB) in southern Victoria, Australia, and major rivers discharging into WPB.Closed circles represent sampling sites where surface water samples were obtained.Values in parentheses represent the % agriculture area in the catchment.

Figure 2 .
Figure 2. The percent agriculture upstream for each of the sampling sites.

Figure 3 .
Figure 3. Spatial and temporal variations of nitrate concentrations and isotope values.Closed circles represent data obtained during the wet periods.Open circles represent data obtained during the dry periods.

Figure 5 .Figure 6 .
Figure 5. Conceptual diagram illustrating the sources and processes of NO − 3 during the wet and dry periods in the Western Port catchment.The values of the enrichment factor (ε) were obtained from the literature(Kendall et al., 2007) to indicate the relative contribution of the transformation processes to the isotopic compositions of the residual NO − 3 .

Table 1 .
The isotopic compositions of potential sources of NO −

Table 2 .
Comparison of NO − 3 concentrations and isotopes across different systems reported in the literature.