| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Agronomy and Range Science, University of California, Davis, CA 95616
* Corresponding author (bedardhaughn{at}ucdavis.edu)
Received for publication February 18, 2004.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: DON, dissolved organic nitrogen
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
There has been limited research on buffer efficiency and capacity in an extensively grazed irrigated pasture system. In California, irrigated pasture provides a relatively low-cost source of green forage during the summer months when surrounding rangelands are dry and dormant. Irrigation rates vary by irrigation method, but for flood irrigation are as high as 70 L s–1 at the top of the slope, applied continuously over an 8- to 14-h period. In the Sierra Nevada foothills, with slopes from 5 to 30%, this can generate runoff losses of up to 70% (Tate et al., 2000b). Given that irrigated pasture is both fertilized and grazed, there is concern that runoff water contains dangerous levels of pathogens and nutrients. This study is part of a larger project examining buffer effectiveness in irrigated pasture for attenuating N, P, and C, as well as indicator bacteria fecal coliforms and Escherichia coli. The component emphasized here is NO–3, a soluble nutrient commonly implicated in eutrophication in seawater and fresh water (Cole et al., 2004); NO–3 concentrations as low as 1 mg L–1 can contribute to algal blooms (Mendez et al., 1999).
Nitrate removal is typically attributed to denitrification, infiltration, or plant uptake. Denitrification, particularly in saturated riparian zones, is frequently viewed as the most effective way to prevent NO–3 contamination of surface and ground water (Casey et al., 2001; Hill, 1996). This presents two concerns. First, denitrification rates vary both spatially and temporally, creating predictive challenges (Hill, 1996). For example, Lowrance et al. (1995) observed much higher denitrification rates in grassed areas compared to either hardwood or pine forest buffers. They also observed significant temporal differences related to timing of N application. In addition, buffer design for NO–3 removal via denitrification can only be effective when site-specific hydraulic characteristics are taken into consideration (Aravena et al., 2002; Leeds-Harrison et al., 1999; Sabater et al., 2003). For example, Wigington et al. (2003) found that even high denitrification potential did not guarantee high levels of NO–3 removal because only a small percentage of the stream flow at their study site intersected riparian soils; the majority of the flow came from ephemeral swales. The second concern is that in landscapes receiving high NO–3 inputs, and where denitrification is dominant, riparian buffer zones can serve as significant sources for N2O, a greenhouse gas with a warming potential 300 times that of CO2 (Groffman et al., 1998, 2000; Hefting et al., 2003). In irrigated pasture, hydrologic patterns and associated denitrification potential can be difficult to characterize because they can change drastically with the rapid wet–dry cycles corresponding to irrigation events. Thus, it becomes important to consider the potential for removing NO–3 via infiltration and uptake as opposed to denitrification. As noted by Verchot et al. (1997), infiltration and vegetative uptake can be the dominant factors for attenuating nutrients in surface runoff.
Previous estimates of buffer NO–3 attenuation range broadly, from buffers serving as a net source of NO–3 to buffering effectiveness of >99% (Dillaha et al., 1989; Dosskey, 2001; Hill, 1996). A similarly broad range of 10 to 90% has been observed for NH+4 (Dillaha et al., 1989; Dosskey, 2001). Although there is very little data available on DON in surface runoff, Dosskey's (2001) review on buffer effectiveness indicates that for total N, buffers can be either a net sink (up to 91% reduction) or a net source, with up to 50% more total N flowing out of the buffer than into it. This broad range of effectiveness values may be attributable in part to the mechanism for N removal. With denitrification, NO–3 is removed from the terrestrial and aquatic systems; in contrast, infiltration and uptake may provide only ephemeral storage. Dillaha et al. (1989) attributed high levels of soluble nutrients leaving buffers to low trapping efficiency for soluble nutrients and release of nutrients previously trapped in the filter. Buffer trapping efficiency may decrease over time, and buffers may ultimately become a source of N rather than a sink (Mendez et al., 1999). The role of N cycling within the buffers must not be neglected when considering potential sources of N, NO–3 or otherwise.
Stable 15N isotopes are used to study the fate and transport of N. Previous buffer studies using 15N have focused on natural abundance methods, using naturally occurring variations in 15N levels to identify pollutant sources that were moving through buffers to adjacent waterways (Chang et al., 2002; Karr et al., 2003; Spruill et al., 2002), or to determine whether or not denitrification was a major factor in NO–3 removal (Dhondt et al., 2002; Ostrom et al., 2002). However, 15N natural abundance provides, at best, semiquantitative estimates of pathways and processes occurring in the field (Bedard-Haughn et al., 2003). If not completely accounted for, background variability in isotopic signatures and fractionating processes that alter those signatures to varying levels can confound interpretation of 15N data. Even when sources of variability are accounted for, natural abundance techniques do not allow differentiation between new sources of N and N already stored within the system. In contrast, using 15N-enriched isotopes allows new N sources to be quantitatively traced through the system and measured in the various potential sinks, and the 15N level of the applied tracer can be predetermined to ensure that the signature is detectable above background variability, even when fractionation occurs (Bedard-Haughn et al., 2003). Isotopic levels are reported as the amount of 15N present relative to the average naturally occurring background 15N levels for a given source. There have been a limited number of studies using 15N-enriched tracers in the field (Davidson et al., 1990; Di et al., 1999; Mulholland et al., 2000), due primarily to high tracer cost. We were unable to find any previous field studies that used 15N-enriched tracers to quantify buffer effectiveness for attenuating NO–3. Previous work by Matheson et al. (2002) to quantify the fate of 15N tracers in riparian zones was performed under controlled laboratory settings as microcosm studies. They determined that soil immobilization and plant assimilation accounted for less than half of the applied tracer; the remainder (61–63%) was assumed to have been lost via denitrification. They could not, however, account for any lateral or vertical movement that might occur in a natural field setting.
Given previous evidence suggesting that vegetative buffers themselves are acting as a pollutant source rather than a sink by releasing previously stored nutrients, the major objectives of this study were to determine (i) if buffers in irrigated pasture were effective in sequestering new sources of NO–3, (ii) where sequestered NO–3 was being stored, and (iii) whether the added NO–3 remained sequestered in the buffers or was subsequently lost, either as NO–3 or as a different form of N (i.e., buffer sustainability). The data were examined both for overall effectiveness in sequestering N over the course of the summer and for general trends in N uptake.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
|
Nitrogen-15 Application and Analysis
Nitrogen isotopes, which are stable and nonradioactive, have been used extensively to follow the dynamics of N in soils and crops (Powlson and Barraclough, 1993). We used 15N enriched material so that the added N could be detected and differentiated from inherent background variability in naturally occurring 15N levels (Bedard-Haughn et al., 2003). Natural abundance background levels of 15N in all N pools were measured before application of 15N-labeled fertilizer to account for natural variability and dilution of the applied 15N fertilizer by background 14N.
In July 2002, 15N-labeled KNO3 was applied in solution at a rate of 5 kg N ha–1 and 99.7 atom % 15N. The rate and atom % concentration were selected to provide an approximation of post-irrigation fertilizer N levels while allowing the tracer to be detectable in all N pools throughout the duration of the experiment. The 15N solution was applied across all nine plots along the entire width of the experiment. The area labeled was 1 by 5 m wide and located 0.75 m above the buffer areas. Following application, the labeled fertilizer was watered in with 20 L of water m–2. Watering in was done by hand with watering cans for maximum precision; 20 L represented the optimum amount to ensure that the applied 15N-labeled KNO3 was rinsed off of the foliar surfaces, but the volume was not so great as to cause deep leaching of the applied fertilizer. The 15N application area was fenced to prevent redistribution of the 15N-enriched material by the cattle.
For a 14-wk period following application, water samples were collected from the installed collection troughs during each irrigation trial (11-d schedule). Water samples (500 mL) were collected as "grab" samples from the V-notch at the end of each collection trough. Samples were taken at 0 (leading edge of runoff), 15, 30, 60, 90, and 120 min following commencement of runoff from each treatment and were stored frozen until analysis. This sampling scheme represented a minimum sample number and is based on previous experience with the timing of runoff and pollutant transport from these systems. At each sampling interval, runoff rate was determined by measuring the volume of runoff draining from the V-notch in the collection trough in a 5-s period. Runoff rate data were used to determine runoff losses (Table 1). Following each irrigation, vacuum was applied to the soil solution sampling tubes and allowed to draw moisture from the soil for 10 d (i.e., until the next irrigation). Although vacuum was not applied constantly over the 10-d period, suction was still present at sampling. Soil water samples were collected just before the subsequent irrigation and were stored frozen until analysis.
Runoff 15N isotope analyses were performed on all three N pools: NO–3, NH+4, and total N for Days 1, 12, 31, 65, and 86 following application of the tracer. For Days 1 and 12, only the 0-, 15-, 60-, and 120-min intervals were analyzed because preliminary experiments indicated that this was sufficient for characterization of maximum variation. For Days 31 to 86, even fewer intervals were needed to acquire sufficient information because there was no longer significant change between sampling days. Samples were filtered to remove sediment and vegetation residues from runoff. Ammonium 15N and NO–3–15N were determined by NH3 diffusion onto polytetrafluoroethylene-encased acid traps (Stark and Hart, 1996). To measure NO–3–15N, the Stark and Hart (1996) method was modified only slightly in that following diffusion of 100-mL samples for NH+4, 1 mL of 5 M NaOH was added to each to bring the pH up to 
12. Samples were heated uncovered at 95°C to remove any trace ammonium or labile organic N (DON) and to concentrate the volume down to 25 mL. In place of Devarda's alloy, TiCl3 (Fisherbrand Titanous Chloride Solution, 20%; Fisher Scientific, Hampton, NH) was then added (typically one-twentieth of the sample volume) to reduce NO–3 to NH3. Soil solution samples (25-mL aliquots) were analyzed for NO–3–15N via the TiCl3 diffusion as above, except no concentrating step was required. Titanous chloride has been found preferable to Devarda's alloy due to its low cost, low N contamination, and availability in solution form (Cho et al., 2002; Cresser, 1977; Crumpton et al., 1987). Samples were sealed and incubated at 50°C for 72 h. Nitrate standards with field-level N concentrations had mean N recovery of 94% (SD ± 5%) using this modified method.
Total 15N was determined on a separate 20-mL aliquot by performing a persulfate digestion (American Public Health Association, 1989) to convert the DON and NH+4 to NO–3, and samples were then diffused for NO–3 as above (without concentration step). The DON–15N for each sample was calculated using an isotope mixing model via difference from total 15N (Shearer and Kohl, 1993):
 | [1] |
Following diffusion, acid disks were removed from polytetrafluoroethylene packets and analyzed via mass spectrometry (Integrated Stable Isotope Analyzer; Europa Integra, Crewe, UK) at the University of California-Davis Stable Isotope Facility. The current sensitivity of our stable isotope ratio mass spectrometers is 0.0002 atom % 15N.
Representative plant samples from the pasture and buffer areas were taken before each irrigation trial. To determine how far the 15N fertilizer had moved into the buffer strip, plants were sampled across the width of the buffer at down slope intervals with a sample spacing of 1 m immediately above and below the zone of 15N application, and spacing of 2 m further into the buffer. The buffer vegetation samples were separated between grasses and verbena, the native shrub in the buffers. Following each grazing (every second irrigation), the fenced 15N application area was clipped and the vegetation removed to simulate grazing. All plant samples were oven-dried at 65°C and analyzed for 15N isotopic composition via mass spectrometry (van Kessel et al., 1994).
Soil samples were taken monthly to a 15-cm depth in two increments (0–7 and 7–15), corresponding to the depth of the A horizon. Samples were taken at 0, 1, and 5 m from the 15N application zone at 12, 43, and 86 d following 15N application. Samples were also taken at 8 and 16 m on Day 86. Sample quantity, depth, and diameter were limited due to concurrent sampling at the site to analyze total suspended sediment in runoff. Soil samples were oven-dried at 40°C and analyzed for total N and 15N via mass spectrometry.
Isotopic levels for the soils and plants are reported as atom % 15N excess, which refers to the amount of 15N present relative to the average naturally occurring background 15N levels for that particular source. Background levels are based on pre-application samples. Where possible, atom % 15N excess amounts were extrapolated to get the total amount of 15N in a given pool by weight and thus to determine a 15N budget. Note that it was not possible to perform budget calculations for the vegetation in the buffer areas as accurate biomass measurements over the course of the summer season would have required destructive sampling that would have confounded subsequent measurements.
Statistical Analysis
The results were analyzed using linear mixed effects model analysis. Linear mixed effects analysis can be applied to both structured and observational studies (Pinheiro and Bates, 2000) and was used here to account for the influence of both fixed (buffer treatment) and random (irrigation date) effects on buffer 15N uptake levels. Treating time as a random effect provided a direct test for whether buffered plots were significantly different from nonbuffered plots when results were considered over the duration of the study. The magnitude and direction (±) of the coefficient for buffer effect was used to define the relationship between 15N loading in runoff and buffer treatment. This approach allowed for robust evaluation of the data while accounting for the repeated measures (group effect–plot identity) embedded in the data structure. This flexible model also allowed within-group variance and correlation structures for handling within-group (plot) heteroscedasticity and temporally correlated errors (irrigation series within year) (Pinheiro and Bates, 2000). This approach has been used in modeling other complex longitudinal datasets (Atwill et al., 2002; Tate et al., 2000a, 2003).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The 15N runoff data showed that buffers were effective for sequestering new NO–3 in irrigated pasture over the course of the summer. The regression coefficients in Table 2 demonstrate that for NO–3, the 8-m buffer decreased 15N load by approximately 28% and the 16-m buffer decreased load by 42%. Indeed, regardless of the form of N, more 15N was lost from the nonbuffered irrigated pasture plots than from those with 8- or 16-m buffers. For NH+4, the decrease was 34% (8 m) and 48% (16 m), whereas for DON, the decrease was 21% (8 m) and 9% (16 m). The net effect on 15N load, illustrated by the total dissolved N analysis, is a decrease of 36% for the 8-m buffer and 28% for the 16-m buffer, suggesting that DON appears to be the limiting factor in the effectiveness of the buffers, particularly the 16-m buffers.
Nitrogen Sequestration
In considering vegetative effects, Schmitt et al. (1999) found that although sorghum [Sorghum bicolor (L.) Moench] was effective as a vegetative filter, grass buffers had no effect on the concentration of dissolved constituents. In this study, however, the primary mechanism for removal of applied 15N–NO–3 was plant uptake; specifically, grass uptake in the zone of 15N application. Within 10 d of application, approximately 40 to 50% of the tracer was removed by plant uptake and a further 23 to 27% was stored in soil immediately below the zone of application, accounting for up to 77% of the applied tracer. A further 3% of the applied tracer was observed in the runoff on the first day after application (1.5% from nonbuffered, 0.9% from 8-m buffers, 0.6% from 16-m buffers), resulting in total 15N recovery of up to 80% just within the pasture and runoff. This is much higher recovery than the 11 to 15% removed by plant uptake and 24 to 26% stored in the soil in the microcosm study by Matheson et al. (2002), which did not account for runoff. The level of plant N uptake in the pasture is less than the 72% measured by Griffith et al. (1997) in a grass-seed production system in western Oregon. However, our irrigated pasture uptake values reflect only the uptake of added 15N, not the uptake of N that was already in the system. The overall high levels of plant N uptake observed in irrigated pasture indicate that although there is not a shallow ground water table at this site, the presence of significant fine root biomass associated with the N–P–K-fertilized annual grasses improves the potential for plant uptake (Cheng and Bledsoe, 2002; Hill, 1996).
Further storage and uptake occurred within the buffer, particularly in the first few meters. Although the total amount sequestered in buffer vegetation could not be definitively quantified due to lack of precise biomass measurements, atom % 15N excess measurements suggest that maximum 15N uptake occurred in the first 4 m of the buffer areas and the overall uptake was less than that of the pasture. A conservative estimate is that approximately 3% of the 15N applied to the buffered treatments was taken up in the first 4 m. Only 1 to 2% of the applied 15N was stored in the upper 15 cm of the soil in the first 5 m of the buffer.
Given the downslope movement of soil 15N (Fig. 5), the expected pattern of plant 15N uptake was a gradual increase in plant 15N further downslope with each subsequent irrigation. Neither the grasses nor the verbena clearly demonstrated this pattern (Fig. 4), instead 15N enrichment decreased slightly as the vegetation took up non-enriched N. There are two possible explanations for this. The first, supported by the runoff data, is that the majority of downslope 15N movement was in the less plant-available DON form, so even though N is present, the plants cannot readily access it. The second is that the vegetation within the buffer was no longer taking up N. Maximum N uptake varies with the N status of the vegetation, NO–3 availability, and plant age. Plant uptake tends to decrease with plant age, which may be related to relative growth (Schenk, 1996). As Jackson et al. (1988) observed in the annual grasslands at the Sierra Foothill Research and Extension Center, even well-watered grasses can senesce within weeks of anthesis.
Buffer Sustainability
The ability of these buffers to remove new N stands in contrast to earlier findings by Tate et al. (2000b) that buffers are ineffective in reducing NO–3 concentrations in irrigated pasture. Although the buffers were effective over the course of the summer, the effectiveness varied in the first few weeks following tracer application. Runoff NO–3–15N was high in the first irrigation, but quickly decreased with subsequent irrigation; data indicate that NO–3 was just being cycled into the other N pools. Within one day of application, some of the NO–3–15N had already been transformed into NH+4 or DON, as shown by measurable levels of excess 15N in these forms. Hill (1996) found that in most riparian buffer studies, loss of NO–3 was not associated with increased NH+4 or DON, but these studies may be failing to recognize the importance of N cycling. By using stable 15N isotopes to examine nitrification rates in the annual grasslands at the Sierra Foothill Research and Extension Center, Davidson et al. (1990) showed that although the size of the NH+4 and NO–3 pools remains relatively constant over time, they turn over about once a day. They also showed that microbial assimilation of NO–3 occurs at rates similar to those for plant uptake, indicating that microbial assimilation of NO–3 is of much greater importance than previously recognized. The path of the 15N over the course of the summer indicates the rapid microbial immobilization of a portion of the applied 15N and its subsequent mineralization and nitrification contributes to the steady low levels of 15N that continue to be released from the buffers over the course of the summer.
This re-release of 15N that had previously been sequestered into the organic and inorganic N pools has contributed to the observation that buffers seem to decrease in effectiveness as more runoff events occur (Barling and Moore, 1994; Dosskey, 2002). As an example, the lower effectiveness of the 16-m buffer for attenuating DON may be attributed in part to the buffer itself acting as a substantial source for N (Dillaha et al., 1989; Mendez et al., 1999). As 15N that was initially stored in the soil beneath the pasture and buffer was gradually transferred downslope via surface and subsurface water movement (Fig. 5), the 16-m buffer had greater area for 15N to be stored initially, but its sequestration was transient. With subsequent irrigations, and particularly later in the irrigation event when runoff levels were at their maximum, more DON was released and transported in runoff. Similarly, the NO–3 and NH+4 were mineralized from the DON pool and mobilized via runoff during subsequent irrigation events. The NH+4 may have been particularly susceptible to nitrification during the dry periods between irrigation events (Barling and Moore, 1994). This pattern of N cycling within the pasture and buffer soils can account for the smaller peak of 15N released in the leading edge of the runoff during each irrigation event over the summer.
The corresponding decrease in the spring measurements of vegetation 15N levels with the stable measurement of 15N soil levels following the winter rainy season indicates that the 15N that was originally stored in the plants was subsequently returned to the soil via decomposition of plant materials during the cooler weather. Harvesting vegetation may remove sequestered nutrients from the buffer before they can be re-released into the system (Dosskey, 2001). Evidence suggests that within two weeks after the cutting of vegetation, uptake of N will increase due to increased NO–3 uptake and assimilation (Ourry et al., 1990). It may be that the limited ability of these buffers to take up new or old N later in the irrigation season could be improved by managing the buffer to increase demand for N.
As Sabater et al. (2003) observed, there can be a very large range of NO–3 removal efficiencies in buffers when NO–3 inputs are very low; when NO–3 inputs increase to greater than 5 mg L–1, NO–3 removal efficiency can decrease exponentially. Runoff NO–3 load in these irrigated pastures tends to be relatively low (<2 mg L–1), but increasing NO–3 inputs might result in much lower buffer efficiency.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The majority of the applied 15N was attenuated via plant uptake within the zone of 15N application; a smaller percentage was stored in the first few meters of the buffer vegetation. However, without proper planning, the N sequestered in vegetation may be lost to decomposition, resulting in net N losses. To maximize long-term effectiveness and sustainability of buffer, the potential for increasing vegetation demand and uptake through buffer management must be explored.
Over the course of the study, buffers were effective for attenuating NO–3–15N, slightly more effective for NH+4–15N, and least effective for DON–15N. For NO–3 and NH+4, the 16-m buffer was slightly more effective than the 8-m buffer, probably due to greater potential for plant N uptake. Nitrogen cycling within the soil was probably the major source of runoff mineral N later in the season. For DON, the 16-m buffer was actually less effective than the 8-m buffer, indicating that the 16-m buffers themselves were serving as a source for this less plant-available form of N.
Nutrients should always be managed first via in-field conservation practices; buffers should only be used as a secondary measure to capture excess. At this site, maximum differences between buffered and nonbuffered plots were observed primarily at the leading edge of irrigation events and in the first few weeks following fertilizer application. Proper timing and management of fertilizer application coupled with improved irrigation practices to decrease runoff could significantly reduce the potential for nutrient losses.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
15N isotopic method to indicate anthropogenic eutrophication in aquatic ecosystems. J. Environ. Qual. 33:124–132.Related articles in JEQ:
This article has been cited by other articles:
|
|
 |
|
  K. W. Tate, E. R. Atwill, J. W. Bartolome, and G. Nader Significant Escherichia coli Attenuation by Vegetative Buffers on Annual Grasslands J. Environ. Qual., April 3, 2006; 35(3): 795 - 805. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Bedard-Haughn, K. W. Tate, and C. van Kessel Quantifying the Impact of Regular Cutting on Vegetative Buffer Efficacy for Nitrogen-15 Sequestration J. Environ. Qual., August 9, 2005; 34(5): 1651 - 1664. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
