| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a U.S. Geological Survey, 333 West Nye Lane, Carson City, NV 89706 USA
b Wairakei Research Centre, Institute of Geological and Nuclear Sciences Ltd., Private Bag 2000, Taupo, New Zealand
c Horticulture and Food Research Institute of New Zealand Ltd., Palmerston North Research Centre, Private Bag 11030, Palmerston North, New Zealand
d Hawkes Bay Regional Council, Private Bag 6006, Napier, New Zealand
* Corresponding author (mrosen{at}usgs.gov)
1 The use of product names in this paper are for identification purposes only and do not imply endorsement by the USGS or any of the authors' institutions. 
Received 24 December 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSCONCLUSIONSREFERENCES |
|---|
Abbreviations: HBRC, Hawkes Bay Regional Council • SR, stocking rate
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSCONCLUSIONSREFERENCES |
|---|
The proper management of feedlot wastes has been an issue in the USA and Australia for a number of years (Gilbertson et al., 1981; Young et al., 1982, 1994), but this problem was only recently identified in New Zealand. The wastes are of major concern because they carry significant nutrient loads of both N and P, as well as some heavy metals such as Cu, Zn, and Fe. In New Zealand, the concentration of NO3–N in groundwater is one of the main concerns with feedlot waste.
For most of the history of European settlement in New Zealand, agriculture-based industries such as dairy and sheep farming have been dominant land uses. Except for small, localized areas on farms, such as sheep and cattle dips for the control of parasites (Hadfield and Smith, 1999), most of the environmental problems related to water quality have been as diffuse, nonpoint source issues related to overgrazing or fertilizing. Changes to farming practices in the 1980s brought about by international market pressures led to the introduction of cattle and sheep feedlots in New Zealand that are used to grain-feed animals before either slaughter or live shipment overseas (Young et al., 1994). Because of this trend, pollution problems related to the livestock industry now include point-source specific problems related to feedlot waste runoff and its impact on groundwater.
From July 1986 to 1998, a sheep feedlot was established for weaning the sheep off grass and onto pellets in preparation for live sheep exports. The feedlot, located approximately 2 km southwest of Maraekakaho Township, was about 1 km2 and held up to 80000 sheep for up to 10 d at a time. Rosen and McNeill (1996) described the feedlot as unlined, having no provision to contain run-off or run-on and had little vegetative cover. Such intense stocking of the land was shown to have an adverse impact on the shallow unconfined aquifer under the feedlot (Rosen et al., 1995; Rosen and McNeill, 1996; Rosen, 1996), primarily because of the large quantity of sheep excretions. This was demonstrated by measured NO3–N concentrations of >140 g m–3 and Cl concentrations >100 g m–3 in the groundwater down gradient of the feedlot (Rosen and McNeill, 1996). The documentation of NO3 concentrations in the groundwater that exceeded the New Zealand drinking water standards by more than 10 times the maximum acceptable value of 11.3 g m–3 NO3–N led the Hawkes Bay Regional Council (HBRC) to impose stricter conditions for the operation of the feedlot. The owners of the feedlot decided to close the operation in 1998 rather than meet the new conditions.
The land once used as the sheep feedlot was turned into a vineyard in July 1998 and the HBRC became concerned about land use intensification in the region. Areas of particular interest were (i) to gain an understanding of the fate of NO3 contamination following land use changes and (ii) to determine the rate of reduction of contamination that may be expected under natural conditions following such changes. The purpose of this study was to examine the current state of the groundwater under the old feedlot and to determine the potential for contaminants that are held in the soil above the groundwater table to be transported into the groundwater system. In particular, we characterized the quantity of mineralizable N in the unsaturated zone that could continue to act as a potential source of NO3 contamination under the former feedlot. We use this information to predict the concentrations of NO3 in the groundwater during the next 5 yr using a simple one-dimensional model. This enables us to generically evaluate the impact of feedlots on alluvial groundwater systems after closure and to determine the contribution of present land uses to the NO3 concentrations in groundwater. This type of information is essential for evaluating the risks of confined sheep feeding operations on long-term groundwater contamination.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSCONCLUSIONSREFERENCES |
|---|
|
Hydrology
Groundwater is the main source of water for domestic, school, and farm use. Groundwater levels range from about 110 m above mean sea level in the west to 83 m above mean sea level to the east. The groundwater table is about 15 m below the ground surface under the former feedlot, but may be <5 m below the land surface to the east. Rosen (1996) mapped groundwater flow as toward the east-northeast (toward the Ngaruroro River). These measurements were based on groundwater level measurements and elevations of the Marakakaho Stream and Ngaruroro River.
Rosen and McNeill (1996) showed that the average gradient of the stream water level in Maraekakaho Stream is approximately 0.005 in the study area. This compares with the average Ngaruroro River gradient of 0.004 at the same distance above the confluence with the Maraekakaho Stream. A higher gradient in the Maraekakaho Stream means that, at similar distances upstream of the confluence, Maraekakaho Stream water levels are above those of the Ngaruroro River. The piezometric contours estimated from river and groundwater levels indicate that groundwater is traveling generally toward the Ngaruroro River and away from the Maraekakaho Stream in the study area. Any water leaking from the Maraekakaho Stream, if it leaked out of the active channel, would tend to travel underground to emerge in the Ngaruroro River channel. Stream gaugings on the Maraekakaho Stream show that it loses approximately 80 L s–1 of flow from Tait Road to near its confluence with the Ngaruroro River. Surface flow is completely lost between the Tait Road Bridge and a location near a gauging site. If approximately 80 L s–1 of flow is lost from the Maraekakaho River below Tait Road Bridge, then this water, assuming no extraction, will travel through the gravels in the direction of the Ngaruroro River.
Land Use
The feedlot was in operation from 1986 to 1998. Seven residents and landowners were interviewed about past and current land use practices around the former feedlot to determine if any observed groundwater chemistry changes were caused by differences in land use that occurred between the 1995 groundwater sampling (Rosen and McNeill, 1996) and the most recent (2001) sampling.
Figure 2 shows general land uses in the Maraekakaho area at the time the feedlot was in operation. Cattle, deer (Cervus elaphus), and sheep farming dominated the area, with a small residential area near the Maraekakaho Stream Bridge. When the feedlot was operating it was located approximately 2 km up Keru Road from the Maraekakaho Bridge. An orchard occupies a large area southeast of the former feedlot and borders the Ngaruroro River. A vineyard was established around 1995 to 1996; it is approximately 1.5 km northwest of the former feedlot. Before this, the land use was sheep and cattle grazing.
|
All residential houses are on septic tanks. An increase in the number of houses near the Maraekakaho Bridge also means an increase in the number of septic tanks. This may cause an increase in nutrients leaching into the shallow groundwater <5 m below land surface to the east of the former feedlot. New wells around the feedlot area have been drilled for commercial and domestic water supplies. One of these new wells has been drilled approximately 200 m southeast of site GMK94-02 and is used to irrigate the orchard. If improperly constructed, these new wells may provide preferential pathways for surface water to drain to the groundwater.
Groundwater Sampling
Characterization of groundwater quality beneath the feedlot before its opening is difficult due to lack of data (Rosen et al., 1995). The GMK94 series of piezometers in and around the feedlot were installed in May through June 1994 to obtain data directly below the feedlot and immediately up and down gradient (Rosen, 1996). Four of the GMK94 sites were drilled with three nested piezometers at different depths (Table 1) in the aquifer (Rosen and McNeill, 1996). Groundwater samples were collected from these piezometers and domestic wells in June, August, and October 1994, and in October 1995. These same wells were resampled in May 2001 to determine if groundwater quality has improved since the feedlot was closed in 1998.
|
In this study, all 12 groundwater-sampling sites were located and, together with an upstream (of the former feedlot) Ngaruroro River site and upstream and downstream sites on the Maraekakaho Stream (Fig. 1), sampled in accordance with procedures outlined in Rosen et al. (1999). Groundwater sampling included purging piezometers and wells of at least three times the casing volume. Samples were collected as close to the wellhead as possible in wells with existing pumps. A 250-mL, untreated, polyethylene bottle and a 100-mL field-filtered (0.45 µm) polyethylene bottle were filled at each site. All samples were chilled below 4°C and analyzed for Cl, NO3–N, and NH4–N using an ion chromatograph, and for alkalinity (reported as HCO3) using an autotitrator (Table 2). Consistent results for repeat analyses and repeat sampling at daily intervals of NO3–N and Cl for many of the samples indicates the reproducibility of our results. Tables of the results from all previous sampling can be found in Rosen et al. (1995), Rosen and McNeill (1996), and Reeves et al. (2001).
|
Samples from the <2-mm size fraction were analyzed for total N and total C with a Leco FP-2000 analyzer (Leco Corp., St. Joseph, MI) according to procedures described by Blakemore et al. (1987). 1 The sample was combusted at 1050°C in a stream of O2, and the CO2 produced was measured by infrared detection while the N gas was measured by thermal conductivity. The instrument was calibrated using standard EDTA and reference soil samples. Data from the soil sampling sites are presented in Tables 3 and 4.
|
|
The computer model links the mechanisms of soil water flow through the root zone with the complex N transformations that result from natural processes, and those resulting from the application of N fertilizer, N uptake and recycling by the vegetation, and the returns of dung and urine from the sheep. The model calculates the balance and movement of water and N through a one-dimensional vertical slab of soil that extends from the surface to a depth of 7 m. The 7-m depth was chosen because it is likely to be above the water table depth, and SPASMO cannot handle dilution with groundwater. Although groundwater levels may be considerably deeper than 7 m, there is likely to be very little attenuation of NO3 below 7 m. A description of the N mass balance portion of the model is given below and a full mathematical description of the SPASMO model is given in Green et al. (2001).
The flow of water through the soil profile is simulated using a capacity model (Hutson and Wagenet, 1993) in which soil water is divided into mobile and immobile phases. The mobile domain is used to represent the soil's macropores (e.g., old root channels, worm holes, and cracks), and the immobile domain represents the soil matrix. After any rainfall or irrigation events, water is allowed to percolate through the soil profile. However, this percolation only occurs when the soil slab is above field capacity. The infiltrating water first fills up the immobile domain and, once this domain is full, it then refills the mobile domain as the water travels progressively downward through the soil profile. Subsequently, on days when there is no significant rainfall, there is a slow approach to equilibrium between the mobile and immobile phases, driven by a difference in water content between the two domains.
The model is run on a daily time step, and the calculations are made in the following sequence:
SPASMO solves the combined transport equations for water flow and the fate of mineral N that is assumed to comprise urea, NH4–N, and NO3–N both in solution and adsorbed onto the soils' clay and mineral particles.
The N transport component of SPASMO is based on a N balance equation that accounts for plant uptake, the application of mineral fertilizer, exchange and transformation processes in the soil, losses of gaseous N to the atmosphere, and the leaching of N below the root zone. The SPASMO model considers both the organic N (i.e., in soil biomass) and the mineral N (i.e., NH4 and NO3 in solution) contained in the soil and the plant biomass. Dissolved NO3 is considered to be mobile and to percolate freely through the profile, being carried along with the invading water. The movement of dissolved NH4 is retarded as it binds to the mineral clay particles of the soil. The soil can receive inputs of organic C and N from decaying plant residues, which is added to the litter layer of the topsoil, and inputs of mineral fertilizer which are applied to the soil surface during the spring and summer periods. Animal urine is assumed to be in the form of urea that is applied uniformly to the pasture surface. Animal dung is recycled back to the soil's organic matter, where it slowly decomposes and releases mineral N in the form of NH4.
Pasture roots play a key role in the water and N dynamics of the root zone. The amount of soil N removed by the pasture is determined by the growth of the above- and below-ground dry-matter, multiplied by their respective N concentrations. Daily biomass production is modeled using a potential production rate per unit ground area, G (kg m–2 d–1) that is related, via a conversion efficiency, 
(kg MJ–1), to the amount of solar radiant energy, 
(MJ m–2 d–1), intercepted by the grass shoots using the equation:
 | [1] |
Here fT, fN, and fW are response functions that range between zero and unity depending on temperature, plant N, and soil water status, respectively (Eckersten and Jansson, 1991). The value of G depends on the daily sunshine and temperature, plus the leaf-area index of the pasture, and is moderated by the soil's water and N status (King, 1993; Thornley et al., 1995). Pasture growth can achieve a maximum only if soil water and soil N are nonlimiting.
A simple allometric relationship is used to partition the daily biomass production into the growth of the foliage and roots. Plant biomass is expressed in terms of the balance between growth and senescence of the plant organs.
Allocation to the roots, 
R, depends on the leaf N content [N]F, having a minimum value [R0] at a maximum leaf concentration [N]Fmax, and increasing as NL decreases (Eckersten and Jansson, 1991), and is written as
 | [2] |
SPASMO accommodates seasonal changes in dry-matter allocation that are associated with changes in the nutrient status of the pasture. For simplicity, the corresponding seasonal changes in senescence rates, 
, have not been included in the model because we are concerned with the long-term consequences of these allocation patterns.
The model assumes that plant growth will achieve the maximum potential only if soil water and soil N (NO–3 and NH+4) are nonlimiting. The net uptake of N from the soil is set equal to the amount of N incorporated into the new biomass minus the fraction of N that has been retranslocated, 
, from the old or senescing tissues.
The pasture's demand for N is set by the maximum N content of the root [N]Rmax and leaf [N]Fmax material. During active growth the pasture tries to supply the new leaf and root material with N corresponding to these maximum concentrations. The potential uptake of N from the soil, PD (kg ha–1 d–1) is defined as
 | [3] |
This uptake can only be met if sufficient N exists in the soil. Otherwise both [N]F and [N]R will be reduced in low-N soils, and pasture growth will be curtailed.
The daily uptake of N from the soil is assumed to be proportional to the depth-wise distribution of the fine roots, and the total amount of NO3 (N) and NH4 (A) in each soil layer (Johnsson et al., 1987). The potential uptake of NO3 is calculated as
 | [4] |
R(z); the proportion of total mineral N as NO3, N (mg L–1); and the total growth requirement for N, PD. However, the actual uptake of NO3 is limited to a fraction, fM = 0.10, of the total NO3 available in each layer. Ammonium uptake, PA, is calculated in a similar way, by assuming PA to be proportional to the relative amount of NH4 in solution, A (mg L–1). If the total N uptake from a given soil layer is below the potential rate, PD, then a compensatory increase in uptake is allowed from other layers deeper in the root zone (Johnsson et al., 1987). This is achieved by adding a fraction (cum) of the deficit to the potential uptake from the next soil layers where more mineral N may be available. Surface roots are the most active (Clothier and Green, 1994) and they preferentially extract soil water and nutrients from the upper soil layers. As water and N stresses develop, the uptake action of roots typically switches to the deeper roots, because more water and nutrients are available there.
Allocation of the daily total N uptake to the new roots and leaves is based on the idea that roots receive N first, until they reach their maximum concentrations, and then that N is allocated to the leaves. Whenever soil N becomes limiting, a reduction factor fN is used to reduce the total N uptake. This reduction function also effectively reduces the leaf N contents and alters the dry-matter allocation pattern (Eckersten and Jansson, 1991). Pasture growth parameters in this study have been chosen to generate appropriate levels of dry-matter production; that is, the model simulates yields between 10 and 15 Mg dry matter (DM) ha–1 from an irrigated pasture and adds about 1000 kg DM for every 100 kg N ha–1 of N fertilizer.
The decomposition of soil biomass and animal manure adds to the amount of mineral N in the soil. This process is known as mineralization, and it is modeled by dividing the soil organic matter into three pools—a fast cycling litter pool, an almost stable humus pool, and a manure pool (Johnsson et al., 1987). The relative amount of organic-N in these three pools changes daily to reflect inputs of fresh biomass and manure, and the losses of older biomass and manure as it decomposes. The N demand for the internal cycling of soil C and soil N is regulated by the C/N ratio of the soil biomass, rO, which is one of the model inputs.
Decomposition of soil litter carbon (CL) is assumed to be a first-order process and is specified by a rate constant (KL) that is influenced by temperature and soil moisture. The products of decomposition are CO2, stabilized organic material (humus), and, conceptually, microbial biomass and metabolites. The relative amount of these products is determined by a synthesis efficiency constant (fE) and a humification fraction (fH). The following mass balance equations, which represent the inputs minus the outputs of soil C and soil N, are used to model the turnover of C and N in the litter pool:
 | [5] |
 | [6] |
 | [7] |
 | [8] |
Lastly, a set of mass balance equations is used to describe the turnover of C and N in the humus pool:
 | [9] |
 | [10] |
Decomposition of soil humus (CH) is assumed to follow first-order kinetics with a specific rate constant (KH) that depends on temperature and soil moisture. The other terms in these mass balance equations have already been described above.
All C and N turnover reactions can result in a net production (mineralization) or a net consumption (immobilization) of NH4, depending on the C/N ratio of the biomass, rO, in the three pools. From a consideration of mass balances, any increase in NH+4–N, due to mineralization, must equal the decrease in organic N from the three organic matter pools. Thus, the following mass balance equation is solved for mineralization:
 | [11] |
Mineralization occurs whenever 
NH+4/t > 0, otherwise immobilization will occur. The model also recognizes that if no NH4 is available for immobilization, then NO3 can be used according to the following equation:
 | [12] |
During all simulations reported here we chose typical values for most of the parameters: the rate constants were set equal to KL = 0.015 d–1, KM = 0.015 d–1, and KH = 0.00005 d–1; constant values were used for the efficiency of C turnover, fE = 0.5; the humification fraction, fH = 0.2; and the C/N ratio of the soil biomass, rO = 10.0, as suggested by Johnsson et al. (1987).
Transformation processes in the soil are regulated by abiotic response functions involving soil temperature and soil moisture. A Q10 relationship (Bunnell et al., 1977) is used to express the effect of temperature:
 | [13] |
The soil moisture factor decreases, on either side of an optimum level, in drier soil or in excessively wet soil (Johnsson et al., 1987); that is,
 | [14] |
S is the saturated water content; H and L are the high and low water contents, respectively, for which the soil moisture factor is optimal; and W is the minimum water content for process activity. The factor fS defines the relative effect of moisture when the soil is completely saturated, and M is an empirical constant.
All N-transformation processes in the soil are assumed to be first-order with rate constants that are regulated by both temperature and moisture status of the soil. The hydrolysis of urea, U (mg L–1), to NH4, A (mg L–1), is modeled as
 | [15] |
 | [16] |
Denitrification is the transfer of NO3 to gaseous N (N2 and N2O) products. This is an anerobic process, and consequently, it is highly dependent on soil aeration. Soil moisture is used as an indirect expression of the O2 status of the soil. The influence on the denitrification rate
 | [17] |
D) to a maximum at saturation (S), where d is an empirical constant. No denitrification occurs below the threshold point. The denitrification rate for each layer is modeled as
 | [18] |
The N transport model allows for an input of mineral N in the form of urea, NH4, or NO3. This option allows us to simulate different forms of mineral fertilizer that are broadcast onto the soil surface. We can also simulate the return of animal urine by assuming that it is all in the form of urea.
Once the urea or urine is applied to the soil surface, its fate is determined by two competing processes:
It is assumed that all of the urea enters the soil, and that any surface runoff of urea is negligible. The total mass of urea, MU (mg m–2), in each soil slab of thickness zR (mm) is found by solving the following mass balance equation:
 | [19] |
(m3 m–3) is the soil's volumetric water content, XU,i (mg m–2) is the mass of urea added to the ith segment (XU,i = 0 if i > 1), k1 (1/d) is the rate-constant describing the hydrolysis of urea to NH4, WPU (mm d–1) represents the percolation of dissolved urea through the soil. Different concentrations are a reflection of soil water content and the mass of urea that is added during each urine event. Urea is rapidly hydrolyzed to NH4, in a matter of a few days. The fate of dissolved NH4 is determined by six competing processes:
The total mass of NH4, MA (mg m–2), in each soil slab of thickness zR (mm) is found by solving the following mass balance equation:
 | [20] |
KD/) is the retardation factor for NH4, (kg L–1 soil) is the soil's dry bulk density, and KD (L kg–1) is the distribution coefficient that determines how much NH4 gets adsorbed to the cation-exchange sites of the soil. A standard batch-isotherm technique was used to determine the adsorption of NH4 (Vogeler et al., 1997). A linear isotherm fitted to the data over the concentration range 0 to 40 mg N L–1, yielded a distribution coefficient, KD, of 3.3 L kg–1. This value indicates that NH4 will be strongly adsorbed by the Takapau sandy loam soil. The fate of any NO3 in the soil water is determined by the following six processes:
The total mass of N, MN (mg m–2), in each soil slab of thickness zR (mm) is found by solving the following mass balance equation:
 | [21] |
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSCONCLUSIONSREFERENCES |
|---|
 | [22] |
The indicators chosen do not specifically identify that the feedlot is the sole source of groundwater contamination. Nitrate-N and Cl could be contributed by other farming practices in the surrounding area. However, the pattern and distribution of the various chemical indicators suggest the feedlot is the likely source of the contamination (Rosen and McNeill, 1996; Rosen, 1996).
Chloride concentration maps are not presented here because they show similar patterns to the NO3 concentrations. Rosen (1996) showed Cl concentration maps of the data collected before 1995.
Comparison of Previous Groundwater Results with May 2001 Sampling
Rosen et al., (1995), Rosen (1996), and Rosen and McNeill (1996) showed that the groundwater under the Maraekakaho area is a mixture of Maraekakaho Stream water, Ngaruroro River water, and rainwater. The groundwater flow direction was mapped flowing to the east-northeast toward the Ngaruroro River under the feedlot, with an approximate 10-m head difference between the southwest and the northeast boundaries of the feedlot. Elevated NO3–N concentrations were highest under the feedlot in 1994 and 1995, and the distribution of elevated concentrations was consistent with calculated groundwater flow directions down gradient of the feedlot.
Rosen and McNeill (1996) found the highest NO3–N and Cl concentrations in the GMK94 piezometers located in the feedlot and down gradient of the feedlot in August 1994, only 3 mo after the June 1994 samples were taken. In particular, Piezometers GMK94-01 and GMK94-06 had NO3–N concentrations >100 g m–3 at this time. Elevated NO3–N and lower HCO3 levels in wells and streams coincided with rainfall events with very little lag time (Rosen, 1996). This indicates that rainwater percolation is the main processes for washing accumulated soil NO3 into the unconfined groundwater aquifer. However, because percolation is rapid, direct rainwater recharge alone cannot account for the long-term spatial variations in groundwater chemistry found by Rosen and McNeill (1996). Groundwater NO3–N concentrations were generally low when sampled in June 1994 (0.3–2 g m–3) at the up-gradient sites and sites around the Maraekakaho bridge area, and generally higher (21–40 g m–3) under and immediately down gradient of the feedlot.
Chloride concentrations were also elevated above background values down gradient of the feedlot. However, there may have been some contribution from the down-gradient orchard. Alkalinity contours over the feedlot area show that there is little evidence of soil ammoniafication reactions causing increased HCO3 concentrations in the groundwater.
Figures 3 and 4 show contoured diagrams of HCO3 and NO3–N for both the May 2001 and June 1994 sampling runs. Where multiple piezometer nests occur at a site, the average concentration was used for the NO3–N maps, and the deep piezometer value was taken for contouring HCO3. This is because there is little variation between the concentrations detected in the deep versus shallow piezometers, except for HCO3 (see the following section). This is consistent with the method used in Rosen et al. (1995). Figures 5A, 5B and 6A, 6B show time trends of HCO3 and NO3–N, respectively, since the June 1994 sampling.
|
|
|
|
Some differences in HCO3 with aquifer depth can be seen at the multilevel nested piezometer sites. Sites GMK94-01 and GMK94-04 have the greatest variations of 15 g m–3 (decreasing with depth) and 30 g m–3 (increasing with depth), respectively. The other two sites (GMK94-02 and GMK94-03) show little variation with depth. The observations above are generally consistent with those observed in the June 1994 sampling round, except for Site GMK94-02. The June 1994 sampling round showed a large difference in alkalinities between the shallow (86 g m–3), intermediate (340 g m–3), and deep (244 g m–3) piezometers at Site GMK94-02. However, these higher concentrations may have been caused by insufficient purging of the deeper wells (Rosen and McNeill, 1996). By the August 1995 sampling, HCO3 concentrations in this well were 40 to 47 g m–3. The May 2001 results show HCO3 concentrations (44–45 g m–3) are about the same as the August 1995 sampling and are the same in the three piezometers at GMK94-02 (Table 2).
One explanation of the relatively low HCO3 concentrations measured in the GMK94 piezometers is that the nitrification reaction produces H+ ions (see Eq. [22]) that could react with the HCO3 in solution to form CO2 gas. This would explain why the piezometers within and directly down the groundwater gradient from the feedlot have the lowest HCO3 concentrations (which are lower than HCO3 concentrations of the Ngaruroro River) and the lowest pH (between 6.0 and 6.3 for GMK94-01, GMK94-04, andGMK94-06). GMK94-05 may have low HCO3 concentrations because of diffusion from the feedlot source, or because of other up-gradient sources of NO3 contamination. Higher HCO3 concentrations (by about 40 g m–3) and slightly higher pH values (by up to 0.2 pH units) measured in May 2001 in the deeper piezometers at GMK94-04 (Fig. 7) and at GMK94-06 indicate that nitrification reactions could be slowing down and the deeper part of the aquifer is beginning to recover. With time, it is expected that HCO3 concentrations in all the GMK94 piezometers should increase.
|
Nitrate-N concentrations are higher than the June 1994 sampling in all groundwater wells (where measured) except at the Maraekakaho Stream and Tait Road sites, which remained the same. However, most wells have lower concentrations than the maximum NO3–N concentrations measured by Rosen and McNeill (1996) in August 1995.
Rosen and McNeill (1996) reported higher NO3–N concentrations in the GMK94 sites during the August 1994 to August 1995 sampling (except at GMK94-02, which had the highest concentrations measured in May 2001), including NO3–N concentrations of 7 to 9.3 g m–3 at Site GMK94-05. The largest NO3–N increases during this time period were at GMK94-01 and GMK94-06 (Fig. 6). Nitrate-N concentrations measured in the May 2001 sampling ranged from 0.03 g m–3 (Maraekakaho Stream and Tait Road) to 48 g m–3 (GMK94-02 Intermediate and GMK94-06) (Table 2). Piezometers in and directly down the groundwater gradient of the former feedlot site had elevated NO3–N concentrations compared with up-gradient piezometers (Fig. 4). The two up-gradient piezometers, Simmond well and GMK94-05, had a large difference in concentrations (0.67 g m–3 and 10 g m–3, respectively). The N concentration in GMK94-05 also increased by about 3 g m–3 NO3–N from 1994 to 2001, indicating some N input up gradient of this piezometer. The highest concentrations of NO3–N in May 2001 (Fig. 6, Table 2) are found at Sites GMK94-06 and GMK94-02, both in the orchard and immediately down the groundwater gradient of the former feedlot. Nitrate-N concentrations at the sites near the Maraekakaho Township are <5 g m–3 except for Barley's well, which was measured at 8.6 g m–3 NO3–N in 2001. Nitrate-N concentrations in the multilevel piezometers GMK94-01, GMK94-02, GMK94-03, and GMK94-04 generally decrease with aquifer depth (Table 2, Fig. 7). Typically, variations in NO3–N concentration with depth are <6 g m–3, except for GMK94-04, where there is an 11 g m–3 decrease from the shallow to the deep piezometer in the May 2001 sampling. All multilevel piezometers showed a decrease in NO3–N concentration from the shallow to the deep piezometers. This decrease in NO3–N concentration was not observed in the 1994 and 1995 sampling because N inputs from the active feedlot were large and penetrated deeply into the aquifer. Because the constant source of NO3 input has decreased (to just soil input) the deeper part of the aquifer is becoming less contaminated more quickly than the upper part of the aquifer, and the NO3 entering the groundwater table now is not penetrating as deeply as when the feedlot was in operation.
Although the deeper parts of the aquifer have lower concentrations of NO3–N than the shallow parts of the aquifer, all three piezometers at GMK94-02 and the deep piezometer GMK94-03 show increasing trends in NO3–N from 1994 to 2001. These two piezometers are directly down gradient from feedlot inputs, indicating that NO3–N is still traveling down gradient to these sites.
Comparison of the groundwater NO3–N results from the 1994 and August 1995 sampling with the May 2001 sampling indicate NO3–N is stored in the unsaturated zone and percolates to the groundwater after large recharge events. This is because high concentrations of these anions persist in the groundwater 3 yr after the main source of these anions (the feedlot) was eliminated.
Chloride
Chloride patterns and values are similar in comparisons of the June 1994 and May 2001 data. The main feature in each contour map is the high Cl concentrations observed at Sites GMK94-02 and GMK94-06, which are both immediately down the groundwater gradient, but outside of the feedlot.
A new groundwater well, used to irrigate the orchard east of the vineyard, is likely to have elevated NO3–N and Cl concentrations given that it is located in the NO3–N and Cl plumes. When the pump is switched on, draw down from this well may alter the groundwater flow direction toward this well, and may provide a source of NO3–N and Cl to the soils in the orchard. This then has the possibility of leaching to groundwater beneath the orchard and recycling the high NO3–N and Cl concentrations in the area for a longer period of time than if the well had not been drilled.
Chloride concentrations in the May 2001 sampling ranged from 5.9 g m–3 in the Ngaruroro River to 81 g m–3 at GMK94-06 (Table 2). Elevated concentrations occurred under and immediately down gradient from the former feedlot site where a concentration of 81 g m–3 was observed at Site GMK94-06. An average concentration of 74 g m–3 was observed at GMK94-02. Typical background concentrations are 19 and 14 g m–3 at the up-gradient sites, Simmond's well and GMK94-05, respectively. Average Cl concentrations in wells sampled in Maraekakaho Township west of the mouth of Maraekakaho Stream (30 g m–3) and water collected from Maraekakaho Stream (36 g m–3) are similar.
No large difference (<10%) in Cl concentrations with depth was observed at Site GMK94-01 (Table 2). The other multilevel piezometers, GMK94-02, -03, and -04 all showed decreasing Cl concentrations with depth with concentrations that varied from 11, 7, and 16 g m–3 over vertical distances of 9, 1.6, and 11.1 m, respectively.
Soil Quality
The data collected during and after feedlot operations on the groundwater concentrations of HCO3, NO3–N, and Cl indicate that the groundwater down gradient of the feedlot is becoming less contaminated. To determine how quickly the groundwater quality would recover, we collected soil N data below the former feedlot to quantify how much N could continue to percolate into the groundwater with time. The soil data presented below provides data on how N moves through the unsaturated zone above the groundwater table. These results are used in our models, as described below, to provide predictions of groundwater quality with time.
Soil Sampling under the Former Feedlot
How much N is likely to leach under the former feedlot in the future is a question that can be answered by sampling the soil profile. In April 2001 we collected samples from two sites (Fig. 1) to a depth of 12 m to establish the depth-wise profile of total N (Tables 3 and 4). These soil samples revealed a total N content of between 0.01 and 0.02% (g/g), a total C content of between 0.07 and 0.12%, and a C/N ratio of between 4 and 7 at depths below about 3 m. Total N includes both organic N, which is immobile, and mineral N, likely to be in the form of NO3. The C/N ratio is calculated on the fine earth fraction (<2-mm size fraction) only. The problem here is to estimate the amount of NO3 that remains in the soil profile. This estimate can be made by comparing the measured C/N ratio (4–7) with the C/N ratio typical of a sandy soil (C/N = 10–12; Smith and Mullins, 1991). A very low C/N ratio, as measured under the feedlot site, indicates an excess of N relative to C in the soil profile. Any excess N is likely to be NO3–N.
The C/N ratio measured under the former feedlot at both soil sampling sites was quite low at depths below about 7 m in the soil profile. Nevertheless, given the measured total C content of the soil, and assuming a typical C/N ratio of 12, we calculated that more than one-half of the total N below 7 m is likely to be NO3 that originated from the former feedlot. The vertical profile of the C/N ratios for GMK01-SOIL1 and GMK01-SOIL2 are shown in Fig. 8 . The bulge in low C/N ratios approximates the depth to which the NO3 has probably leached during the previous 3 to 4 yr. This profile depth is consistent with output from the SPASMO model. Site GMK01-SOIL2 has a similar profile in total N, but our calculations of NO3 results in a smaller bulge at the 7- to 10-m depth, possibly due to errors in our simple procedure to estimate the C/N ratio. It could also be true that the GMK01-SOIL02 site may have received fewer nutrients from the feedlot operations because the site was located on the edge of the former feedlot.
|
Feedlot Nitrogen Balance
Historically, the feedlot at Maraekakaho was operated for about 13 yr, between 1986 and 1998, with the animals being grazed just once or twice each year. We do not have records of the actual grazing dates, so for the purpose of demonstration we have assumed here that a single 10-d grazing event occurred each October for 13 consecutive years. Thereafter, in the model we altered the land use from a feedlot to a vineyard, and reran the calculations of NO3 leaching under the vineyard using the next 14 yr of weather data. We used this second simulation to assess the potential for continued NO3 leaching under the former feedlot.
The calculations assume a sheep stocking density of 800 head ha–1 and a standing time of 10 d. The feedlot is a dryland pasture that is grazed just once per year, in the month of October. Pasture growth and N cycling within the root zone are predicted using daily sunshine, temperature, and rainfall at the site, with each process being moderated by the moisture and nutrient status of the soil (King, 1993). Sheep feed requirements are met first by the available pasture and then by pellet supplements, which accounts for most of the feed given to the sheep. We have no information on the quality and quantity of the pellet feed, so we have made some approximations to estimate the dung and urine returns to the soil. These approximations are based on normal levels of feed intake and assume simple ratios of N excretion by the grazing sheep (Whitehead, 1995). Each sheep urinates 3 L d–1, in the form of urea at a concentration of 10 g L–1. The daily return of fecal N from each sheep is taken to be 10 g N d–1 (Parsons et al., 1991). At a stocking rate (SR) of 800 sheep ha–1, the total urine N loading is 24 kg N ha–1 d–1 and a total fecal N loading of 8 kg N ha–1 d–1. Over the 10 d that the sheep are on the feedlot, the total N loading is assumed to be 240 kg N ha–1 in the form of urine and 80 kg N ha–1 in the form of dung. Thus, each 10-d grazing event adds about 320 kg N to the feedlot pasture. These rates indicate that there is the potential for excessive NO3 leaching to occur under the sheep feedlot.
The feedlot is assumed to be a dryland pasture that is grazed just once per year, in October. However, based on our observations of the feedlot when it was in operation, very little pasture grass was growing in the enclosures. This indicates that our model will be a conservative estimate of N leaching to the groundwater.
Pasture growth and N cycling within the root zone are predicted using daily sunshine, temperature, and rainfall at the site, with each process being moderated by the moisture and nutrient status of the soil (Green et al., 2001). The feedlot is assumed to be 1 km2 in size and to have 80000 sheep on the pasture site for a period of 10 d. Thus, the SR is set at 800 sheep ha–1, which is very high compared with a SR of 10 to 20, which is more typical of a grazed sheep pasture. The task here is to estimate the feed intake and the corresponding amount of dung and urine that is returned to the pasture by the grazing sheep.
The feedlot at Maraekakaho was operated between 1986 and 1998, and animals were grazed once or twice each year. Because we do not have records of the actual grazing dates we have assumed here, for the purpose of demonstration that a single 10-d grazing event occurred each October for 13 consecutive years. Thereafter, in the model we altered the land use from a feedlot to a vineyard, and ran the calculations of NO3 leaching under the vineyard using the next 14 yr of weather data. We used this 27-yr simulation to assess the potential for continued NO3 leaching under the closed feedlot after changing to vineyard land use.
The temporal pattern of NO3 leaching under the sheep feedlot is shown in Fig. 9 . For these calculations we used a low initial concentration of 5 g m–3 in the soil profile. For this simulation, the site was assumed to be operating as a feedlot for approximately the first 13.5 yr (Days 1 to 5000) and then was converted to a vineyard for the next 13 or so years (Days 5000 to 10000).
|
85%) is likely to be returned back to the soil as organic N in the form of decaying plant material as the ungrazed pasture sward senesces. The organic N that is returned to the soil biomass will slowly decompose and release mineral N in the form of NH4 in solution. This NH4 is strongly bound to the soil particles, yet it is quite quickly transformed into NO3 that is mobile and will move through the soil profile along with the percolating drainage water. In this model scenario, the animals are returning some 320 kg N back to the pasture each time the feedlot is in operation. So N input is well in excess of the nutrient requirements of the pasture. After 13 yr of a feedlot operation, the NO3 concentrations are predicted to be in excess of 100 g m–3 down to a depth of at least 7 m (Fig. 9). These predictions are the result of considering just a single grazing each year and assuming that some N is taken up by pasture growth. The actual NO3 concentrations under the feedlot may well have been much higher because there was probably more than one grazing event in some of the years and there is very little pasture available for the sheep. Toward the end of the feedlot operation, NO3–N concentrations of more than 140 g m–3 were recorded in the shallow groundwater at a depth of 12 m under the feedlot (Rosen and McNeill, 1996). Our model results support the measurements that show that NO3 concentrations in excess of 10 times the New Zealand drinking water standard are likely in the drainage water under the feedlot.
Once the land use was converted from a feedlot to a vineyard we see a very rapid decrease in soil NO3 concentrations (Fig. 9). This is because of a combination of two factors: (i) more water is draining through the soil profile under a vineyard (223 mm yr–1 compared with 74 mm yr–1 for the dryland pasture) and (ii) the total NO3 loading from fertilizer and animals is much reduced (30 kg N yr–1 compared with >300 kg N yr–1 under the feedlot operation). The NO3 concentration at a depth of 3 m has reduced from >200 to <50 g m–3 during the first 3 to 4 yr after conversion to a vineyard. This represents a reduction of about 80% in soil NO3 at 3 m.
Because of the stony nature of the underlying soil, and assuming piston flow, any drainage water that travels through the soil profile will advance, on average, at a rate of about 2.5 to 3 m yr–1 (223 mm yr–1 through an 8–10% transport volume). An 80% reduction in NO3 concentration at the 7-m depth would appear likely 4 to 5 yr after the land use conversion. An impact of the NO3 leaching from the former feedlot on NO3 concentrations in the ground water, at depths up to 14 m, is expected to be observed for a period of up to 10 yr after closure of the feedlot. Dilution of the drainage by lateral transport of low-NO3 groundwater will also occur. This dilution has not been estimated, but this additional low NO3 groundwater may speed up the renovation of the groundwater quality under the feedlot site.
Vineyard Nitrogen Balance
The conceptual model used to describe the vineyard land use included grape vines alternating with grass between rows, assuming irrigation and fertilizer management practices that are typical of a commercial vineyard (Green et al., 2000). For modeling purposes, the grape vines are irrigated automatically with an aliquot of 2.6 mm of water whenever the water deficit within the root zone exceeded one-half of the total available soil water. A single dressing of CaNH4[NO3]2 is applied once per season, in October, at a prescribed rate of 30 kg N ha–1. However, no fertilizer is applied to the vineyard on the former feedlot, because of the high total N status of the soil. Pasture growth and N cycling within the root zone are predicted using daily sunshine hours, temperature, and rainfall at the site, with each process moderated by the moisture and nutrient status of the soil (Green et al., 2001).
Vegetative and fruit growth of the grapevines is simulated to estimate the total N uptake by the plants. Vine growth is modeled using daily weather data and is moderated by the moisture and nutrient status of the soil. The vines were assumed to have been thinned during late summer to reduce their vigor, and pruned in mid winter to remove the old shoots. All leaf and shoot material was recycled back into a pool of soil organic C and N that slowly
