SSSAJ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via ISI Web of Science (6)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Jenkinson, D. S.
Right arrow Articles by Powlson, D. S.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Jenkinson, D. S.
Right arrow Articles by Powlson, D. S.
Agricola
Right arrow Articles by Jenkinson, D. S.
Right arrow Articles by Powlson, D. S.
Related Collections
Right arrow Isotopes
Right arrow Nitrogen
Right arrow Other Forage Crops
Right arrow Nutrient Cycling
Published in Soil Sci. Soc. Am. J. 68:865-875 (2004).
© 2004 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA

DIVISION S-4—SOIL FERTILITY & PLANT NUTRITION

Turnover of Nitrogen-15-Labeled Fertilizer in Old Grassland

D. S. Jenkinson*, P. R. Poulton, A. E. Johnston and D. S. Powlson

Rothamsted Research, Harpenden, Hertfordshire, AL5 2JQ, UK

* Corresponding author (David Jenkinson c/o Kevin Coleman, kevin.coleman{at}bbsrc.ac.uk).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
In this paper we follow the fate of single applications of 15N-labeled fertilizer to old grassland, over a period of nearly 20 yr. In 1980 and 1981, 15N-labeled N was applied to two of the treatments on the Park Grass Continuous Hay Experiment at Rothamsted, started in 1856. The labeled N was applied at the same rate (nominally 96 kg ha–1 yr–1) and in the same chemical form (NH4 or NO3) as the unlabeled N normally applied as fertilizer to the selected treatments. After 19 yr, 69.6% of the N applied in 1980 as 15NH4 had been harvested in successive cuts of herbage, with a further 16.5% remaining in the soil. For 15NO3, 64.3% had been harvested and 13.8% remained in the soil. The 15N data were then used to calculate annual inputs of nonfertilizer N, annual losses of N and N turnover times in old grassland, assuming that the selected treatments were under steady-state conditions. The annual input of N from nonfertilizer sources (rain, dry deposition, N fixation by leguminous components of the herbage, etc.) was large: 39 kg N ha–1 yr–1 for the NH4 treatment and 31 kg for the NO3 treatment. Leguminous plants made up <2% of the herbage in both the NH4 and NO3 treatments. The annual loss from the NH4 treatment was 19 kg N ha–1 yr–1 and 24 from the NO3 treatment. The gross turnover time of N in the root compartment (which included plant crowns) was 1.41 yr for the NH4 treatment and 0.42 yr for NO3. The gross turnover time of soil microbial N was 2.13 yr (NH4) and 1.83 yr (NO3): for humus N (i.e., soil N not in roots or microbial biomass) it was 181 yr (NH4) and 116 yr (NO3).

Abbreviations: NUE, nitrogen use efficiency


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
THIS PAPER IS ABOUT the fate of single additions of 15N-labeled fertilizer to old grassland, followed over a period of nearly 20 yr. Our first aim was to measure how much of the labeled N was taken up by the herbage in the year of application. Our second was to follow the fate of the labeled N remaining belowground over the following years, measuring (at intervals) the 15N removed in the herbage and that remaining in the soil. The third was to draw up balance sheets for 15N applied to the different treatments and to use these balance sheets to quantify the turnover of N in old grassland, with particular attention to the annual input of nonfertilizer N.

The 15N experiment was superimposed on selected areas of the Park Grass Experiment at Rothamsted, in southern England. This experiment, started in 1856, tests the long-term effects of various combinations of fertilizer and lime on the yield and botanical composition of permanent grassland (Lawes and Gilbert, 1900; Warren and Johnston, 1964; Tilman et al., 1994). Herbage yields and detailed botanical records for the different fertilizer and liming treatments are available for well over a century.

One of the areas selected for our work with 15N-labeled fertilizer had received N as ammonium every year since 1856 and the other N as nitrate since 1858. The labeled fertilizer was applied in the same chemical form and at (nearly) the same rate as received by the selected treatments, so that our applications of 15N-labeled fertilizer would not disturb the N balance in the chosen treatments.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
The Park Grass Experiment
The experimental site has not been cultivated for at least 300 yr, although the characteristic ridge-and-furrow marks of medieval plowing are still faintly visible. At present there are 25 main fertilizer treatments; most are now subdivided for liming tests. On Park Grass, the fertilizer treatments are designated by numerals and the liming status is indicated by the lower case letters a, b, c, and d. Subplots a, b, c, and d are laid across the fertilizer treatments, sufficient CaCO3 being applied at intervals to give a target soil pH of 7.0 for the a subplots, 6.0 for b, and 5.0 for c (Thurston et al., 1976). The d subplots have not been limed since the 1880s, when the whole experiment received a small amount of chalk. The harvesting scheme used on Park Grass since 1960 is to take two narrow strips of herbage by forage harvester across each subplot for fresh yield and dry matter determination. The position of the strips on each subplot varies from year to year. The remainder of the subplot is then mown, the herbage left to dry in situ and then removed as hay. This allows seeds to be returned to the soil, as in the pre-1960 period, when the whole of each subplot was made into hay and then weighed. The herbage is cut twice a year, once in June or early July, followed by an autumn cut, usually in late October or early November. The autumn cut is weighed green and removed without being made into hay.

An account of the experimental layout and an overview of the yields is published at intervals (Lawes Agricultural Trust, 1991); detailed year-by-year yields are published annually (Results of the Field Experiments. Published annually by Lawes Agricultural Trust, Harpenden, UK).

Park Grass Soil
The soil is a moderately well-drained silty clay loam (22% clay, 29% silt, and 49% sand), overlying Clay-with-flints, which in turn overlies chalk at a depth of several meters (Avery and Catt, 1995). Two soil series occur on the Park Grass Experiment: the Batcombe Series and the Hook Series. Both have similar topsoil but differ in the subsoil, the depth to Clay-with-flints being more than 80 cm in the Hook and less in the Batcombe. Both are classified as stagnogleyic paleo-argillic brown earths by the Soil Survey of England and Wales; the USDA classification is Aquic Paleudalf.

Location of the Nitrogen-15 Experiment
Two subplots were chosen for the 15N experiment: 9b, which has received 96 kg NH4–N ha–1 every year since 1856 and 14d, which has received 96 kg NO3–N ha–1 since 1858. A third subplot (3d, unfertilized since 1856) was used as control; it did not receive labeled N. In 1990, subplots 9b and 14d were split, parts (now called 9/2b and 14/2d, respectively) continuing to receive N in the customary form and parts (now called 9/1b and 14/1d, respectively) receiving no further N. Our 15N microplots were located on areas that now receive no N. For convenience, treatments that receive (NH4)2SO4 every year will be referred to as the NH4 treatment, those that receive NaNO3 as the NO3 treatment, and those receiving no N as unfertilized or none, depending on the context.

Mean yields for the 20 yr 1980–1999 and soil analyses for these subplots are given in Table 1. Both dry matter and N are lost from the herbage during the haymaking process; calculated values for these losses are also given in Table 1. Haymaking losses (first cuts only) were obtained by comparing dry matter yield, measured on herbage cut green by forage harvester from narrow strips across each subplot, with dry matter yield, measured on hay cut and dried over the remainder of the same subplot. Both herbage and hay samples were analyzed for percentage of N and the N lost during haymaking calculated. The haymaking losses given in Table 1 are averages, measured on four unfertilized Park Grass subplots in 1992, 1993, and 1994 and on eight fertilized subplots over the same run of years.


View this table:
[in this window]
[in a new window]
  		Table 1. Treatments, soil characteristics, and yields in the Park Grass subplots used for the 15N experiments.

 
Nitrogen-15 Microplots
The microplots were laid out on a 4 by 13 m strip along the eastern edge of each of the selected subplots, centered along that edge. The labeled fertilizers were applied in solution to 1 by 2 m microplots, using a precision spreader specially designed to give uniform application (Woodcock et al., 1982). Each microplot was separated by a 1 m discard from its neighbors and from the edge of the strip. There were six replicates of each microplot, the three odd-numbered microplots receiving their labeled N on 17 Apr. 1980 and the three even-numbered on 15 Apr. 1981. The microplots received unlabeled N every year until 1989, in the customary form and rate given to the subplots on which they were located, except in the year when the labeled N was applied. In that year the microplots were covered, so that they did not receive any of the unlabeled N being given to the rest of the subplot. There was therefore no N gradient between each microplot and its surrounds. Only the central area was harvested from the microplots, leaving a 25 cm discard all round (see below). Experiments done in 1980 on winter wheat growing at Rothamsted on the same soil type, using the same techniques used here, showed that this discard was adequate to avoid measurable dilution of 15N in the harvested area by migration of 15N outwards and of unlabeled N inwards (Powlson et al., 1986).

The atom% excess of the (15NH4)2SO4 applied in 1980 was 4.901 and 4.813 in 1981; the corresponding values for Na15NO3 were 4.851 and 4.822.

Botanical Composition of Subplots
By the time the 15N experiment was laid down in the early 1980s, large differences in botanical composition had developed between the various Park Grass treatments and subplots; for details see Thurston et al. (1976) and Williams (1978). Averaging over all the data listed by Williams (1978), hay from the unfertilized treatment (3d) contained 13 species of grass, of which Festuca rubra L. usually made the biggest contribution; three species of legume, of which Lotus corniculatus L. made the biggest contribution and 14 forb species, with major contributions from Plantago lanceolata L. and Leontodon hispidus L. This plot contained 56 ± 11.0 (SD)% grasses by mass (average of 68 samplings over the 1877–1976 period), 8 ± 3.5% legumes and 36 ± 11.3% forbs. Hay from the NH4 treatment (9b) contained 10 grass species, (main contributors Arrhenatherum elatius and Alopecurus pratensis L.); one legume (Lathyrus pratensis L.) and four species of forb, none dominant. Hay from this treatment contained 95 ± 6.4% grasses (average of 32 samplings, 1914–1976), 2 ± 3.7% legumes and 3 ± 3.1% forbs. Hay from the NO3 treatment (14d) contained 13 grass species (main contributors Alopecurus pratensis L. and Arrhenatherum elatius); one legume (Lathyrus pratensis L.) and seven forb species, Anthriscus sylvestris (L.) Hoffm. being the most common. The hay from 14d contained 93 ± 4.8% grasses (average of 42 samplings, 1877–1976), 2 ± 2.2% legumes, and 5 ± 3.7% forbs.

After the first 20 yr or so of the main Park Grass Experiment, there were no long-term changes in the proportions of grasses, legumes, and other species in the subplots we used (Silvertown, 1980). There are of course considerable year-to-year variations in the proportion of these three constituents (plant guilds) and in the species composition within a guild (Dodd et al., 1995).

There were no significant changes in herbage yield over the period 1890–1992 on the unfertilized treatment (Plot 3d) or on the NO3 treatment (14d) (Jenkinson et al., 1994; yields on the NH4 treatment [9b] were not examined).

Sampling of Herbage and Soil
The microplots were cut by mechanical scythe, just before the rest of the subplot, in the year the 15N-labeled fertilizers were applied and in the following 2 yr. After that, the microplots were harvested along with the rest of the subplot, regardless of their 15N content, except in those years in which they were to be sampled and analyzed. The herbage on the microplots was cut just above ground level, leaving no more than 1 to 2 cm of stubble. Herbage from the outer 25 cm of each microplot was first cut away and removed, leaving a central area of about 0.5 by 1.5 m for yield measurement and analysis. The exact area sampled was always measured for each individual microplot. The whole of the herbage from the sampling area was dried at 80°C for 24 h and put through a hammer mill. A 30-g subsample was then finely ground for 3 min in a disk mill (Tema model T 100, Tema [Machinery] Ltd., Woodford Halse, Northants, UK).

Soil samples were taken in spring, before the subplots were due to receive their annual application of N fertilizer. The samples were taken from the central area of the microplots, excluding the outside 25 cm. Ten 4.75-cm diameter cores were taken per microplot at each sampling, the 0- to 23- and the 23- to 50-cm layers being kept separate. About 2.4% of the soil surface inside the 0.5 by 1.5 sampling area was thus removed each time the soil was sampled. Uptake of 15N by subsequent herbage samples was corrected to allow for these removals. In 1999 the 50- to 100-cm layer was also sampled, using a 2-cm diameter corer. The position of each core was recorded, so that earlier sampling points could be avoided. The moist soils were weighed and sieved (12.7 mm) in the field and the soil then dried at 40°C. Soil weights are expressed on an oven-dry basis at 105°C. The roots (which included crowns and stubble) were dried at 80°C. Roots from the 23- to 50-cm layer were added to those from the 0- to 23-cm layer. The dry root mass was roughly crushed and as much soil as possible shaken off. All soil thus removed from the roots was recovered and analyzed separately. The roots were then washed in an ultrasonic bath and redried at 80°C. Dry roots (whole sample) and soil (100 g subsamples) were ground for 3 min in a Tema disk mill before analysis.

Herbage samples (two cuts) were taken in 1980, 1981, 1982, 1986, 1991, and 1997: the soils were sampled in 1982, 1987, 1991, and 1999.

Analytical Methods
Herbage, root, and soil samples taken in 1980, 1981, 1982, 1986, and 1987 were analyzed for total N by Kjeldahl and 15N measured on a portion of the Kjeldahl distillate using a MicroMass 602 isotope ratio mass-spectometer (Pruden et al., 1985). All samples taken in 1991, 1997, and 1999 were analyzed for total N content and 15N enrichment using an automatic N analyzer linked to a mass spectrometer (Roboprep-Tracermass, Europa Scientific, Cheshire, UK). Herbage samples from the forage harvester strips across the main plots were dried at 80°C and analyzed for total N by LECO combustion (LECO CNS-2000, Leco Corp., St. Joseph, MI).

Soil microbial biomass C was measured by the fumigation/incubation method (Jenkinson and Powlson, 1976) on moist soil that had been stored at –15°C, thawed and given a 7-d conditioning incubation at 25°C before fumigation with CHCl3. Microbial biomass C was calculated from the relationship: Biomass C = (CO2–C from fumigated soil minus CO2–C from unfumigated control)/0.45. Biomass N was measured by the fumigation/extraction method (Brookes et al., 1985), from the relationship: Biomass N = (N extracted from fumigated soil minus N extracted from unfumigated soil)/0.54. Labeled biomass N was measured in the same way (Shen et al., 1989).

Statistical analyses were done using Genstat (Lane and Payne, 1997). Significant means P < 0.05.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Removal of Nitrogen in Herbage
Labeled and Unlabeled Nitrogen Removed in the Herbage in the Same Year that the Labeled Fertilizer was Applied
About 47% (mean of the 1980 and 1981 data for both the NH4 and NO3 treatments) of the fertilizer N was recovered in the first cut and a further 5% in the second. There were however minor differences between years and treatments (Table 2). Recoveries of fertilizer N were significantly greater in 1980 than in 1981, in both treatments and with both cuts. Conversely, first cut yields were less in 1980 than in 1981, but this effect was reversed in the second cut, even though the second cut was taken much later in 1981 than in 1980 (see Fig. 1) .


View this table:
[in this window]
[in a new window]
  		Table 2. Yields of herbage and removal of herbage N in the year that labeled fertilizer N was applied.

 


View larger version (19K):
[in this window]
[in a new window]
  		Fig. 1. Cumulative rainfall during the period from application of labeled fertilizer (17 Apr. 1980 and 15 Apr. 1981), up to the time the herbage was cut for the second time.

 
These differences in fertilizer N recovery and yield are related to rainfall. There was nearly four times as much rain in the period between fertilizer application and the first cut in 1981 than in 1980 (Fig. 1) so that losses by leaching and/or denitrification could well have been a little greater in 1981. Recoveries of labeled N were significantly larger with the NH4 treatment than with NO3, apart from the second cut in 1980.

Recoveries of Labeled Fertilizer Nitrogen in Herbage over the Whole Experimental Period
As the removal of labeled N was only measured every few years, removals in the years when measurements were not made were calculated by linear interpolation. Figure 2 shows the percentage of recovery thus calculated for the whole period. By the end of the 16- to 17-yr period, 64% (mean of 1980 and 1981 starts for both NH4 and NO3 treatments) of the labeled fertilizer N had been removed in the herbage (both cuts). However the salient feature of Fig. 2 is that the pattern established during the first year persisted throughout the whole experiment. Thus the NH4 treatment started in 1980 maintained its lead over the NO3 treatment started at the same time, from beginning to end.



View larger version (23K):
[in this window]
[in a new window]
  		Fig. 2. Cumulative recovery of labeled fertilizer N in herbage, with standard error of the mean (SEM).

 
Herbage Yields and Total Nitrogen Removal
Figure 3 shows the relationship between yield and N removal (both cuts and all three treatments considered separately) for the 6 yr in which the microplots were harvested. A close relationship: dry mass (DM) yield (kg ha–1) = 64.4 (N removal, kg ha–1); r2 = 0.936 exists between yield and N removal for all the microplots that did not receive N in the period immediately before cutting. Data from these particular microplots (i.e., all first and second cuts on the unfertilized treatment, second cuts alone on the NH4 and NO3 treatments between 1980 and 1986 and all cuts on all treatments after 1990) are shown as open squares on Fig. 3.



View larger version (20K):
[in this window]
[in a new window]
  		Fig. 3. Nitrogen removal and dry matter (DM) yield from microplots in 1980, 1981, 1982, 1986, 1991, and 1997. First cuts that received fertilizer N are shown by solid squares. First cuts that did not receive fertilizer and all second cuts are shown by hollow squares. The line shows the relationship: dry mass yield (kg ha–1) = 64.4 (N removal, kg ha–1); r2 = 0.936. Note that the regression is only for the microplots without N (hollow squares).

 
Labeled Nitrogen Remaining in Soil, Roots, and Microbial Biomass
Soil Mass and Total Nitrogen Content
Table 3 shows the masses of soil in the 0- to 23- and 23- to 50-cm layers and the total soil N content of these layers, averaged over the 1982 and 1987 samplings of the 15N microplots. Values for 1876 and 1999 are given for comparison: the 1876 measurements were made on the whole plot (then undivided) and the 1999 measurements on the appropriate subplot. The total N content of the top 50 cm of soil from the microplots varied from 8 to 10 Mg ha–1, in the order NH4 > unfertilized > NO3.


View this table:
[in this window]
[in a new window]
  		Table 3. Soil mass and total N content.

 
The data in Table 3, as well as other unpublished measurements from the Rothamsted Archives, indicate that there have been no consistent changes over the last 120 yr in the total N contents of soil (0–23 cm) from the unfertilized, NH4, and NO3, treatments in the main experiment, as would be expected if steady-state conditions had been attained. However Table 3 also shows that our measurements of N in soil from the NO3 microplots were markedly lower than other measurements on this treatment. Local spatial variability is the most likely explanation; the microplots represent only a very small part of the main plot area.

Belowground Labeled Nitrogen
The decline in belowground labeled N (i.e., labeled N in roots, soil from roots, soil 0–23 cm and soil 23–50 cm) over the whole course of the experiment is shown in Fig. 4 . There were no consistent significant differences between the two treatments, although toward the end of the experiment there was a tendency (significant at the last sampling) for belowground labeled N to be a little less in the NO3 treatment than in the NH4 treatment. After 1 yr, 36.3% of the original labeled fertilizer N remained belowground in the NH4 treatment, falling to 16.6% after 18 yr. The corresponding figures for the NO3 treatment were 37.6 and 13.4%.



View larger version (20K):
[in this window]
[in a new window]
  		Fig. 4. Recovery of labeled N in roots and soil (0–50 cm) including standard error of the mean.

 
After 1989, fertilizer N was no longer applied to those sections of the NH4 and NO3 treatments where our microplots were located. This change affected total N uptakes (not shown) and botanical composition but is unlikely to have had any major effect on the decomposition of the 15N-labeled residues then remaining in the soil—although we cannot be certain of this.

In 1999, the 50- to 100-cm layer contained 3.55 Mg N ha–1 (mean for both fertilizer treatments, data not shown); this layer contained <1% of the added 15N-labeled fertilizer and will not be considered further.

About one-tenth of the labeled N in the 0- to 50-cm layer was found below 23 cm in 1982 (mean for NH4 and NO3 treatments started in 1980 and 1981) and a similar proportion in 1999 (again, mean of all measurements).

Labeled Nitrogen Remaining in Roots
There was more than twice as much root dry matter in the NH4 treatment (19.8 Mg ha–1; mean of all microplots sampled in 1982 and 1987) as in the NO3 treatment (7.2 Mg ha–1). The total N contents of these roots were 255 and 80 kg N ha–1 for the NH4 and NO3 treatments, respectively. Not surprisingly, 1 yr after the labeled fertilizer was added, roots from the NH4 treatment contained more than twice as much labeled N as roots from the NO3 treatment (Table 4). This difference persisted for at least the first 7 yr (Table 4). In both treatments, the amount of labeled N in roots at the end of the first year fell by more than half during the following year: thereafter the rate of decline slowed. Throughout the experiment, N in soil shaken from roots was always less heavily labeled than root N, but more heavily labeled than bulk soil (data not shown). Much of the soil shaken from roots would have come from the rhizosphere. These findings suggest that such soil mixes very slowly with the surrounding soil, as might be expected in old uncultivated grassland.


View this table:
[in this window]
[in a new window]
  		Table 4. Labeled N in roots.

 
Soil Microbial Biomass
About 2.5% of the total soil N in the NH4 treatment was in the soil microbial biomass; the NO3 treatment contained a little more (3.6%; Table 5). One year after the application of labeled fertilizer, a much greater percentage of the total labeled N then remaining in the soil was present in microbial biomass; 17.8% for the NH4 treatment and 21.2% for the NO3 treatment. After another year, these percentages had fallen to 14.5 and 12.4%, respectively. These biomass figures are a little low because they were measured on the sieved soil alone: biomass was not measured in the soil that adhered to the roots and was subsequently shaken off. Although the mass of soil shaken off was <2% of that in the main bulk of sieved soil, it was more heavily labeled and would have contained >2% of the soil microbial biomass. Although our measurements were made on a single soil sampling in spring, other work (Patra et al., 1990) on an adjacent subplot of the NH4 treatment that has a slightly higher pH (Plot 9a), indicates that the quantity of microbial biomass varies relatively little throughout the year.


View this table:
[in this window]
[in a new window]
  		Table 5. Microbial biomass C and N in soil (0–23 cm), 1 and 2 yr after application of labeled fertilizer. All measurements made on soil sampled in 1982.

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
CONCLUSIONS
 REFERENCES
 
Overall Recovery of Labeled Nitrogen
These experiments show that NH4–N can be recovered very efficiently—after 19 yr, 86.1 ± 0.75% of that applied in 1980 was either still in the top 50 cm of soil or had been harvested in successive cuts of herbage (Table 6). Nitrate N was recovered somewhat less efficiently; the corresponding figure was 78.1 ± 1.07%. These high recoveries of labeled fertilizer N in grassland herbage and soil are by no means exceptional—see for example, Dowdell and Webster (1980), Power and Legg (1984), Whitehead and Dawson (1984), Bristow et al. (1987), and Cookson et al. (2000). Even applications of fertilizer five times larger than those we used can sometimes be quantitatively recovered from grassland herbage and soil (Delgado et al., 1996). High recoveries do however depend critically on time of application; thus Ledgard et al. (1988) found the largest recoveries when the fertilizer was applied just when active growth was starting, as in our experiments. Likewise, recoveries of 45 to 55% of labeled N by herbage in the year fertilizer was applied have been widely reported for temperate grassland (Dowdell and Webster, 1980; Whitehead and Dawson, 1984; Triboi, 1987; Bristow et al., 1987, Whitehead, 1995). Assuming that 15N unaccounted for was actually lost (i.e., that the deficit was not an artifact of our sampling and analytical techniques), the overall loss of 15N from the NH4 fertilizer added in 1981 was greater than from fertilizer added in 1980. Thereafter losses were very similar. Losses from NO3 fertilizer were consistently greater than from NH4 and greater when the fertilizer was added in 1981 than in 1980. Most of the loss occurred in the year the fertilizer was added.


View this table:
[in this window]
[in a new window]
  		Table 6. Nitrogen-15 balance sheet.

 
There was a soil moisture deficit throughout the 8 wk after addition of labeled fertilizer in 1980, but not in 1981 (Powlson et al., 1986). Leaching (but not gaseous losses of N as N2O and N2) can therefore be ruled out as a cause of loss in 1980 but not in 1981. Nitrous oxide can be lost from microsites in relatively dry soil in the course of both denitrification and nitrification (Stevens et al., 1997); such losses could well have occurred in both years. Our fertilizer N was at risk for a very short part of the year. Richardson (1938) showed just how quickly the fertilizer was taken up in spring by the plant–soil system on the same Park Grass treatments that we used. Averaging over 1931, 1932, and 1933, he found a half-life of 6 d for NH4 and 4 d for NO3.

Nitrogen Use Efficiency
The removal of labeled and unlabeled N from the NH4 treatment in the year of application (mean of 1980 and 1981 results, see Table 2) was 146.0 kg ha–1 and for the unfertilized plot was 56.6 kg ha–1. Calculated by the difference method, using the unfertilized plot as control, mean fertilizer N recovery (or N-use efficiency [NUE]) for a fertilizer application of 97.0 kg ha–1 (mean of 1980 and 1981 applications) is then 92.2%. This result must be treated with caution; the unfertilized plot is not a true control because of the differences that have developed over the years between it and the fertilized plots in botanical composition, pH, etc. Nitrogen use efficiency calculated from 15N data is much less (53.5%, mean for both years of the NH4 treatment, see Table 2) because the 15N in the herbage came from a single application of labeled fertilizer. The corresponding figures for the NO3 treatment are: NUE by difference, 90.7%; and from labeled N recovery, 49.8%. Had the application of labeled fertilizer been repeated, year after year, the NUE, as measured by 15N uptake, would slowly increase to a value closer to that given by the difference method. The true NUE from successive additions of 15N-labeled fertilizer would then be the sum of the labeled N harvested from the current year, plus that from the year before, plus that from the year before that, and so on. In other words, the NUE would be given by the asymptotes to the horizontal of the lines on Fig. 2 that show how cumulative recovery of labeled fertilizer increases with time.

Comparison between Grassland and Arable Land
It is instructive to compare the results from the Park Grass experiment with those from the sister experiment on winter wheat (Triticum aestivum) on Broadbalk (Table 7). Treatments given an (nominal) application of 96 kg N ha–1 in both experiments gave almost identical uptakes of total N. The overall recovery of labeled N (in aboveground removal, roots, and 0- to 23-cm soil) was also similar on both experiments: 84.4% on Broadbalk and 84.9% on Park Grass. However, aboveground removal of labeled N was markedly greater for winter wheat on Broadbalk (61.1%) than for herbage on Park Grass (47.9%). In contrast, more labeled N was retained in roots and soil on Park Grass (37.0%) than on Broadbalk (23.3%). Some (but not all) of this difference arises because the Park Grass roots still contained 7.8% of the labeled N originally added, whereas the wheat roots on Broadbalk were long dead when the soil was sampled. The key difference between the experiments is that the annual return of organic debris to soil is much greater in grassland than in arable land.


View this table:
[in this window]
[in a new window]
  		Table 7. Uptake of N by herbage growing on old grassland and by winter wheat growing on old arable land. Nitrogen-15 labeled fertilizers applied in same year to field experiments located 1 km apart on same soil type.

 
Annual Nitrogen Removals
Nitrogen is the nutrient that limits plant growth in most unfertilized nonarid ecosystems (Vitousek et al., 1997; Ågren et al., 1999). In our experiments herbage yields were governed by the amount of N the herbage was able to take up—provided that those herbage cuts that had just received fertilizer N were excluded (Fig. 3). Morrison et al. (1974) obtained a similar relationship for a wide range of grasslands in the UK. Nitrogen was therefore the key factor that limited the yield of unfertilized herbage in these microplots, whatever the differences in botanical composition, whatever the season, and in first and second cuts alike.

There was a response to fertilizer N (96 kg ha–1) in all the first cuts in which it was tested. However, although a few of the first cuts that had received fertilizer N (shown as solid squares on Fig. 3) followed the same relationship between yield and N uptake as cuts that had not received fertilizer N (open squares), most did not, indicating that in these plots N was no longer the key limiting factor. Whatever this limiting factor, it was not P or K, as both were in adequate supply (Table 1). First cut yields on plots receiving N fertilizer are frequently limited by water in the Park Grass Experiment (Cashen, 1947).

Calculating Annual Inputs of Nitrogen to the Soil–Plant System and Nitrogen Turnover Times
It is possible to calculate annual inputs of nonfertilizer N to the NH4 and NO3 treatments from these removal data, knowing the recovery of NH4 and NO3 fertilizer N from our 15N experiments. Figure 5 is an idealized diagram of the soil–plant system, in which each of the concentric shaded areas (i.e., the root and crown pool, the soil microbial biomass pool, and the humus pool) is under steady-state conditions. The humus pool is taken as the total N in the soil, less that in roots, crowns and microbial biomass. Defining the humus pool in this way is an oversimplification: it will contain free and bound fractions, as well as NH+4 fixed on the soil colloids. Fixed ammonium accounts for some 200 to 300 kg N ha–1 in the 0- to 23-cm layer of soil at Rothamsted, that is, about 3 to 5% of the N in our Park Grass soils (Bremner, 1959).



View larger version (41K):
[in this window]
[in a new window]
  		Fig. 5. Steady-state model for the turnover of N in fertilized grassland. Pools (shaded) are unchanged on a year-to-year basis—but not necessarily during the course of a year. Flows (in italics) are in kg N ha–1 yr–1. F is input of fertilizer N, I is input of nonfertilizer N, H is removal of N in herbage, D is herbage N returned to the soil during haymaking, L is loss of N from the soil/plant system, S is net formation of microbial biomass and humus N (=net mineralization of N) and P is formation of humus N (=decomposition of humus N).

 
Let
  1. F = annual input of N from fertilizer
  2. I = annual input of N from nonfertilizer sources
  3. H = N in both cuts of herbage, as measured in fresh green plant material
  4. D = herbage N returned to the soil during haymaking
  5. L = annual loss of N from the crop/soil system
  6. S = net annual mineralization of N from microbial biomass and humus. Under steady-state conditions, S is also equal to the annual formation of microbial biomass and humus N from decomposing roots.

All values are in kg N ha–1 yr–1, with the year running from March, shortly before the fertilizer was applied, to the following March, when the soil was sampled.

Consider the outer ring (i.e., the root and crown pool) in Fig. 5. For steady-state conditions inputs equal outputs, or

[1]

Let x be the fraction of the fertilizer N input (F) recovered in soil, crowns, roots, and herbage 1 yr after application.

Assuming, for the purpose of these calculations, that F, I, and S all behave alike, then

[2]

The errors introduced if F, I, and S do not behave alike (as in reality they do not) are considered later.

If y is the fraction of the fertilizer input (F) that is retained in microbial biomass and humus (but not roots and crowns) at the end of 1 yr, then, again assuming that F, I, and S all behave alike,

[3]

From Eq. [3]

[4]

Substituting S from Eq. [4] into Eq. [2]

[5]

Substituting L from Eq. [5] into Eq. [1]

[6]

Solving for I

[7]

This expression is formally similar to that derived by Powlson et al. (1986) for winter wheat. D, the quantity of N in the first cut of green herbage that is returned to the soil during haymaking, can be taken as the total loss of N during haymaking (as given in Table 1), if it is assumed that all the missing N is returned to the soil as shed seeds, broken-off leaves etc., and none is lost to the atmosphere as NH3 or other volatiles. Measurements of NH3 losses by Whitehead and Lockyer (1989) suggest that this is a reasonable assumption. They found that, at most, 10% of the N in cut grass was lost during drying to the atmosphere as NH3, under conditions that were much more favorable for losses than ours—higher N concentration in herbage, longer period of drying, exposure to successive periods of drying and wetting.

Since H, D, and F are known (Table 1), it now remains to estimate x and y from our 15N measurements. Consider first the NH4 treatment and using (where possible) a mean of the 1980 and 1981 data. The recovery of 15N in herbage after 1 yr was 53.5% (mean for both cuts and both years: Table 2). That in soil, crowns, and roots was 36.3% (single value for 1981 start, from Table 6), so that x = 0.535 + 0.363 = 0.897. A value for y can be obtained from the recovery of labeled fertilizer N in soil (0–50 cm) and roots after 1 yr (36.3%; Table 6), of which 11.1% was in roots (Table 4), so that y = (0.363 – 0.111), or 0.252.

Substituting these values for x and y in Eq. [7] gives 39 kg N ha–1 yr–1 for I, from which S is 46 kg N ha–1 yr–1 and L is 19 kg N ha–1 yr–1 (Table 8). The corresponding values for the NO3 treatment are I, 31; S, 63; and L, 24 kg N ha–1 yr–1.


View this table:
[in this window]
[in a new window]
  		Table 8. Inputs and outputs of N in fertilized grassland.

 
Sensitivity Analysis
We now address the question of the reliability of these calculations. The key assumption is that F, I, and S all behave alike, as set out in Eq. [2] and [3]. The assumption that x (the overall recovery of N in herbage, crowns, roots, and soil) was the same for all three was tested by varying x above and below its measured value. If losses from I were zero, (i.e., x = 1, rather than 0.897) then the calculated value of I for the NH4 treatment is 35, of S, 44, and of L, 14, all in kg N ha–1 yr–1. If the losses from I were 50% greater than those from F (i.e., x = 0.8455 rather than 0.897), then the calculated value of I is 42, of S is 46, and of L is 21. Applying the same arguments to S, if losses from S were zero, the calculated value of I is 34, of S is 44, and of L is 13. If the losses from S were 50% greater than those from F, then the calculated value of I is 42, of S is 47 and of L is 21. These calculations indicate that I and S are relatively insensitive to changes in x. A similar analysis showed that varying y up and down by 50% had relatively little effect on the calculated values of I and S. Once N is in the plant, whether it came from fertilizer, nonfertilizer input, or from mineralization of soil organic N, it is likely to be similarly distributed between parts that are harvested and parts that are not—that is, y is similar for F, I, and S.

Calculations of I from Eq. [7] are very sensitive to the assumption that all the N lost during haymaking is returned to the underlying soil surface in seeds, broken-off leaves, etc. Making the extreme assumption that none is returned and all is lost to the atmosphere, gives a value for I of 70 kg N ha yr–1 for the NH4 treatment, compared with 39 if all is returned. For reasons discussed earlier, at most 10% of the N lost during haymaking is volatilized, so I will be much nearer 39 than 70.

In applying steady-state conditions, as in Eq. [1], it must be remembered that our calculations are on a March-to-March basis. In perennial herbage, root N decreases in spring, as growth gets under way, and rises in the following autumn as N is translocated downwards to the roots (see Chaplin et al., 1990)—although Bausenwein et al. (2001) showed that with grasses the mobile N is held in overwintering leaves and crowns. Steady-state conditions can only be assumed if measurements are made at the same time each year.

Inputs of Nonfertilizer Nitrogen
These nonfertilizer inputs of 39 kg N ha–1 yr–1 for the NH4 treatment and 31 for the NO3 treatment are much larger than the input in rain at Rothamsted (9.9 kg N ha–1 yr–1 over the period 1980–1999: P.R. Hargreaves, personal communication, 2003). Legumes are rare and very patchy in the NH4 treatment and even more so in the NO3 treatment. The herbage of the NH4 treatment did however contain on average 2% of legumes (see Table 1) and these may have made a small contribution to the input of nonfertilizer N. Even larger inputs of N have been calculated for the Broadbalk long-term experiment on winter wheat at Rothamsted, where legumes are absent. Thus I for the plot on Broadbalk that receives 96 kg N ha–1 yr–1 as NH4NO3 was calculated to be 51 kg N ha–1 yr–1 (mean of measurements made in 1980 and 1981; Powlson et al., 1986). Inputs of this size are not confined to Rothamsted; Weigel et al. (2000) measured inputs of 50 kg N ha–1 yr–1 in one of the unfertilized plots on the Bad Laüchstadt long-term arable experiment in Germany. In our experiments, the nonfertilizer N cannot have come from depletion of the reserves of soil organic N, since there is no evidence of any great change in the size of these reserves over the last 130 yr (see Table 3).

The annual removal of N from the unfertilized treatment was 38 kg N ha–1 yr–1 over the period 1980 to 1999, deducting the N lost during haymaking (Table 1). This is rather more than the 27 kg N ha–1 yr–1 mean for the 1920 to 1923, 1940 to 1943, and 1956 to 1959 periods, when yields and N removals were measured directly on dry hay (Thurston et al., 1976).

The mean annual removal in grain and straw on the Plot (051) not receiving N on the Broadbalk Winter Wheat Experiment over the period 1980 to 1999 was 20 kg N ha–1 yr–1, with an additional 11 kg in drainage (mean of measurements over the 1991–1998 period: Goulding et al., 2000). The organic matter content of this particular plot has changed little over the last 100 yr, as in our Park Grass treatments. Nitrogen has been accumulating at slightly lower rates in two nearby old arable sites that were abandoned 120 yr ago and allowed to revert to woodland: 26 kg N ha–1 yr–1 on Geescroft Wilderness (in soil and trees) and 23 kg on Broadbalk Wilderness, both measured over the 1969 to 2001 period (Poulton et al., 2003). Clearly there are inputs of atmospheric N at Rothamsted that are largely independent of cropping. Goulding et al. (1998) estimated the deposition of combined N to a crop growing at Rothamsted in 1996 to be 43 kg N ha–1 yr–1, made up of inputs from rain (9 kg), and from dry deposition of NO2 (14 kg), HNO3 (16 kg), particulates (3 kg), and NH3 (1 kg). This figure for deposition of combined N implies that there is little room for biological N fixation by blue-green algae (Cyanobacter spp.) or other free-living organisms such as Azotobacter, in accord with Stevenson and Cole's (1999) estimates of fixation by nonsymbionts.

It should be noted that the results in this paper and Goulding's estimated annual input of 43 kg N ha–1 for Rothamsted are markedly greater than Metcalfe et al.'s (1999) estimate for lowland England and that for Western Europe by Holland et al. (1999).

Turnover Time of Roots
The gross turnover time of N in a given pool can be defined as (N in pool)/(quantity of N moving through pool in 1 yr). For roots (including crowns), Table 9 gives the gross turnover time thus defined as 1.41 yr for the NH4 treatment and 0.42 yr for the NO3 treatment. If N is lost from roots by exponential decay, the corresponding rate constant r for the NH4 treatment is 1/1.41 or 0.71 yr–1; for the NO3 treatment r is 2.38 yr–1. Thus root turnover is markedly faster in the NO3 treatment—almost certainly an indirect effect caused by differences in the botanical composition of the plots (see Williams, 1978) rather than by the direct exposure of the roots to NH+4 or NO–3 ions for a week or so once a year.


View this table:
[in this window]
[in a new window]
  		Table 9. Turnover of root N, microbial biomass N, and humus N in the 0–23 cm layer of fertilized grassland.

 
Turnover Time of Microbial Biomass
Table 9 gives the gross turnover time of N held in the soil microbial biomass to be 2.13 yr in the NH4 treatment and 1.83 yr in the NO3 treatment. Assuming that N is an integral part of the microbial cell, that all members of the soil population are alike and that death follows an exponential process, the corresponding rate constants for microbial death are 0.47 and 0.55 yr–1. The half-lives are then ln2/r or 1.48 and 1.27 yr, respectively. These turnover times are comparable with those used for microbial C in RothC, the Rothamsted model for the turnover of organic in topsoils (Coleman and Jenkinson, 1999). At any one time, most of the microbial C and N in these grassland soils are held in organisms that are not actively dividing (Patra et al., 1990).

Decay of Humified Soil Organic Nitrogen
Let the quantity of N entering the humus pool in the top 23 cm of soil each year be P (Fig. 5) and z be the fraction of the added labeled N retained in this humus pool at the end of 1 yr. For the NH4 treatment, z can be taken as the fraction of the added labeled N remaining in the 0- to 50-cm layer of soil (0.363, as given in Table 6), less that in the 23- to 50-cm layer (0.012), less that in roots (0.111, as given in Table 4), less that in microbial biomass (0.033, as given in Table 5), or 0.207. Then, assuming, as before, that F, I, and S all behave alike,

[8]
assuming that the loss of label from the humus pool is negligible during the year following the addition of fertilizer, P is then 37 kg N ha–1 yr–1, the gross turnover time 181 yr (Table 9) and the rate constant 0.0055 yr–1. The corresponding values for the NO3 treatment are z = 0.250, P is 48 kg N ha–1 yr–1, the gross turnover time 116 yr and the rate constant 0.0086 yr–1.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
REFERENCES
 
Roughly one third of the 15N-labeled N fertilizer applied to old grassland remained in roots, plant crowns, soil microbial biomass and humus by the end of 1 yr, with just over half harvested in the herbage. By the end of 19 yr the fraction in roots, crowns, and soil had fallen to some 15%. Because the old grassland in this long-term experiment could reasonably be assumed to be under steady-state conditions, these 15N data can be used to calculate the flux of N through roots, microbial biomass, and humus, as well as the annual input of N from non-fertilizer sources.

In a treatment that had received 96 kg NH4–N every year since 1856, the calculated annual input of nonfertilizer N was 39 kg N ha–1 yr–1 over the 20 yr 1980–1999. The corresponding input of nonfertilizer N to a treatment that had had 96 kg NO3–N since 1858 was 31 kg N ha–1 yr–1. These N inputs to old grassland are of similar magnitude to those measured recently in woodland and in arable experiments at Rothamsted and imply a large annual input of combined N from the atmosphere, much greater than the 10 kg N ha–1 yr–1 in rain.


   ACKNOWLEDGMENTS
 
We are indebted to the late Gordon Pruden for the 15N analyses done during the early years of this study. We thank A.R. Todd for statistical analyses. We are grateful to the following who assisted with field, laboratory and library work: D.S. McCann, J. Brown, Jean Devonshire, the late E. Bird, N. Mepham, A.J. Macdonald, P.A. Cundill, and Maggie Johnston. We also thank the many visiting scientists who have contributed to the work, both practically and in discussion, in particular R.H. Fox, P.B.S. Hart, D.D. Patra, H. Setatou, S.M. Shen, K.R.Tate, and P.B. Tinker. We appreciate the thoughtful comments made by the three reviewers. Rothamsted Research receives grant-aided support from the UK Biotechnology and Biological Sciences Research Council. The Lawes Agricultural Trust also provided support.

Received for publication April 15, 2003.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 





This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via ISI Web of Science (6)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Jenkinson, D. S.
Right arrow Articles by Powlson, D. S.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Jenkinson, D. S.
Right arrow Articles by Powlson, D. S.