The effect of winter cover crop management on nitrate leaching losses and crop growth

Transcription

1 Journal of Agricultural Science, Cambridge (1998), 131, Cambridge University Press Printed in the United Kingdom 299 The effect of winter cover crop management on nitrate leaching losses and crop growth G. S. FRANCIS*, K. M. BARTLEY AND F. J. TABLEY New Zealand Institute for Crop & Food Research Limited, Private Bag 474, Christchurch, New Zealand (Revised MS received 12 May 1998) SUMMARY Two field experiments in Canterbury, New Zealand, were conducted during following the ploughing of temporary pasture leys. These experiments investigated the effects of cover crop management on the accumulation of soil mineral N and nitrate leaching during winter, and the growth and N uptake of the following spring cereal crop. The cover crops used were ryegrass (Lolium multiflorum L.), oats (Avena sativa L.), lupins (Lupinus angustifolius L.), mustard (Sinapis alba L.) and winter wheat (Triticum aestivum). Ploughing of temporary pasture in autumn (March) resulted in extensive net N mineralization of organic N by the start of winter (June). In fallow soil, mineral N in the profile in June ranged from 98 kg N ha in 1993 to 128 kg N ha in When cover crops were established early in the autumn (March) in 1993, both the above-ground dry matter production ( kg DM ha) and its N content ( 71 kg N ha) were substantial by the start of winter. In 1994, establishment of cover crops one month later (April) resulted in very little dry matter production and N uptake by June. In both years, compared with fallow soil, winter wheat planted in May had little effect on soil mineral N content by the start of winter. Compared with fallow, cover crops had little effect on soil drainage over winter. Cumulative nitrate leaching losses from fallow soil were much smaller in 1993 (23 kg N ha) than in 1994 (49 kg N ha), mainly due to differences in rainfall distribution. Cover crops reduced cumulative nitrate leaching losses in 1993 to 1 5 kg N ha and in 1994 to 22 3 kg N ha. When cover crops were grazed, soil mineral N contents were increased due to the return of ingested plant N to urine patch areas of soil. Elevated soil mineral N contents under grazing persisted throughout the winter. Grazing had little effect on cumulative nitrate leaching losses, mainly because of the small amount of drainage that occurred after grazing in either year. Compared with fallow, incorporation of large amounts of non-leguminous above ground dry matter depressed the yield and N uptake of the following spring-sown cereal crop. Where cover crops were grazed, yields of the following cereal crops were similar to those for soil fallow over the winter. INTRODUCTION Ley arable (or mixed cropping) farming is widespread on the Canterbury Plains of New Zealand, a region that occupies 7 ha. Grazed ryegrass (Lolium perenne) white clover (Trifolium repens) pastures are usually grown for 2 5 years, followed by 2 5 years of arable crops. The nitrogen (N) fertility of the soil increases during the pastoral phase of the rotation due to symbiotic N fixation by the clover (Haynes & Francis 199). Pastures are commonly ploughed in late summer early autumn and left fallow until a cereal crop is planted in the spring. * To whom all correspondence should be addressed. Francisg crop.cri.nz A drawback of this ley arable system is that substantial amounts of nitrate can be leached during the winter fallow period, with nitrate concentration in potable water in Canterbury (Adams et al. 1979) often greater than the Drinking-Water Standards for New Zealand of 11 3 mg nitrate-n l (Anon. 1995). In Canterbury, large amounts of nitrate are leached in the first and second winters after leys are ploughed (Francis et al. 1995). Cover crops established early in the autumn can reduce leaching losses in some years, mainly due to their uptake of soil mineral N before the start of winter leaching (Meisinger et al. 1991; Francis et al. 1995; Davies et al. 1996). However, the incorporation of large amounts of cover crop residues in the spring can cause significant yield reductions in the following spring-sown crop (Martinez & Guiraud

2 3 G. S. FRANCIS, K. M. BARTLEY AND F. J. TABLEY 199; Francis et al. 1995; Davies et al. 1996). A possible solution to this problem is to graze cover crops over autumn winter to reduce the amount of residue that needs to be incorporated in the spring. However, during grazing, organic N contained in the cover crop residues will be returned to the soil in concentrated mineral form in urine patch areas. Large amounts of N may be leached from these urine patch areas if significant drainage events occur after grazing. There is no published information on the effect of grazing cover crops on nitrate leaching losses over autumn winter in New Zealand. This 2-year study investigated the effects of cover crop management after the ploughing of leys on nitrate accumulation in the soil profile, nitrate leaching losses, and yield and N uptake of a following spring-sown cereal crop. MATERIALS AND METHODS Site and soil This experiment was undertaken from 1993 to 1995 at the AgResearch Templeton Research Farm, Canterbury, New Zealand (43 38 S, E), on a Templeton silt loam soil (Udic Ustochrept; USDA 1983). The site had been under grazed ryegrass white clover pasture for 4 5 years before the experiment began. Mean annual rainfall is 68 mm, and is relatively evenly distributed throughout the year. Soil drainage is likely to occur from about June to September (Cox 1978). Rainfall was recorded daily at the experimental site, with other meteorological data recorded daily at a station 15 km away. Experimental treatments Pasture (c. 4% clover) was mouldboard ploughed to c. 2 mm depth in March of each year. Plots were Table 1. Grazing dates for the treatments in 1993 and 1994 Treatment Dates grazed 1993 experiment Pasture Jun, Oct Fallow Winter wheat Ryegrass Aug, Oct Oats Jun Lupins Aug Mustard Jun 1994 experiment Fallow Winter wheat Ryegrass Sep, Oct Oats Sep, Oct either left fallow over winter or secondarily cultivated and sown with cover crops. In 1993, ryegrass, oats, lupins and mustard were sown as cover crops in March, with winter wheat sown in May. Unploughed pasture control treatments were also included in the experimental design. Based on results collected in 1993, only ryegrass, oats (both sown in April) and winter wheat (sown in May) were grown as cover crops in Except for the fallow and winter wheat, cover crop main plots in both years were split into grazed and ungrazed subplots. Cover crops were grazed once or twice during the winter (Table 1), depending on the amount of dry matter produced. Stocking density was 4 sheep ha in 1993 and 19 sheep ha in At each grazing event, small (c. 25 m ) areas of the grazed subplots were protected from sheep. After removal of the sheep from the plots, herbage in the protected subplots was cut to grazing height. Simulated urine patches were made in half of these subplots by applying 2 cm of sheep urine to an area of 49 cm a typical size and amount for sheep urinations (Williams & Haynes 1994). The N concentration of the urine was 5 6 gn l, which is typical for sheep eating herbage with the N concentrations in these cover crops (Haynes & Williams 1993). Except for fallow and winter wheat, cover crop plots were mown and surface mulched in spring (October). These plots were then ploughed, secondarily cultivated and sown with a test crop of either spring wheat in 1993 or spring barley in At tillering, half of each main plot received urea fertilizer ( kg N ha), with crops harvested in the autumn (February March). Each experiment was a randomized block, split-plot design with three (in ) or four (in ) replicates of the main treatments. Cover crops were the main plots, with crop management the split-plots. Main plots were 2 m long and 1 m wide. Sampling procedures Soil samples (four bulked, 25 mm diameter cores) to 1 m depth were taken from each ungrazed plot for ammonium- and nitrate-n analysis at the start (June) and end (October) of winter leaching. Additional samples were also taken from the grazed subplots the day after grazing at the end of winter leaching. In 1993 some treatments were grazed only once, with samples from these treatments taken from both simulated urine patch areas and areas that had been protected from receiving excretal returns. For treatments grazed twice in either year, the area of soil covered by urine patches was calculated from the area of the grazed plots, the stocking density and the duration of grazing (Haynes & Williams 1993). In 1993, c. % of the soil area would have been covered by urine patches. In this year, the distribution of urine

3 Cover crop management and nitrate leaching 31 patches at sampling was not apparent due to rain (2 4 mm) that fell between grazing and sampling. A stratified sampling strategy was not possible for the separate collection of soil from urine patch and nonurine patch areas. Thus, in 1993, 2 soil cores were taken randomly from each plot to sample both urine patch and non-urine patch areas. In 1994, the soil covered by urine patches in treatments grazed twice was calculated to be c. 3%. Urine patch areas were still discernible at sampling as areas of wet surface soil, allowing the collection of separate samples from urine patch and non-urine patch areas. Thus, at the end of winter leaching in 1994, 14 soil cores were taken from urine patch areas and six from non urine patch areas. Each set of these samples were subsequently mixed thoroughly and subsampled before analysis. Soil samples after harvest of the cereal crops were taken randomly from the unfertilized areas of both grazed and ungrazed plots. Two porous ceramic samplers (25 mm diameter 55 mm long) were installed in each plot at 6 mm depth the day after cover crops were sown. Additional soil solution samplers were installed in simulated urine patch areas the day after the application of urine. Soil solution samples were extracted after significant ( 2 mm) rainfall events by the application of a 7 kpa suction for 6 min (Francis et al. 1992). Soil moisture content was measured in each main plot using time domain reflectometry ( 2 mm depth) and neutron probe (2 8 mm) techniques. Nitrate leaching losses were calculated as the product of the mean soil solution nitrate concentration at successive sampling events and the calculated drainage between these samplings (Francis et al. 1992). A simple water balance model was used to calculate drainage, with evapotranspiration of the cover crops calculated from a modified SIRIUS model (Jamieson 1989). Total above-ground biomass of the cover crops was measured ( 25 m quadrat) on the day before each grazing and before mowing in spring. Wheat yields were determined from harvests of two randomly chosen, 5 m quadrats per plot. Plant material was dried (7 C) and ground for subsequent analysis. Ammonium- and nitrate-n in soil extracts, soil solution samples and total N in plant samples were analysed as described by Francis et al. (1992). RESULTS Soil temperature, rainfall and drainage Soil temperature at 1 mm depth followed a similar pattern in 1993 and 1994, decreasing from 1 12 C in March to 2 3 C in June and increasing to 6 7 C in September (data not shown). Although total rainfall amount (May September) was near to the long-term mean (293 mm) in both 1993 (3 5 mm) and 1994 (318 3 mm), its distribution was markedly different Rainfall (mm) (a) (b) May Jun Jul Aug Sep Fig. 1. Daily rainfall (mm) from May to September in (a) 1993 (cumulative total 3 5 mm) and (b) 1994 (cumulative total mm). between years (Fig. 1). Calculated total cumulative drainage from the fallow treatment was similar in both years, although the pattern of drainage varied between years (Fig. 2). In 1993 drainage was least from the pasture control. In 1993, compared with fallow, cover crops only significantly reduced (P 5) calculated cumulative drainage in September. All cover crops were as effective at reducing drainage, with drainage from lupins, mustard and ryegrass (data not shown) being similar to that from oats. In 1994, cover crops had no significant effect on cumulative drainage throughout the experiment. Dry matter production, N uptake and soil mineral N In June 1993, above-ground dry matter (DM) production varied significantly (P 1) between cover crops (Table 2). Dry matter production by winter wheat and its N uptake was much less than for the other crops due to its later sowing (May). Due to variations in cover crop herbage N concentrations (data not shown), the herbage N content was greater in the mustard than the oats, lupins or ryegrass. Except for winter wheat, N uptake by cover crops significantly reduced soil mineral N content ( 1 m depth) compared with fallow at this sampling time. For all crops, cultivation of soil and sowing of cover crops resulted in a greater soil mineral N content than for the undisturbed pasture control. In October 1993, although the amount of DM and its N content varied between the ungrazed cover crops, soil mineral N

4 32 G. S. FRANCIS, K. M. BARTLEY AND F. J. TABLEY Cumulative drainage (mm) (a) (b) 2/5 8/6 29/6 22/7 16/8 7/9 28/9 16/5 13/6 3/6 28/7 16/8 3/9 Date Fig. 2. Cumulative drainage (mm) from May to September for ungrazed pasture ( ), fallow ( ), winter wheat ( ) and oats ( ) treatments in (a) 1993 and (b) Vertical bars represent S.E.s (D.F. 12 for 1993; D.F. 9 for 1994). contents were similar under all cover crops and were significantly less (P 1) than under fallow (Table 2). In 1994, cover crop DM and N content were not determined in June as crops had grown very little by this time. Soil mineral N content in fallow soil in June 1994 was similar to that in June By October 1994 all ungrazed cover crops had significantly lower soil mineral N contents than fallow, although the extent of their DM accumulation and N content varied (Table 2). In 1993, the oats and mustard treatments were grazed in June, but the lupins and ryegrass were grazed in mid-august when they had produced 4494 and 3412 kg DM ha respectively (containing 132 and 74 kg N ha). The oats, lupins and mustard crops died after grazing, with little residue remaining above ground by the spring. Pasture and ryegrass survived grazing, with sufficient regrowth to require a second grazing in October when they had produced 3924 and 443 kg DM ha respectively. In 1994, the ryegrass and oats treatments were first grazed in September when they had produced 117 and 1116 kg DM ha respectively. Both cover crops were grazed a second time in October when they had produced 1748 and 1647 kg DM ha respectively. Cover crop management had a marked effect on soil mineral N content in spring (October). In 1993, soil mineral N contents in the ryegrass and pasture treatments grazed twice over winter (mean of urine patch and non-urine patch areas) were greater (P 5) than in the ungrazed treatments (Table 3). For oats and mustard, soil mineral N contents were not significantly different in the ungrazed plots and the grazed plots that had not received urine returns. For lupins, however, soil mineral N contents were greater in the non-urine patch areas of the grazed plots than in the ungrazed plots. The application of simulated urine to grazed plots increased soil mineral N contents Table 2. Soil mineral N ( 1 m depth) and above-ground herbage dry matter and N content in ungrazed treatments at the start (June) and end (October) of winter leaching in 1993 and 1994 Soil mineral N Herbage DM Herbage N (kg N ha) (kg DM ha) (kg N ha) Treatment Jun Oct Jun Oct Jun Oct 1993 experiment Pasture Fallow Winter wheat Ryegrass Oats Lupins Mustard S.E. (D.F. 12) experiment Fallow Winter wheat ND 76 8 ND 1686 ND 75 5 Ryegrass ND 43 2 ND 3558 ND 88 Oats ND 52 1 ND 6965 ND S.E. (D.F. 9)

5 Cover crop management and nitrate leaching 33 Table 3. Soil mineral N content ( 1 m) in ungrazed and grazed cover crop plots in spring (October) 1993 and 1994 Soil mineral N Soil mineral N Treatment (kg N ha) Treatment (kg N ha) 1993 experiment Pasture, ungrazed 34 9 Oats, ungrazed 27 1 Ryegrass, ungrazed 18 9 Lupins, ungrazed 35 7 Pasture, grazed* 75 2 Mustard, ungrazed 45 3 Ryegrass, grazed* 78 3 Oats, grazed urine 45 6 S.E. (D.F. 4) 9 11 Lupins, grazed urine Mustard, grazed urine 68 4 Oats, grazed urine 97 2 Lupins, grazed urine Mustard, grazed urine 66 6 S.E. (D.F. 2) experiment Ryegrass, ungrazed 43 2 Oats, ungrazed 52 1 Ryegrass, grazed* Oats, grazed* S.E. (D.F. 6) 34 5 * Mean of urine patch areas. 4 (a) significantly greater (P 1) in the grazed plots (mean of urine patch and non-urine patch areas) than in the ungrazed plots (Table 3). Cumulative nitrate leaching loss (kg N/ha) (b) 2/5 8/6 29/6 22/7 16/8 7/9 28/9 16/5 13/6 3/6 28/7 16/8 3/9 Date Fig. 3. Cumulative nitrate leaching losses (kg N ha) from ungrazed pasture ( ), fallow ( ), winter wheat ( ) and oats ( ) treatments in (a) 1993 and (b) Losses from the grazed oats treatment ( ) in 1993 are also shown. Vertical bars represent S.E.s (D.F. 12 for 1993; D.F. 9 for 1994). in the oats and lupin treatments only. In 1994, the oats and ryegrass were both grazed twice during the winter. Soil mineral N contents in October were Calculated N leaching losses Cumulative N leaching losses were much lower in 1993 than in 1994 (Fig. 3). In 1993 losses from all treatments were very low until September. Total cumulative losses were greatest from the fallow treatment, with slightly, but significantly (P 5), greater losses from winter wheat than from ungrazed oats. Losses from the ungrazed ryegrass, lupin and mustard plots (data not shown) were similar to those from the ungrazed oats. Leaching losses were not affected by grazing until the last leaching event of the winter. At this time, leaching losses were significantly greater (P 5) from the oats plots that received simulated urine than from the ungrazed plots (Fig. 3a). Similar, but non-significant, effects were apparent for ryegrass, lupins and mustard (data not shown). In 1994 leaching occurred earlier in the winter than in Ungrazed oats (Fig. 3b) and ryegrass (data not shown) significantly reduced (P 5) leaching losses by a similar amount compared with fallow. A similar, but non-significant, trend was apparent for winter wheat. Grazing of cover crops did not affect leaching losses at any stage of the experiment (data not shown). Dry matter production, test crop grain yield and total N yield For ungrazed treatments, the amount of aboveground DM incorporated into the soil in spring in

6 34 G. S. FRANCIS, K. M. BARTLEY AND F. J. TABLEY Table 4. Harvest grain yield and total (i.e. grain and straw) N yield in the spring wheat ( ) and spring barley ( ) test crops Grain yield (kg ha) Total N yield (kg N ha) Grazed Ungrazed Grazed Ungrazed Treatment N* N N N N N N N experiment Pasture Fallow Winter wheat Ryegrass Oats Lupins Mustard S.E. (D.F. 48) experiment Fallow Winter wheat Ryegrass Oats S.E. (D.F. 33) * Amount of fertilizer N applied to the test crops in kg N ha. Table 5. Soil mineral N contents ( 1 m) in unfertilized, ungrazed and grazed treatments at test crop harvest in the experiment. Mean values for cover crops Soil mineral N (kg N ha) Depth (mm) Ungrazed Grazed S.E.(D.F. 1) both 1993 and 1994 varied greatly between cover crops (Table 2). Much less DM was produced by spring in all crops in 1994 than in Aboveground DM incorporated in grazed plots in spring was small ( 3 kg DM ha) in each year (data not shown). Harvest grain yield and total N uptake (i.e. N in the grain and straw) of the spring wheat ( experiment) and spring barley ( experiment) test crops are shown in Table 4. In both years, where cover crops were grazed and no N fertilizer was applied, the yield and total N uptake of the test crop was not significantly different between cover crop species. In addition, the application of N fertilizer ( kg N ha) had no significant effect on test crop grain yield or plant N uptake in the grazed cover crop treatments. In 1993, test crop (spring wheat) yields from the nil N, ungrazed oats, ryegrass, pasture and mustard treatments were less than following fallow. With the exception of lupins, all ungrazed cover crops showed a trend towards greater test crop yields when N was applied, although this effect was not significant for the ryegrass or pasture. Except for winter wheat, test crop yields were not significantly different between cover crops when fertilizer was applied to ungrazed treatments. The total N yield from ungrazed plots was least following oats, ryegrass and pasture and greatest following fallow, winter wheat and lupins. Application of fertilizer N increased total N yield in most ungrazed treatments, although this was only significant following winter wheat, mustard and pasture. For all treatments, grazing increased total N yield of the test crop, although this was not significant following lupins or mustard. In 1994, yield and total N uptake of the test crop (spring barley) was significantly greater (P 5) following ryegrass than oats in the nil N, ungrazed treatments. The application of N fertilizer to the ungrazed oats treatment significantly increased both the test crop yield and its total N yield. Where no fertilizer N was applied, yield of the test crop was not significantly different between grazed and ungrazed treatments. In contrast, total N yield was greater following grazing than incorporation.

7 Cover crop management and nitrate leaching 35 Mean soil mineral N contents for the ungrazed and grazed pasture, oats, lupins, ryegrass and mustard treatments at the harvest of the test crop in the experiment are shown in Table 5. Distribution of mineral N through the soil profile was similar for both ungrazed and grazed treatments, with greatest N contents at 2 4 mm depth. Mineral N contents at all depths were greater where cover crops had been grazed than where they were incorporated. DISCUSSION Cover crops can be effective at reducing nitrate leaching losses compared with bare fallow through reducing soil drainage and reducing soil mineral N content when drainage occurs (Alvena s & Marstorp 1993; Shepherd et al. 1993; McCracken et al. 1994; Shepherd & Lord 1996). However, cover crops usually affect leaching losses more through reducing soil mineral N content than through reducing drainage volumes (Meisinger et al. 1991; Davies et al. 1996). In this experiment, we calculated that soil drainage was reduced only a small amount by the growth of cover crops compared with fallow. In the autumn and winter of both years, relatively regular rainfall and low potential evapotranspirative demand resulted in similar amounts of soil water to be lost through cover crop evapotranspiration and bare soil evaporation. Cover crops did not affect drainage amounts until potential transpiration demand by the crops increased in the spring (September). Thus, as in previous experiments (Francis et al. 1995), reductions in drainage were only significant when major rainfall events occurred in spring (i.e. in 1993). As grazing was not expected to have a large effect, drainage was not measured from the grazed subplots. The importance of sowing cover crops early in the autumn is evident from the differences between 1993 and 1994 in DM production and crop N uptake. Cover crops sown in March produced substantially more above ground DM and removed considerably more mineral N from the soil by the start of winter than crops sown in April or May. As in previous experiments with winter cover crops (Francis et al. 1995), DM production of winter wheat (traditionally sown in late May) was very small by the start of winter in either year, and soil mineral N content was greater than in the fallow treatment. This greater soil mineral N content under winter wheat is attributed to additional net N mineralization caused by the secondary tillage needed to create a seed bed for this crop (Francis et al. 1995). In comparison, fallow soil was only ploughed in the autumn and then left undisturbed until the following spring. The time of grazing varied between crops and between years in response to the rate of DM production. Soil mineral N contents after grazing were only measured in the spring, so the time between grazing and sampling varied. Nevertheless, in all treatments, there was a trend towards greater soil mineral N contents where cover crops were grazed compared with ungrazed. In ungrazed cover crops, soil mineral N contents in spring (October) were relatively low due to substantial uptake of mineral N by the cover crop. In the grazed cover crop plots, soil mineral N contents were greater due to the return of N to the soil in urine patch areas at rates of 3 1 kg N ha (Ball & Ryden 1984). For treatments grazed twice, the soil mineral N content in the grazed areas was much greater in 1994 than in This was largely due to the different sampling strategies used in each year. In 1993, the area of soil covered by urine patches during the grazing events was calculated to be c. %. Random sampling of soil should therefore have equally sampled areas that had or had not received excretal returns. However, the stratified sampling method used in 1994 appeared to be much more efficient at sampling urine patch areas than the random method used in In the treatments grazed only once, soil mineral N contents were similar in the ungrazed subplots and the grazed subplots that did not receive urine applications. The exception was the lupin treatment that had a much greater mineral N content in grazed non-urine patch areas than the ungrazed areas (Table 3). In 1993, the oats, lupins and mustard died after their first grazing in winter. Soil temperatures during winter were too low (2 3 C) for extensive net N mineralization to occur in any treatment (Stanford et al. 1973). It appears that as soil temperatures rose in the spring the extent of net N mineralization varied between the grazed treatments. During the decomposition of residues with low N concentrations, extensive net N immobilization often occurs (Haynes 1986). At the time of their grazing, the N concentration in plant residues was greater for the lupins (2 9%) than the mustard (2 5%) or the oats (1 7%). As a result, the period for net N immobilization was probably shorter during the decomposition of lupins than mustard or oats (Jensen 1992), leading to greater net N mineralization by sampling in October. As expected, soil mineral N contents were greater in the urine patch than ungrazed areas for the oats and lupin treatments. It is unclear why this effect was not also observed for the mustard treatment. Soil mineral N contents were greater in urine patch areas under lupins than under oats or mustard, largely due to the different grazing times for these treatments. Cover crops that were grazed earliest in the season had the lowest soil mineral N content at sampling in the spring. Oats and mustard were grazed in June, and it appears that by October some N returned to the soil in the urine patch areas was no longer present in mineral form. As no plant uptake occurred in these plots and leaching losses were small, some incorporation into the soil organic N pool is likely to have

8 36 G. S. FRANCIS, K. M. BARTLEY AND F. J. TABLEY occurred (Williams & Haynes 1994). In contrast, the lupin plots were grazed 2 months later (in August) and it appears that more of the returned N in the urine patch area was still in mineral form in the spring. Nitrate leaching losses in both years were less than the losses of 75 1 kg N ha often measured following the ploughing of temporary leguminous pastures in March (Francis 1995). Leaching losses in 1994 were particularly low, and this was mainly due to the low amounts of drainage during late autumn early winter. As in other experiments (Shepherd et al. 1993; Francis et al. 1995; Davies et al. 1996; Shepherd & Lord 1996), cover crops reduced nitrate leaching losses when drainage events occurred after the crops had taken up considerable amounts of soil N. In general, leaching loss reduction was broadly related to the DM production and the N content of the cover crop at the time of leaching. The grazing of cover crops increased the mineral N content of soil through the return of N in urine patch areas. Mineral N contents remained elevated under urine patch areas throughout the winter as net N immobilization of this urine N into the soil organic matter and or its uptake by the cover crops was less than that applied. However, leaching losses of N from cover crops were largely the same from grazed and non-grazed areas (Fig. 3). This contrasts with other experiments in wetter winter environments where nitrate leaching losses from grazed pastures were greatest from urine patch areas of soil (Cuttle et al. 1992; Ruz-Jerez et al. 1995). The main reason for the limited effect of grazing on leaching losses in this experiment was the small amount of drainage after grazing in either year. During a urination event, the maximum depth of movement of urine N into this soil type is c. 1 mm, with most of the urine N present above mm depth (Williams & Haynes 1994). In this soil, substantial macropore flow of water and solutes is not expected to occur for most rainfall events (Francis et al. 1995). Thus in 1993, when c. 4 mm of drainage occurred after grazing, convective-dispersive flow was likely to have transported nitrate 1 mm through this soil, which has a volumetric water content of c. 3 at field capacity (McLaren & Cameron 1996). The initial depth of movement of the urine N and its leaching during subsequent drainage events is clearly not sufficient for most of it to reach the solution sampler depth (6 mm) during the experiment. However, it does appear that a limited amount of preferential flow of nitrate may be occurring in this soil as nitrate concentrations in the soil solution at 6 mm depth under urine patches were greater than under nonurine patch areas at the last sampling event in In 1994, grazing did not increase leaching losses. This is because there were no major drainage and leaching events after grazing in this year. Soil mineral N contents after grazing, however, show that there is potential for increased nitrate leaching losses from urine patch areas of grazed plots. This is especially the case if large drainage events occur soon after grazing. For cover crops to be most effective at reducing nitrate leaching losses they need to be planted as early as possible in autumn (Meisinger et al. 1991; Francis 1995). Consequently, compared with bare fallow, the common local practice of planting winter wheat in May has little effect on reducing leaching losses as this crop has very little effect on soil mineral N content by the start of winter (Shepherd & Lord 1996). Early planting results in greater DM production and N uptake before the start of drainage. However, planting early will also require the early grazing of these cover crops. Urine patch areas of soil with high mineral N contents will then be present for longer periods over the winter, increasing the potential for elevated leaching losses from these areas of soil. Under the same environmental conditions, nitrate leaching losses vary between urine patches produced by different animals. In comparison with cows, sheep urinations are much smaller in volume and cover a smaller area of soil. The depth of urine movement into the soil and the subsequent amount of leaching loss is much less for sheep than cow urine patch areas (Williams & Haynes 1994). Grazing cover crops with sheep rather than cows is therefore important to minimize winter nitrate leaching losses. The timing of grazing in relation to rainfall events is also important in determining the effects of grazing on leaching losses. Delaying grazing for as long as possible in winter will minimize the potential leaching losses, but this needs to be balanced against the possible decline in feed quality with time. Although the grazing of cover crops has the potential to increase nitrate leaching losses, it can have a very beneficial effect on the grain yield and total N yield of the following spring-planted cereal crop (Table 4). Both grain yield and total N yield were significantly reduced when large amounts of aboveground herbage DM were incorporated in the soil in spring, compared to grazed plots. The reduction in yield generally increased with the amount of DM, although this varied somewhat between crops and between years. Similar depressions in yield have been reported in other experiments (Martinez & Guiraud 199; Jensen 1991; Wallgren & Linde n 1994; Francis et al. 1995; Davies et al. 1996), and have been attributed to extensive net N immobilization during the decomposition of cover crop residues (Jensen 1991; Davies et al. 1996), resulting in temporary N deficiency in the following spring crop. In this experiment, net N immobilization following the incorporation of large amounts of non-leguminous cover crop residues had a residual effect until the harvest of the following spring cereal crop (Table 5). However, the application of small ( kg N ha) amounts of N fertilizer overcame the yield reduction.

9 Cover crop management and nitrate leaching 37 In contrast, incorporation of large amounts of lupin residues did not affect the yield of the following crop. As legume residues often have higher N concentrations than non legumes, they often decompose faster with shorter periods of net N immobilization than non legume residues (Jensen 1992; Amato et al. 1987). Where cover crops are not grazed over the winter, the growth of legume cover crops with high residue N concentrations is important for overcoming net N immobilization in the spring. Results from these experiments suggest that in this environment it is preferable to graze rather than incorporate winter cover crops. Long-term ( ) meteorological records (Anon. 1986) show that drainage in August is likely to occur in 83% of years, whereas drainage in September is only likely to occur in 4% of years. Thus delaying grazing for as long as possible in winter is important in minimizing subsequent nitrate leaching losses from urine patch areas of soil. We thank Y. LeWarne and R. Gillespie for assistance with trial establishment, field sampling and chemical analyses, and A. Wallace for assistance with the statistical analyses. ADAMS, J. A., CAMPBELL, A. S., MCKEEGAN, W. R., MCPHERSON, R. J.& TONKIN, P. J. (1979). Nitrate and chloride in groundwater, surface water and deep soil profiles of central Canterbury, New Zealand. In The Agricultural Industry and its Effects on Water Quality: Progress in Water Technology 11 (Ed. S. H. Jenkins), pp Oxford: Pergamon Press. ALVENA S, G.&MARSTORP, H. (1993). Effect of a ryegrass catch crop on soil inorganic-n content and simulated nitrate leaching. Swedish Journal of Agricultural Research 23, AMATO, M., LADD, J. N., ELLINGTON, A., FORD, G., MAHONEY, J. E., TAYLOR, A. C.& WALSGOTT, D. (1987). Decomposition of plant material in Australian soils. IV Decomposition in situ of C- and N-labelled legume and wheat materials in a range of southern Australian soils. Australian Journal of Soil Research 25, ANON. (1986). Summaries of Water Balance Data for New Zealand Stations. New Zealand Meteorological Service, Miscellaneous Publication 189. Wellington, New Zealand: Ministry of Transport. ANON. (1995). Drinking-Water Standards for New Zealand. Wellington, New Zealand: Ministry of Health. BALL, P. R.& RYDEN, J. C. (1984). Nitrogen relationships in intensively managed temperate grasslands. Plant and Soil 76, COX, J. E. (1978). Soils and Agriculture of Part Paparua County, Canterbury, New Zealand. Wellington, New Zealand: New Zealand Soil Bureau Bulletin 34. CUTTLE, S. P., HALLARD, M., DANIEL, G.& SCURLOCK, R. V. (1992). Nitrate leaching from sheep-grazed grass clover and fertilized grass pastures. Journal of Agricultural Science, Cambridge 119, DAVIES, D. B., GARWOOD, T. W. D.& ROCHFORD, A. D. H. (1996). Factors affecting nitrate leaching from a calcareous loam in East Anglia. Journal of Agricultural Science, Cambridge 126, FRANCIS, G. S. (1995). Minimising nitrate leaching losses following cultivation of temporary leguminous pastures in mixed cropping rotations in Canterbury, New Zealand. Journal of Contaminant Hydrology 2, FRANCIS, G. S., HAYNES, R. J., SPARLING, G. P., ROSS, D. J. & WILLIAMS, P. H. (1992). Nitrogen mineralization, nitrate leaching and crop growth following cultivation of a temporary leguminous pasture in autumn and winter. Fertilizer Research 33, FRANCIS, G. S., HAYNES, R. J.& WILLIAMS, P. H. (1995). REFERENCES Effects of the timing of ploughing-in temporary leguminous pastures and two winter cover crops on nitrogen mineralization, nitrate leaching and spring wheat growth. Journal of Agricultural Science, Cambridge 124, 1 9. HAYNES, R. J. (1986). The decomposition process: mineralization, immobilization, humus formation, and degradation. In Mineral Nitrogen in the Plant Soil System (Ed. R. J. Haynes), pp New York: Academic Press. HAYNES, R. J.& FRANCIS, G. S. (199). Effects of mixed cropping farming systems on changes in soil properties on the Canterbury Plains. New Zealand Journal of Ecology 14, HAYNES, R. J.& WILLIAMS, P. H. (1993). Nutrient cycling and soil fertility in the grazed pasture ecosystem. Advances in Agronomy 49, JAMIESON, P. D. (1989). Modelling the interaction of wheat production and the weather. In Integrated Systems Analysis and Climate Impacts (Ed. R. W. M. Johnson), pp Wellington, New Zealand: MAF Technology. JENSEN, E. S. (1991). Nitrogen accumulation and residual effects of nitrogen catch crops. Acta Agriculturae Scandinavica 41, JENSEN, E. S. (1992). The release and fate of nitrogen from catch-crop materials decomposing under field conditions. Journal of Soil Science 43, MARTINEZ, J.& GUIRAUD, G. (199). A lysimeter study of the effects of a ryegrass catch crop during a winter wheat maize rotation, on nitrate leaching and on the following crop. Journal of Soil Science 41, MCCRACKEN, D. V., SMITH, M. S., GROVE, J. H., MACKOWN, C. T.& BLEVINS, R. L. (1994). Nitrate leaching as influenced by cover cropping and nitrogen source. Soil Science Society of America Journal 58, MCLAREN, R. G.& CAMERON, K. C. (1996). Soil Science. Sustainable Production and Environmental Protection. Auckland, New Zealand: Oxford University Press. MEISINGER, J. J., HARGROVE, W. L., MIKKELSEN, J. R. & BENSON, V. W. (1991). Effects of cover crops on groundwater quality. In Cover Crops for Clean Water (Ed. W. L. Hargrove), pp Ankeny, IA, USA: Soil and Water Conservation Society. RUZ-JEREZ, B. E., WHITE, R. E.& BALL, P. R. (1995). A comparison of nitrate leaching under clover-based pastures and nitrogen-fertilized grass grazed by sheep. Journal of Agricultural Science, Cambridge 125, SHEPHERD, M. A. & LORD, E. I. (1996). Nitrate leaching

10 38 G. S. FRANCIS, K. M. BARTLEY AND F. J. TABLEY from a sandy soil: the effect of previous crop and postharvest soil management in an arable rotation. Journal of Agricultural Science, Cambridge 127, SHEPHERD, M. A., DAVIES, D. B.& JOHNSON, P. A. (1993). Minimizing nitrate losses from arable soils. Soil Use and Management 9, STANFORD, G., FRERE, M. H.& SCHWANINGER, D. H. (1973). Temperature coefficient of soil nitrogen mineralization. Soil Science 115, UNITED STATES DEPARTMENT OF AGRICULTURE (1983). Keys to Soil Taxonomy. Soil Management Support Services Technical Monograph No. 6. WALLGREN, B.&LINDE N, B. (1994). Effects of catch crops and ploughing times on soil mineral nitrogen. Swedish Journal of Agricultural Research 24, WILLIAMS, P. H.& HAYNES, R. J. (1994). Comparison of initial wetting pattern, nutrient concentrations in soil solution and the fate of N-labelled urine in sheep and cattle urine patch areas of pasture soil. Plant and Soil 162,