1994

Journal of

Hydrology ELSEVIER

Journal of Hydrology 162 (1994) 23-46

[2]

Field measurements of water and nitrogen losses under irrigated maize L. Kengni a, G. Vachaud a'*, J.L. Thony a, R. Laty a, B. Garino b, H. Casablanca c, P. Jame c, R. Viscogliosi ¢ ~Laboratoire d~tude des Transferts en Hydrologie et Environement ( L THE/ IMG ) ( INPG, UJF, CNRS URA 1512), BP 53 X, 38041, Grenoble, France bLyc& Agricole, 38260 La Cdte Saint-An&& France ¢CNRS, Service Central d'Analyses, BP 22, 69390 Vernaison, France

Received 24 January 1994; revision accepted 9 April 1994

Abstract

An intensive multidisciplinary experiment has been conducted over several years at La C6te Saint-Andr6, near Grenoble, France. The major objective is to determine an optimal fertilizer application scheme for an irrigated agricultural system. Such a scheme would not degrade the quality of the environment, and yet would maintain a profitable level of crop production. This study is explicitly related to the cultivation of irrigated maize, a major crop in the area. The various terms of the water balance (consumption, drainage, soil storage) and of the nitrogen cycle (mineralization, plant uptake, leaching) were obtained from intensive monitoring in the upper layer of the 0.8 m of soil which corresponds to the root zone of the crop. This entailed the combined use of a neutron moisture meter, tensiometers and soil suction cups. To determine the specific effects of fertilization and crop growth, there were different treatments. These corresponded to a traditional fertilizer application of 260 kg N ha -1, no fertilization, and bare soil. carried out within an area of approximately 2 ha. Several sites were instrumented on each treatment, one of them being specifically for the application and the monitoring of 15N-tagged fertilizer. The results have shown that, in terms of the water balance, irrigation water management is extremely efficient, as drainage losses under the maize culture are negligible during the crop cycle. The situation is totally different, however, during the intercrop period (October-April I. owing to rainfall. Then the soil is left bare and evaporation is very small, and now the drainage corresponds to about 90% of total inputs from precipitation. In terms of the nitrogen cycle, the results showed clearly that up to 150 kg N ha -~ was produced by mineralization in the soil. Nitrogen leaching beyond the root zone during the

* Corresponding author. 0022-1694/94/$07.00 © 1994 - Elsevier Science B.V. All fights reserved SSDI 0022-1694(94)02526-H

24

L. Kengni et al. / Journal of Hydrology 162 (1994) 23-46

crop cycle is negligible, regardless of the rate of fertilizer application, as a result of the very small amount of drainage, despite irrigation. A very important contrast was found, however, between the fertilized and unfertilized treatments at harvest. There was a residue of 182 4- 64 kg N ha -I in the fertilized sites, but none for the others. The whole quantity remaining in the root zone at harvest was then totally leached by winter rains. To decrease the risk of groundwater pollution, a reduction of about 100 kg N ha -1 from the traditional application rate has been recommended. Finally, the method of estimation of N balance has been successfully validated by a comparison between N uptake determined by direct analysis of the whole plant and the value estimated from the temporal variations of the N content in the soil.

1. Introduction

Non-point-source agricultural pollution of ground water has become a real threat to the environment over the last 30 years. Although subject to some controversy, the relationship between nitrate pollution and agricultural practices is uncontested (Hrnin, 1980, 1981; Srbillotte, 1987). Furthermore, little is known of the impact of irrigation on the degree to which leaching of fertilizers can degrade the environment. In Europe, studies have shown that only 50-70% of the applied fertilizer is used by crops, the rest being volatilized, denitrified or leached (Guiraud, 1984; Chotte, 1986; Martinez, 1989). Leaching of nitrates is responsible for ground water pollution and may be increased by irrigation. This stresses the need to understand drainage from the root zone and to determine optimal fertilization rate in such a way that the amount of N remaining in the soil after harvest, and potentially leachable during winter, will be minimized, without unacceptably reducing crop production. Our contribution to the development of such a strategy has been made within the framework of a research programme initiated by the Commission of European Communities (DGXII-STEP). This multidisciplinary research involved teams from the University of Wageningen, the Catholic University of Leuven, the Agricultural University of Athens, the Spanish Science Research Council Institute of Natural Resources and Agrobiology Sevilla, and ourselves, and the expertise of Cornell University and DSIR, New Zealand. Several experimental sites were established, but a common approach has been used for the different edaphic situations, so as to characterize the various terms of the water balance and of the nitrogen cycle, and to determine the effect of fertilization on crop production, and to use a common simulation model. The latter was derived from L E A C H M , developed by Wagenet and Hutson (1989) at Cornell University, and will be used as a tool to predict the possible effects of a modification of cultivation and water management techniques. Our experiment began in 1991 and will run to 1994. The results which will be presented in this paper relate only to the first year, from April 1991 to February 1992. They have recently been reported in detail by Kengni (1993).

L. Kengni et al. / Journal of Hydrology 162 (1994) 23-46

25

2. Materials and methods

2.1. The soil The study was conducted on the Experimental Farm of the Lycre Agricole of La Crte Saint-Andrb, located 40 km northwest of Grenoble, France. The site is a typical glacial terrace, with soil of approximately 1 m thickness resting on a layer of gravel and pebbles of high permeability, of 10-20 m thickness. There is a water-table aquifer which varies in depth between 9 and 15 m from the soil surface. It has a high sensitivity to nitrate pollution. A large number of wells reach a nitrate concentration close to, or higher than, the European limit of 50 mg 1-I . Typical characteristics for the soil are listed in Tables 1 and 2, an important point being the increase, in percentage and size, of gravel and stones with increasing depth. As a result, the root zone is practically no deeper than 0.80 m. Augering is scarcely possible past 1 m. Because of the coarseness of the soil, and the dryness of the climate, irrigation is traditionally used in this region.

2.2. Spatial structure and sampling A plot of 2 ha was selected for this experiment on the basis of systematic sampling of soil cores, performed on a 17.5 m × 17.5 m grid (see Fig. I(A)). This was used to characterize the variability at the field scale and to define criteria for site selection (Kengni, 1993). The selection procedure was based on a geostatistical analysis of the value of the water storage at a given date, over a depth of 0.8 m, and followed the method described previously by Vachaud et al. (1985). The statistical results showed a unimodal distribution corresponding to a Gaussian form with mean 213.5 mm and a standard deviation I 11.5 mm. The skewness and kurtosis values, respectively near zero and three, confirmed this normality hypothesis. In terms of spatial structure, whatever the direction, the semi-variogram was flat. Table 1 Bulk density, coarse percentage, and soil granulometry (in % of the fine fraction) in the experimental field Layer 0-30 cm

6d Coarse A Lf Lg Sf Sg

Layer 30-60 cm

Layer 60-90 cm

M

a

CV

M

~

CV

M

1.339 40.0 17.5 23.3 17.7 16.6 24.9

0.069 7.0 1.0 0.7 1.4 0.7 1.3

5.2 17.5 5.7 3.2 7.9 4.5 5.1

1.362 71.5 18.9 22.3 15.3 15.7 27.7

0.045 2.8 0.9 1.1 1.9 1.3 3.3

3.3 3.9 4.9 4.9 12.2 8.3 11.9

1.270 61.0 13.9 17.7 8.6 13.8 46. I

~ 0.024 3.4 4.4 2.1 2.0 11.4

CV 1.9 24.5 25.0 24.3 14.8 24.8

~d, Bulk density On g ~TI-3); coarse, > 2 ram, in per cent o f the total weight; A, clay ( < 0.002 ram); Lf, fine loam (0.02-0.002 mm); Lg, coarse loam (0.05-0.02 ram); Sf, fine sand (0.2-0.05 mm); Sg coarse sand (20.2 mm); M, mean; o, standard deviation; CV, coefficient o f variation (%).

26

L. Kengni et al. / Journal of Hydrology 162 (1994) 23-46

Table 2 Soil chemical properties in the experimental field Layer 0-30 cm

CEC Ca Mg K Na Bases pH OM N C/N Calc

Layer 30-60 cm

Layer 60-90 cm

M

a

CV

M

a

CV

M

a

CV

8.48 7.26 0.73 0.47 0.03 8.49 7.0 2.57 0.12 11.7 0.9

0.7 0.8 0.2 0.1 0.0 1.0 0.1 0.7 0.0 3.4 0.3

8 11 23 29 13 12 1 28 23 29 37

7.63 6.19 0.78 0.31 0.05 7.32 7.0 1.00 0.09 7.7 0.8

0.1 0.9 0.1 0.1 0.0 0.9 0.2 0.6 0.0 1.2 0.4

10 14 16 22 13 12 2 60 14 16 57

5.44 7.59 0.60 0.12 0.05 8.36 7.4 0.69 0.05 7.4 2.0

1.3 3.0 0.I 0.0 0.0 2.8 0.3 0.1 0.0 0.4 1.1

24 39 21 29 26 34 4 12 15 6 56

In milliequivalents per 100 g of the soil (< 2 mm): CEC, cation exchange capacity; Ca, Mg; K; Na; Bases, exchangeable bases. In per cent of weight of the soil ( < 2 mm): OM, organic matter; N, total nitrogen; C/N, total carbon vs. organic nitrogen ratio; Calc, calcium carbonate. This is p r o o f that within the sampling distance there was no autocorrelation between the m e a s u r e m e n t sites. As a consequence, it was decided that if the distance between m o n i t o r i n g sites was higher than the grid distance chosen (17.5 m), their measurements could be considered as randomized, and analysed with classical statistics. W i t h that criterion, and for technical reasons, measurement sites were installed along two transects, parallel to the sowing lines a n d to the trajectory o f the irrigation system, and covering the different treatments selected for this experimentation. 2.3. The treatments

Different levels o f treatment were considered, to determine the a m o u n t o f leaching (water a n d nitrate) for conventional farmers' practices and to characterize the potential o f the soil to produce nitrogen by mineralization o f its organic matter, and also to obtain i n f o r m a t i o n on the response o f the crop to fertilization. First the plot was divided, as shown in Fig. I(A), into two subplots according to the fertilization rate. One p a r t received the application rate that is usual in this region, 260 kg N ha -1 o f a m m o n i u m nitrate, applied at sowing here on 22 April. The second part remained unfertilized. This partitioning was in the east-west direction. Perpendicular to this, there was a second partitioning. M o s t o f the plot was cropped with maize, but at the n o r t h e r n end, there was a small b a n d o f bare soil. Finally, within the cropped and unfertilized treatment, there were two other experiments: (1) on three replicate microplots (each one approximately 5 m 2 in area), 260 kg N ha - l o f tagged a m m o n i u m nitrate, enriched with 1.2788 a t o m % o f 15N, was applied to calculate the rate o f utilization o f the fertilizer nitrogen. Only one o f these microplots (T6) was fully instrumented, a n d is represented in Fig. I(A).

L. Kengni et al. / Journal of Hydrology ]62 {1994) 23-46 N o fertilization

Fortlllzatlon

( ..............................................................................

180.5

125.2

............................

27

107.7

~ ~< ...........................................................

90.2

72.7

55.2

37.7

20.2

Origin

)

2.7

: .......... :~": .............. ~NT8 "nn " 1" : .............. ~ T;J" "''':" In "~nn "n"nnn: ................ :''~" nnln''~5 .......... ~

~ .

~ '

;L ,!

o

oil

....

°

°il .I

. . . .

o

o ;i

°

o

o

~,.5

I.

•.

°

,o o °,.o.oo

"~,°

~°~

;

,

~ ~.

"~J~)':

"

.. ~ T ' ~ ° . ° o

. i'.;".'°

~'°,~*°

"°"ii: °','. ,,oo ,,

: o"o" °

~°,.°°

oo :

..i:o :

%.o

", o

° " "°

~,

o

° " i .... "

:~° : o' ~,~,,, :,, :~..:--,, ................~

Croppedso,

~

o-

so

il

. o

;

" " "~

':

o

o

--

107.5

....................... D l s t a n o e ( m )

P a t h o f r a v e l l i n g irrigator

LEGEND

Fe~lized cropped subpk)t L a b e l l e d 1 5 - N site

®

[]

M o n i t o r i n g site

o

Gmvirnet~

80 cm

sampling

>~

~q(

Pluviometer

NI

N2 :.

®

®o ii15

o > ~

neutron access tube

•

~0

Tensioleterlocation s a d

depths

• i70

®

0 i90 tm_J

$3 (~

\

S1

O

$2

(~ Suction cups and depths

NI, SI

30 c,,,

N 2, S2

so ,-,.

N3, $3

80 ca

Sowing lines

Fig. 1. Experimental field design in 1991. A, general plan; B, details of a monitoring site.

L. Kengni et al. / Journal of Hydrology 162 (1994) 23-46

28

On nine other replications not represented in Fig. I(A), various amounts of nitrogen, ranging from 0 to 275 kg N ha -I were applied in a classical agronomic fertilizer response trial. The corresponding results will not be analysed here. 2.4. Measurements

Altogether eight sites were instrumented for soil moisture and soil solution concentration measurements. Two were on bare soil (one fertilized: T1, the other unfertilized: T8); three on fertilized maize (T2-T4); two on unfertilized maize (T5 and T7); and one had 15N fertilization (T6). Their locations are given in Fig. I(A), and they were all instrumented as follows, as shown in Fig. I(B): one neutron access tube was installed to allow soil water monitoring every 10 cm down to 80 or 90 cm; five mercury tensiometers were installed vertically at 15, 30, 50, 70 and 90 cm; six suction cups at 30, 50 and 80 cm were installed and replicated in two series (northern and southern), to measure the variability of nitrogen concentration in the soil solution, even over such a short distance. There was also one rainfall recorder, at the level of the canopy, at each site. Furthermore, at Site T6, temperature gauges were installed at 2, 4, 8, 16, 30, 60 and 90 cm in the soil, and at 2 m in the air. Finally, a simple meteorological station was also instrumented close by. This provided measurements of Class A pan evaporation, air temperature and humidity, and sunshine duration. Values o f Penm a n - M o n t e i t h evapotranspiration (PET) were obtained from the local meteorological station o f the Saint Etienne de Saint Geoirs Airport (5 km away). Measurements o f soil moisture and samples of soil solution were taken weekly from June to October 1991 and then every 2 weeks until mid-February 1992. Tensiometer readings were recorded daily from June until harvest, and then every week until the onset of frost. Temperature, rainfall-irrigation and micro-meteorological data were obtained every 30 min with the use of a data acquisition system connected by modem to the laboratory. Agronomic measurements were made every 2 weeks on samples of the whole plant at every site with the exception of T6, the site with 15N fertilization. This allowed assessment of the total dry matter and nitrogen uptake. At harvest (4 October), yield and grain analyses were also performed with the same objectives. At T6, plant samples were collected only at harvest, to determine their nitrogen isotope ratio. The same level o f irrigation was applied on cultivated and bare sites o f each subplot, by the use of high-pressure apparatus, running at 8.5-9 kg pressure and a discharge o f 60 m 3 h -1 . The farmers followed a water rotation scheme according to the availability of the apparatus. Altogether, the total amount o f water input during the crop cycle was 540 mm, 216 mm of which was by irrigation from mid-June to September, applied in six separate applications. 2.5. Determination o f water balance

The water balance was calculated from the mass conservation equation: A S = R - D - AET

(1)

L. Kengni et al. / Journal of Hydrology 162 (1994) 23-46

29

where AS is the change in water storage in the root zone, R the rainfall and irrigation amount, D the drainage at the depth Zr of the root zone, and AET the actual evapotranspiration during a given period of time At. Runoff was neglected, as this was practically nil on this particularly permeable field site. Rainfall or irrigation were measured with standard rain gauges. The change in water storage was obtained by the difference of water storage in the root zone (from 0 to 0.8 m) as measured by neutron probe between two consecutive dates. The remaining unknown terms are D and AET. For most of this work, the drainage component D was calculated from Darcy's law: D = qAt = -K(O) grad H A t

(2)

where q is the mean volumetric flux density (mm day -1) during At, K(O) is the hydraulic conductivity (mm day -1) corresponding to the water content at Zr, and grad H is the hydraulic head gradient at this same depth. In Eq. (2), 0 and grad H are averaged over At, so it is therefore necessary to have small changes of water content at Zr (maximum 0.01 cm 3 cm -3) during At. This is due to sensitive dependence of K on 0 in the K(O) relationship. Furthermore, this method requires that this relationship be known at Zr for every site. The K(O) relationship was determined through the use of the 'zero flux plane' method, as described in detail by Vachaud et al. (1978), based on an analysis of soil moisture and soil hydraulic head profiles during periods of soil drying in the absence of rainfall, when there are no plants, or plants with very shallow root distributions. This essentially happened during spring and autumn for the cultivated sites, and after each irrigation for the bare sites. A typical result is given in Fig. 2 for the bare Site T1; every measurement site has been characterized correspondingly (Kengni, 1993). For simplicity of calculations, each curve was fitted with a power law as suggested by previous results (Vachaud et al., 1991). To increase the accuracy of estimation of the drainage, and because of the high sensitivity of K to 0, D was calculated on a daily basis from June to September. To do so, water content values were interpolated between weekly measurements using daily tensiometer readings and the water retention curve, obtained at the same depth and site. This used the correlation between all water content and hydraulic head measurements available. A fit of this relationship with the well-known Van Genuchten ana~ioo.o

4OO

10.0

q,&

,' i

1.o

a "~

:ffi 0.1 0.15

if K,.6.7E+II~m^

iiiiiiiiiiiii iiiiiiii

14k91 0

0.20 0.25 0~0 0.;15 W a t e r content, cm31cm3

.

0.1

.

.

.

.

0.2 O~ W a t e r content, c m ] / e m 3

Fig. 2. Soil water characteristics obtained in situ; an example for a bare site (T1).

'

'

0.4

L. Kengni et al. / Journal of Hydrology 162 (1994) 23-46

30

20

.

40

600

-200

t~

j

Rainfan

,

~

,

-

]

,

.

~

,

,

•

-300

.

.

.

.

,

.

.

.

.

D D D D [] [] • i t e m •

T e n s i o m e t e r depth, c m I •15 D3o o70 Ago

-400 0.30

~ ~

/'

i

~"D-Q"D--1..[]

o.2s

,~

::

:~

I -- --

-

i

..... ~ ..... ~.-..

I

~

5

0 l~un

20~un

2S.lun

30-Jun

~ui

Fig. 3. Example of reconstitution of the daily water content (B) from tensiometer readings (A). The calculated daily values of drainage at 80 cm (C) are shown for the bare soil site for TI over the period 19 June-4 July.

L. Kengni et al. / Journal of Hydrology 162 (1994) 23-46

31

lytical approximation (Van Genuchten, 1980) was used to obtain the water content value easily. An example of the application of this method is given in Fig. 3 for a short period (19 June-4 July). During late autumn and winter for the soil then bare, when measurements were not so intensive and evapotranspiration was small, another method was used to calculate D. This was based on the estimation of AET from the Penmann-Monteith estimation (PET), with an algorithm derived from that of Chopart and Siband (1988), and suggested that AET = PET on the day of the rain

(3)

AET(j) = PET(j)/2j on the following days

(4)

and

where j is the number of days after the last rain event. An empirical lower limit of AET = PET/8 was fixed. This drop-offis reasonable for a dry soil. Drainage was then calculated from Eq. (1). It is obvious that this estimation is better when rainfall (Table 3) is large with respect to evapotranspiration and changes of storage. That is why it was used in winter when the soil was bare, and in our conditions changes of storage are always close to zero. 2.6. Nitrogen movement and balance

The soil solution extracted from ceramic porous cups was immediately sterilized in situ with the use of a 0.4 #In Millipore filter (MILLEX SLHV 025LS, Millipore, Bedford, MA, USA) and then deep-frozen. Nitrate concentrations were measured at the Service Central d'Analyses CNRS, at Solaize, using liquid chromatography. The amount of nitrate nitrogen in a soil layer was then calculated from the formula N = [NO3-]0A,AZ

(5)

4.42 where [NO3-] is the nitrate content in the soil solution, 0• the mean water content of the layer Az and N is the nitrogen amount expressed in mass per area unit, 4.42 being the molar ratio between NO~ and N. It was assumed that the nitrate content of the soil solution extracted at 30 era was representative for the 0-30 cm layer. Likewise, we calculated that 50 cm applied for the 30-60 cm layer, as did 80 cm for the 60-90 cm Table 3 Water balance during the 1991-1992 experiment (mean value and associated standard deviation) Treatment

Bare sites Fertilized maize Unfertilized maize

16 April--4 October

4 October-12 February

Rainfall (ram)

Irrigation (mm)

Drainage (mm)

AET (ram)

Rainfall (mm)

Drainage (ram)

AET (ram)

323.4 323.4 323.4

227.4 216.6 238.2

292 (29) 83.3 (14.3) 87.0 (22.6)

236 (14.8) 457 (23.6) 460 (17.7)

320.6 320.6 320.6

292.5 (2.1) 278.5 (4.2) 279.0 (2.8)

29 29 29

L. Kengni et al. / Journal of Hydrology 162 (1994) 23-46

32

layer. With measurements available for all three depths, their summation gives the total amount of nitrate nitrogen in the soil profile, which can also be identified as the amount of mineral nitrogen, if it can be proved that volatilization losses are negligible, and that ammonium is not present in the soil solution. The Laboratoire d'Ecologie Microbienne, Universit6 Claude Bernard Lyon 1, carried out a series of tests on soil samples under laboratory conditions to quantify the processes of mineralization, denitrification, volatilization and carbon production as a function of temperature, and for various water content values. These tests indeed indicated that volatilization losses were negligible, and that ammonium was generally not present in the soil solution. Finally, the rate of nitrogen leaching below the root zone was obtained from the relationship LN = DCo.8

(6)

where D is the water drainage rate at 0.8 m as calculated above, and C0.s is the N content as N O ; at that depth. Obviously, this equation does not consider the effect of dispersion, but it has been shown by Kengni (1993) that the dispersive term of the full transport equation represents at most only 6% of the value of the convective term given by Eq. (6). 2.7. Fertilizer balance

Labelling with 15N is probably the only tool able to allow an evaluation of the major phenomena of the N cycle (Guiraud, 1984). Indeed, the selective labelling of the fertilizer N permits one to distinguish between the nitrogen in the plant and soil solution derived from the soil, and that coming from the fertilizer. During the last decade, many studies have been conducted using this approach. However, up to now, the 15N enrichment technique has only been applied to drainage water from lysimeters (Guiraud, 1984; Waiters and Malzer, 1990), on plants (Hahne et al., 1977; Bigeriego et al., 1979; Olson, 1980; Feigenbaum and Hadas, 1980; Recous, 1988; Varvel and Todd, 1992; Recous et al., 1992; Francis et al., 1993) or from soil samples (Bigeriego et al., 1979; Broadbent et al., 1980; Recous et al., 1992; Francis et al., 1993). Weaknesses in these approaches are the lack of representativity of lysimeter studies and the lack of reproducibility in time and space of destructive sampling techniques. We have tried here to take advantage of our available measurement techniques to characterize the change in the isotopic lSN dilution in the soil solution that we extracted via the suction cups. With this measurement, combined with the determination of the water flux as described above, we can not only determine the proportion of the fertilizer N that leaves the system through the drainage water during and after the growing season but also assess the amount of fertilizer N taken up by the crops. For the practical reason of the time of receipt of the tagged material, the lSN fertilizer was not applied until 5 June, 4 weeks later than the main fertilization. Then, a 15NH~--15NO~ fertilizer (260 kg N ha -1, with a mean isotopic excess of 1.2788 atom % of 15N) was applied using a hand sprayer, with successive crossed applications, over a surface of 2 m x 2.4 m at Site T6. As for the conventional experiment, soil solution samples were then extracted on a weekly basis, from

L. Kengni et al. / Journal of Hydrology 162 (1994) 23-46

33

ceramic porous cups installed at 30, 50 and 80 cm and duplicated to give two series. The samples were first analysed by liquid chromatography, to obtain the nitrate concentration, then reduced by dehydration and the isotope ratio was determined by mass spectrometry, at the laboratory of the Service Central d'Analyses, CNRS, at Solaize. At harvest, the dry matter content of the aerial parts (cut immediately above the soil) was determined. Soil samples from the 0 - I 0 cm and 10-30 cm layers were obtained by mixing nine separate cores taken from each microplot. Six cores were taken from the 30-60 cm and the 60-90 cm layers. The above-ground parts and roots (0-30 cm layer) were analysed for dry matter, total N and 15N. Soil inorganic N and organic C and 15N were determined according to Recous et al. (1992), at the laboratory of the INRA, Laon-Prronne. The unit of measure used here is the 15N atom per cent enrichment compared with a standard: 15N E% = (14N +15 N × 100) - 0.3663

(7)

where 0.3663 is the reference standard of air (Junk and Svec, 1958; in Mariotti, 1982). The amount of nitrogen derived from fertilizer in a given soil layer is then NE =

NT x E

E0

(8)

where NE is the mineral N issued from the fertilizer, Na- the total mineral N in the layer considered, E the ~SN enrichment in the same layer, and E0 the 15N enrichment of the fertilizer. Then, the amount derived from the soil mineralization, Ns is obtained through the difference Ns = NT -- NE. In the same way, LE =

LN x E

E0

(9)

LE is the amount of fertilizer leached.

3. Results and discussion 3.1. W a t e r balance Drainage

This permeable soil responds very quickly to any water input either of rain or irrigation, owing to its sandy and rocky nature. An example is given in Fig. 4 for a bare site (T 1) and for a cropped site (T3). Clearly, the tensiometer at 90 cm responded to a water application the same day. It should be noted, however, that the behaviour between the two sites was similar only up to 22 June, the date at which the root extraction term started to be important on the cropped site, and again after harvest. It is because of this rapid dynamic behaviour that it was necessary to establish on a daily basis the water balance with Darcy's law. On bare soil, the hydraulic head gradients showed that the soil was almost always subject to drainage. The drainage flux was a maximum on the day of rainfall; it

34

L. Kengni et al. / Journal of Hydrology 162 (1994) 23-46

I~ • • •¢m m~mm~

-100

-~oo

%

'-300 2-

~

[]

0

0

~-soo

[]

0

-4O0 []

o o o°

[]

0 [] o

= -600

~

sou

o

-700

[3

.800

-900

i

0

I

I

.101 ~i~ li~lll. --ii~~'~ ~ -300

~

=t

-600" 0.

ni~~

~ ~ ~I

o

.

.

.

.

.

.

.

.

.

-1oo i

4-200

|oj ,z: -300

-100 :: _2001 -300 2.o t ~

i

[::]

1.o ~. . . . . . . . . . . . . . . . . . . . . . . J o.o!

o .................

|

.0

o---o--~-

....

-~-

Water movement up

c~...o_.... ~ ............. o..o - . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . c~

o

.,.0

. . . , ~ ~

-2.0

,

,

'°L

June

.... o .... ...... ~ , ,

i

' ° f .............. T "

\

..............................................

i

in- ~o~,ql

....... *........... , i .............. ~.............. i J - ! ~ J l

July

August

September

October

November

Fig. 4. Daily tensiometer and hydraulic head gradient at 80 cm; responses for a bare soil (T1) and cropped site (T3).

35

L. Kengni et al. / Journal of Hydrology 162 (1994) 23-46

diminished rapidly and became negligible after a week (see Fig. 3). The same phenomenon was observed at each water application. The cumulative drainage during the period of reference (April-February) represented approximately 52% of total input. This was 540.0 mm for T1 and 561.6 mm for T8. The corresponding curve for T1 is given in Fig. 5. Clearly, these values are of no practical interest, as they are related to irrigated bare soil sites, but they are scientifically interesting, for the soil and cropped sites comparison is enlightening. On the cropped sites, cumulative drainage during the same period (see Fig. 5) was hardly more than 20% of total input despite six irrigations. Drainage occurred essentially before emergence of the plant, or at the end of the cycle when its physiological activity had stopped, or obviously after harvest when rainfall increased as well. It is important to note that during the active crop cycle, drainage losses were virtually negligible: they ranged from 0 to 20 mm, depending upon site and treatment. This highlights the efficiency of the sprinkler irrigation even on this permeable soil. In contrast, during the intercrop period, water losses were important, and were 800 lu 700

--

600 ii1

~,,;,,,-"i~ ":

_ _

~ .......

~. 41111

....... i ....... ~....... i:.:-~--;-;

:

:

k . , ~ l

"

:

"'

:

"i~-~i

i .......

.: ........

" .......

i

i

::

:

.... ~ ~---

~

.......

~.7]

.

.

.

.

.

.

.

:

:

........

~.......

~. . . . . . . . . . . . . . .

~. . . . . . . .

" .......

~ ...............

" .....

]"

]

- . . . . . . .

:. . . . . . . .

:

',

" ........

i

" ~

.

:

i

~ :

....... ~....... i

:

'

:

i

:

..... !i..... :: "

"

300

i

200

. . . . . . . .

,: . . . . . . .

- . . . . . .

:. . . . . . .

:. . . . .

- . . . . . . .

:

',

- . . . . . . . .

~' . . . . . . .

.: . . . . . . . .

-!

100 0 800

;

:

,,

:

700 ...

g soo 400 300

~"~l.i ,,

,,

,,

,

~

!!iiiii!iiii!!! !iiiiiiiiiiiiiiiiiii!!i!!!!!!i ii;ii

E 200 g~ 100 0

Apr

Miy

,Tun

Jul

Aug

Sep

Oct

Nov

I1~

ian

Feb

Fig. 5. Cumulative water losses on a bare and a cropped soil from 16 April-12 February (PET, potential evapotranspiration).

36

L. Kengni et al. / Journal of Hydrology 162 (1994) 23-46

similar f r o m one site to another, whatever the fertilizer treatment, representing a p p r o x i m a t e l y 90% o f the rain. These results are summarized in Table 3. Evaporation-evapo tr ansp ir at ion In terms o f evaporation or evapotranspiration, the c r o p p e d and bare sites behaved similarly, regardless o f the rate o f fertilization, until the end o f June. After this period, e v a p o r a t i o n fluxes increased significantly, as expected, on c r o p p e d sites. This was due to the establishment o f soil cover. F o r bare soil sites, evaporation was always lower than or equal to the P E T calculated for grass, whereas, for the cropped sites, this value was often reached and even exceeded at critical stages, in particular at flowering. Cumulative evapotranspiration values during the crop cycle m o u n t e d to 457 (+24) m m and 460 (4-18) ram, respectively, on fertilized and unfertilized sites, compared with 236 (4-15) m m on bare soil (see Table 3). 3.2. Dynamics o f nitrogen on bare soil The use o f bare sites was extremely useful to determine the rate o f p r o d u c t i o n o f 500

~e

400 .~. . . . . . . ' . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . a00

m 200

T. . . . . . . . . . . . . . . . . . . . . . . .

........................

" ....... ?....................................

........................

i. . . . . . . .

r . . . m. 30"60 . . . .I . .

• ~

I ........

,, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

,, ...................

~.x~.~.-:~-~

300

...................

~

200

. . . . . . . • . . . . . . . ~............. E2.~ ~. . . . . . . . . . . . . . . . . . . . . . . O tJ', . . . . . . . ~',.... .~.~,................. ',

100

..............................................

~. . . . . . . .

0 500 400

i

X !~K

.........

~-

[]

i

ix ~7 1 0 0

........

...

Apr

" .......

~xx,~,.j

"~-",,=

i

~ o . . . . . . i---m-~t- ~ L , - ~ i , _ A I_Nil " '~ATno:

" - T ' .

May

:

x , x - ! . . . . . . . q. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

,

Jun

;

,

,

Jul

;

,

•

Aug

i~

!

, ' " - , - x - ~ ~::- ~ . . . . . . . . . . . . . . . . . . . . . ""i.'A: • x x x i

•

-

Sep

i

_r,~]t

Oct

it3

~

Nov

.

Dec

~. . . . . . . . ~ --.

Jan

,

- -

Feb

Fig. 6. Temporal evolution of the amount of nitrate nitrogen in soil layers for the bare sites without (T8) and with (T1) fertilization.

L. Kengni et al. / Journal of Hydrology 162 (1994) 23-46

37

nitrogen for this particular soil by the mineralization of the organic matter. These sites were also used to characterize some of the parameters related to nitrate transport. Mineralization In Fig. 6 is given the evolution with time of the amount of nitrogen available in the layers 0-30, 30-60, 60-90 and 0-90 cm for the bare sites with fertilization (T1), and those without fertilization (T8). These data result from use of Eq. (5). It is clear that the production of nitrogen at T8 only results from mineralization, whereas at T1 the values obtained are the result of transport, transformation and degradation of the fertilizer in the soil, as well as mineralization of the organic matter. Thus, the amount mineralized at T8 was about 123 kg N ha -I during the period April-June, with a second peak of approximately 50 kg N ha -1 in autumn. This appears to be typical for this region, which is known to have soils with a good capacity for mineralization, owing to aerobic conditions and a substantial organic matter content (2-3%) in the upper horizon (0-30 cm). These results have been validated by incubation tests of soil samples at the Laboratoire d'Ecologie Microbienne, Universit6 Claude Bernard Lyon (internal report, 1991). These data showed that the rate of nitrification could reach 0.2 kg ha -1 day -1, within the temperature range 15-20°C. It was also clear that under normal conditions of water content prevailing in this permeable field, denitrification was not expected to occur. Finally, volatilization was considered a negligible process. Nitrate movement and leaching A simplified method of analysis, based on the curves given in Fig. 6, was used to characterize the transport of applied fertilizer in the soil. The assumption was made that the two sites were similar, and that the application of fertilizer had no effect on the mineralization rate. We know that these two assumptions might be subject to dispute, but, as a first approximation, by using them it was possible, at any time and 300

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x . . . ~ . ~ . . _ :. . . . . . . . . . . . . . . . . . . . . . . . . Z:

200

~ 150

..................

i ....

..................

4. . . . . . . . : []

U

~100

. . . . . . . . . . . . . .

......

~<

On

! ........

4. . . . . . O q . . . . . . . . O

. . . . . . . . . . . . . . .

": .........................

i )KX~<

4. . . . . . .

~ .........

. . . . . . . .

. . . . . .

/

',. . . . . . .

:. . . . . . . . .

:. . . . . . . . .

:. . . . . . . . .

.~

50 ........................

Apr

May

i ................

Jun

Jul

Aug

! ..........................................

Sep

Oct

Nov

F i g . 7. N e t t r a n s p o r t o f fertilizer f o r t h e b a r e sites.

Dec

Jan

Feb

L. Kengni et al. / Journal of Hydrology 162 (1994) 23-46

38

depth, to subtract from the value obtained at T l that measured at T8. The result (Fig. 7) was analysed in terms of the displacement o f a pulse of fertilizer applied at the soil surface and moving under the effect of rain and irrigation. It should at this point be added that measurements o f the changes of water content and of water pressure for the two bare soil sites showed that soil water flow was almost steady. A few important points then result from Fig. 7. First, the total mass of nitrogen remained constant, and approximately equal to that applied, at least until leaching started at the deepest layer in mid-August. Second, the displacement of fertilizer corresponded clearly to a convective-dispersive scheme. The responses at 30, 50 and 80 cm were fitted to the analytical solution corresponding to the displacement of a pulse applied at the soil surface moving under the steady flow o f water, and coupled with mass conservation o f the solute. The best fit corresponded to an average velocity of 0.5 cm day-] and a coefficient of dispersion in the range 0.7-1 cm 2 day -1 . It is obvious from these results that nitrate movement was very slow compared with the water dynamics in which the response time of the tensiometer at 90 era was less than 1 day. To explain this difference, the soil was regarded as consisting of two porosity regions. First, we consider here a structural porosity, where water transfers 500

I

400

....... ....... ................!.......]................:...............I-,°-,°,

....

300 e~

200 i

100

i

......

i .......

!

:

i_..~...:: ........ , ;g, : r-~ , ~

; ....... ,

i .......

i ........ ,

:

:

............ ,

+

~

: : :

0 500

.

(~

(s)

:

---. 400 ,~....... ~. . . . . . . . . . . . . :,:. . . ,::~.............. ,e~~

300

........

÷. . . . . . .

~. . . . . . . .

~. . . . . . ~-: . . . . . . . . . . .

200 ;i .......

'2....... _:........ !.... ~--':-

too + .......

~ ....... :i~-

:

Apt

'

May

i........................

..! ..........

i ....... Z---~---, ~ ....... i--~,--2•

Jun

Elm

',

Jul

•

Aug

"

Sep

-; ...............

;~

:

Oct

Nov

Dec

~<,

Jan

~ .........

I

~(

Feb

Fig. 8. T e m p o r a l c h a n g e s o f t h e n i t r o g e n a m o u n t , p e r soil layer, for u n f e r t i l i z e d (T7) a n d f e r t i l i z e d (T3) c r o p treatment.

L. Kengni et al. / Journal of Hydrology 162 (1994) 23-46

39

are rapid, induced by the heterogeneous nature of the alluvial material, perhaps through cracks and channels (Booltink and Bouma, 1991). Second, there is a secondary porosity, within the soil matrix or aggregates, in which movement of the solute is essentially occurring by convection-dispersion at smaller velocity. This is coupled with biochemical reactions and exchange with the porous matrix. This is essentially the mobile-immobile scheme described by Coats and Smith (1964).

3.3. Dynamics of nitrogen on cropped soil Local estimation of nitrate balance In Fig. 8 are given for one unfertilized (T7), and one fertilized (T3) site, the temporal changes of nitrogen under maize. The same layers as in Fig. 7 are considered. By comparison with the bare soil sites the following results are obvious. For the unfertilized site (Fig. 8(A)), the production of nitrogen obtained by mineralization of the organic matter was 140 kg N ha -1 . This amount was rapidly used by the plant. There was no more nitrogen available in the soil after early August, with the exception of a small production by secondary mineralization in September. For the fertilized plot (Fig. 8(B)), the pattern of nitrogen increase in the early stages of plant growth (from sowing to early July) was very similar to that obtained on the bare soil. Past this date, there was a strong decrease of nitrogen available in the soil, owing mainly, as we will show, to a large uptake by the plant. Finally, at harvest, a residue of 130 kg ha -1 was still available. This amount decreased quickly in October, owing to rainfall, and leaching loss. In terms of nitrate flux, leaching was small during the crop season regardless of the N application, mainly because drainage of water was almost negligible and the movement of nitrate was slow. Only 39 kg N ha -1 and 11 kg N ha -1, respectively, were leached under the two sites. From October to February, strong contrasts existed between the two treatments, as the corresponding numbers were 101 kg N ha -1 400

[ - 'hi - Cropped, feurflUzed 300

.

.

.

.

o-

C~p~,

~ , , ~

i ...............................

? .......

! .......

i

.

.

.

.

.

.

.

.

',

~200

,,

',

:

uu-:.--n--ni----i

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

,,l!!

t

m-m 0

A~r

'

~i'y--

~un'

Jul

Aug

Sep

Oct

Nov

Dec

Jan

Feb

Fig. 9. Cumulative nitrogen leaching at 80 cm for a cropped soil without (T7) or with (T3) fertilization treatment.

40

L. Kengni et al. t Journal of Hydrology 162 (1994) 23-46

and 15 kg N ha-1 (Fig. 9), respectively. The quantity of nitrogen remaining in the soil after harvest was totally leached by April 1992. This is a clear indication that the traditional application rate of 260 kg N ha-t is indeed excessive. Field N spatial and temporal variation

A similar analysis was attempted at the scale of the field. Following the pioneering work of Nielsen et al. (1973), it has been well known that extrapolation from point values to spatial averages is a real challenge. This was also particularly the case here, owing to the very strong variability found in the nitrogen concentration of the soil solution at any given depth, for a given treatment. It may be worth noting again that, for every date of sampling, six replications were available for the fertilized treatment, as against four rephcations for the unfertilized treatment. For these two treatments, an example of the temporal change in the nitrate concentration at 30 cm from April 1991 to February 1992 is given in Fig. 10. Clearly, apart from a few measurements in July, which can probably be explained by differences in root growth, the variability between the unfertilized sites was very small; most of the time the concentration was below 500 mg 1-1 . In contrast, the variability on the fertilized sites was higher, and the values could reach and even exceed 1500 mg 1-1 . It is also important to note that the 2000 Fertilized crop

]

I

i

laizs

t¢~ ~ 500

....................

...................

0 2000

i

........... * .....................................

I

t.

% .... i ....... t-t-r-ri? -

500

I

I

L, I

1ooo

h

nl

I

.............................

.

i ,o± ,,ii+

.................. ~: .......... II-.,.l~

I

<> 4N

IIII-

I °4' I

i l<..... V~ , ili,

i,i,a,i,

i

!

i

Unfertilized crop ]

1500 E 1000

- ~ ~nnn m

500

.

.

Apr

.

.

.

May

J ,

,"x m~,'~'-mmi,,,,,mam~ ,~,,mnn, n . • ,

.

Jun

Jul

Aug

Sep

Oct

Nov

i.

[],

Dec

Fig. 10. Temporaland spatial evolutionof the nitrate concentration under the f e ~ crop treatments at 30 cm.

•

Jan

Feb

and the unfertilized

41

L. Kengni et al. / Journal of Hydrology 162 (1994) 23-46

Table 4 Cumulative water and nitrogen losses under the maize root zone (mean value and associated standard deviation) Treatment

Fertilized maize Unfertilizedmaize

16 April-4 October

4 October-12 February

Water (mm)

Nitrogen (kg ha-1 )

Water (ram)

Nitrogen (kg ha-l )

83.3 (14.0) 87.0 (22.6)

22.6 (16.5) 14 (3.7)

278.5 (4.2) 279.0 (2.8)

173.9 (91.1) 15.4 (5.7)

trend was the same for the other depths; however, it was damped with increasing depths. A first conclusion is that the high variability in nitrate concentration is probably not endogenous through production of N by the soil. Rather it is exogenous and related to external sources. It is probable that the highest source of noise was the heterogeneity of fertilizer application, compounded by the non-uniform water distribution in the soil. For example, granular ammonium nitrate was used, and the existence of local pockets of higher than normal application is therefore not excluded. There was also the problem of the dilution o f the soil solution when samples were taken immediately after a water application event, with a high probability that the solution was not yet in equilibrium with the soil. Finally, on some sites, microheterogeneities in drainage owing to the variation in clay content were also probably responsible for the apparently low movement of nitrate. Owing to these possibilities, and because of the large standard deviations, it was decided to determine individually the water and nitrate balance at each site, and then to determine the average values on the fluxes. It was found that, in agreement with the previous results, during the growing season, leaching was rather small at the cropped sites, regardless of the N application (Table 4). Amounts of 22.6 (+16.5) kg N ha -1 and 14 (+3.7) kg N ha -1, respectively, were leached under the fertilized and unfertilized sites. F r o m October to February, losses were 173.9 (+31) kg N ha -1 and 15.4 (-4-5.7) kg N ha -I, respectively. The great importance of leaching on fertilized sites is highly correlated to the N remaining in the soil at harvest: 182 (+64), vs. 2 ( + l ) kg N ha-I on unfertilized ones.

Table 5 Estimation of plant N uptake per treatment during the 199I- 1992 experiment(mean value and associated standard deviationper treatment) Treatment

Dn (kg ha- 1)

Nr (kg ha-i )

Np (kg ha-l )

Fertilized maize Unfertilizedmaize

22.6 (16.5) 14 (3.7)

182 (64) 2 (1)

251 (63) 176 (4)

D,, leachingduring the cultural cycle;Nr, residue at harvest; Np, plant N uptake.

L. Kengni et at. / Journal of Hydrology 162 (1994) 23-46

42

Plant N uptake and yield Table 5 shows the estimates o f plant N uptake computed using the method described above. The nitrogen produced by mineralization of the soil organic matter appeared to be completely incorporated into the unfertilized crops, so that the residue at harvest was nil. The plant consumption reached 176 (+4) kg N ha -1 . This was not the case for fertilized crops, where plant uptake was 251 (+63) kg N ha -I from the 450 kg N ha -1 total nitrogen input from fertilizer and mineralization. By comparison, analyses performed independently at harvest on the whole plant gave 151 kg N ha -I and 293 kg N ha -I uptake, respectively, under unfertilized and fertilized crops. This appears to validate our method o f measurement. Finally, in terms of the plant response to fertilization, the differences in yield between the two treatments were small: 10.6 tons ha -1 vs. 13.1 tons ha -I (dry grain yield) for the unfertilized and fertilized treatments, respectively. This is another reason to suggest a substantial reduction of the fertilizer input.

3.4. Fertilizer recovery and use of ]SN Evolution of the fertilizer N The evolution with time of total N available, together with the partition between the amount derived from the fertilizer (filled symbols) and that derived from the soil (open symbols) is shown in Fig. 11. The behaviour here obtained is different from that shown in Fig. 8 for the traditional fertilization, owing to the late fertilization. In particular, owing to interactions between nitrogen transformations and plant uptake, the peak o f total nitrogen is smaller, followed by a faster decrease o f total nitrogen available in the soil. Thus, at the end o f the period of maximum plant uptake (mid-July), the soil is almost totally depleted. Two characteristic points merit discus250 xi

i

............................................

m 200

.....

o Son

,,,,r m is0

...................................

. ..... X

O ',O

gl 100

. . . . . . . . . . . . . . .

.o

.__~_._

:X--,.X ~ , ~ . . . . . .

.:i"

~ .........

, R"

0

~ . . . . . . . . . . . . . . . . . . . . . . . .

9-o---'~

....

o:

....

o-~

. . . .

o

:

z 50

"........

....

--"

: " - - ~

o

. . . . . . . .

:

i, ~, :m ix i [] . . . . . . . . . . . . . . . . . ~ . . . . . . . . . . ~-m~r~-m---:--m--- ~,-m-.....m',".... • ...... "i . . . . . . . . . ~. . . . . . . . , ',

•

,.

•

Jan

i

, , Feb

mm 0

A pr

iVlay

. . . .

Jun

J'ul

'

'

' Aug

L -

Sep

. . . Oct

.

. . . Nov

.

Dec

Fig. l 1. Evolution with time of N derived from fertilizer,from the soil and the total N available in the soil profile on the labelled site.

L. Kengni et al. / Journal of Hydrology 162 (1994) 23-46

43

sion. First, for an input of 260 kg N ha -1, and a plant uptake of about 220 kg N ha -1 from sowing to mid-July, the total amount of N fertilizer remaining as nitrate in the soil in mid-July is almost nil. The deficit in mass balance, at least 40 kg N ha -l, can only be explained by conversion of the fertilizer from mineral to organic form. Second, during a second mineralizing flush between mid-August and mid-September, there is an increase of about 50 kg N ha -1 of nitrogen fertilizer in the soil, owing to the mineralization of previously immobilized N fertilizer.

Nitrogen and fertilizer leaching As we have stated above, nitrogen losses are rather negligible before harvest, when the mineral N residue is 155 kg N ha -1, of which 32% is from the fertilizer N. Most of the N is found in the layer 0-60 cm. During the intercrop period, climatic conditions are, as we saw previously, favourable for nitrate leaching. With the continuous monitoring of the isotopic ratio in the soil solution concentration at 80 cm, and the use of Darcy's law, it has been possible to estimate the amount of nitrogen derived from the fertilizer that is leached. The result is summarized in Fig. 12. The proportion amounts to 28% for the entire intercrop period.

Recovery of the fertilizer N by the plant Finally, the fertilizer uptake was calculated as the amount of fertilizer N in the soil profile at sowing (260 kg N ha-l), less the fertilizer nitrate remaining at harvest (50 kg N ha-l), less the amount leached during the crop cycle (zero), less any denitrification or immobilization. Because of the climatic conditions, denitrification was assumed to be negligible (Grundmann and Chalamet, 1991), whereas at this site the equivalent of 41 kg N ha -1 of the fertilizer has been found incorporated into soil organic compounds at the end of the growing season. This value was obtained from nitrogen analysis (mineral and organic, including lSN) on soil samples taken on this site just after harvest by the Department of Agronomy, Institut National de la Recherche Agronomique, Laon. Thus the estimated plant fertilizer uptake was 169 kg N ha -1, i.e. 65% of that applied. This is in agreement with the result obtained directly from plant analyses conducted in the laboratory. I00 80 U

60

,

.........

,

..... o.i .........

f-~-----| "l

•

,

i

AddedN

I--'~'

,

! ....................

° .........

,

i

_

~' . . . . .

i

~'J'-"~

i ......... o

i

.........

,~ . . . . . . . . .

~J CJ m.

40 20 0

. . . . . . . . .

.........

Sep

, . . . . . . . . .

; .........

i

i

i ......

Oct

,. . . . . . .

•

s-~ .........

Nov

i

',

•

~i. . . . . . . . . .

Dec

i;;

..... ! : ~

~-. . . . . . . . .

Jan

....

~.........

Feb

Fig. 12. Partition between nitrogen from the fertilizer and nitrogen from the soil in the leached water during winter under the labelled site.

44

L. Kengni et al. / Journal of Hydrology 162 (1994) 23-46

To conclude, the total plant N uptake, originating from fertilizer and soil mineralization, calculated on this particular site by Kengni (1993) was 288.5 (4-20.5) kg N ha -1, compared with 300 kg N ha -1 on plant analysis. This comparison would appear to validate both computation methods.

4. Conclusions The experiment described here has demonstrated that the tensiometer-neutron probe method, coupled with the use of ceramic cup samplers, is a suitable method for assessing the water and nitrogen balance in field monitoring. The extreme variability of the soil necessitated, however, hydrodynamic characterization of each site. The water balance was rather different from one site to another, but at all sites it revealed an efficient use of the irrigation water. The use of ceramic porous cup samples was found to be satisfactory. Methodologically, this approach is convenient to characterize the time-dependent behaviour of the nitrogen in field conditions. Sampling can be repeated during the growing season without altering the water and nitrate dynamics, as plant and soil sampling at regular intervals might do. The nitrogen balance was subject to considerable variability at the field scale. However, in spite of this great variability, N leaching during the crop cycle was negligible, because of the small amount of drainage, whatever the rate of fertilization. There was in consequence a substantial residue of nitrate in the soil at harvest at the fertilized sites, although the difference in yield between the two treatments was small. This confirmed that the rate of fertilizer application was excessive, and indicated that more leaching below the root zone can be expected if the potential mineralization of the soil is not taken into account when choosing the N application rate. Substantial N leaching occurred during the intercrop period, such that by February the remaining nitrogen had passed below 60 cm depth. The 15N-labelled fertilizer monitoring gave good results. Finally, some additional analyses such as the amount of plant or other organic matter in the soil could lead to a better understanding of the N dynamics. The predominance of the mineral form vis-~t-vis the organic form of the nitrogen in the water flux should also be studied in more detail.

Acknowledgements The authors wish to thank the farmer's association in collaboration with the Lycre Agricole for encouraging the study and furnishing data as soon as required, the Laboratoire d'Ecologie Microbienne Lyon 1 for providing nitrogen transformation data on soil samples, and the Laboratoire d'Agronomie, INRA-Laon, for providing several comments on crop N uptake estimation and nitrogen balance in the soil. They are also grateful to Richard H. Cuenca, Oregon State University, USA, and Peter

L. Kengni et al. / Journal of Hydrology 162 (1994) 23-46

45

Ross, CSIRO, Australia, for helpful comments and advice. Funds for this work were provided by the Commission of the European Communities through the STEP-DGXII programme, by the Conseil Grnrral de l'Is&e and by the Charnbre Regionale d'Agriculture Rh6ne Alpes. A scholarship was kindly offered to the senior author by the Cameroon Government and the Centre International des Etudiants et Stagiaires.

References Bigeriego, M., Hauck, R.D. and Olson R.A., 1979. Uptake, translocation and utilization of JSN depleted fertilizer in irrigated corn. Soil Sci. Soc. Am. J., 43: 528-533. Booltink, H.W.G. and Bourna, J., 1991. Physical and morphological characterization of bypass flow in a well-structured clay. Soil Sci. Soc. Am. J., 55: 1249-1254. Broadbent, F.E., Rauschkolb, R.S., Lewis, K.A. and Chang, G.Y., 1980. Spatial variability of nitrogen-15 and total nitrogen in some virgin and cultivated soils. Soil Sci. Soc. Am. J., 44: 524-527. Chopart, J.L. and Siband, P., 1988. PROBE; programme de bilan d'eau. Mrmoires et travaux de I'IRAT, Montpellier, France, No. 17, pp. 76. Chotte, J.L., 1986. Evolution d'une biomasse racinaire doublement marqure (14C, 15N) dans un systrme sol-plante: 6tude sur un cycle annuel d'une culture de ma'is. Thrse Doct., Univ. Nancy, 169 pp. Coats, K.H. and Smith, B.B., 1964. Dead end pore and dispersion in porous media. Soc. Pet. Eng. J., 4: 73-84. Feigenbaurn, S. and Hadas, A., 1980. Utilization of fertilizer nitrogen-nitrogen-15 by field grown alfalfa. Soil Sci. Soc. Am. J., 44: 1006-1010. Francis, D.D., Doran, J.W. and Lohry, R.D., 1993. Immobilization and uptake of nitrogen applied to corn as starter fertilizer. Soil Sci. Soc. Am. J., 57: 1023-1026. Grundmann, G. and Chalamet, A., 1991. Transfert hydrique et transformation biologique de l'azote. Rapport d'activit6 1991, Univ. Claude Bernard Lyon, Laboratoire d'Ecologie Microbienne, 10 pp. Guiraud, G., 1984. Contribution du marquage isotopique ~ l'rvaluation des transferts d'azote entre les compartiments organiques et min&aux dans les systrmes sol-plante. Thrse Doct. d'Etat-Es Sciences Naturelles, Univ. P. et M. Curie, Paris, 335 pp. Habne, H.C.H., Kroontje, W. and Lutz, J.A., 1977. Nitrate fertilization. I: Nitrate accumulation and losses under continuous corn cropping. Soil Sci. Soc. Am. J., 41: 562-567. Hrnin, S., 1980. Rapport du groupe de travail "Activitrs agricoles et qualit6 des eaux', Minist~re de l'agriculture et de l'environnement, Paris, Tome 1, 58 pp. Hrnin, S., 1981. Le cycle de l'azote, les problrmes de fertilisation et de pollution. Comptes Rendus Acad. Agriculture de France, 67: 846-856. Kengni, L., 1993. Mesure in-situ des pertes d'eau et d'azote sous culture de mais irrigure: application ~ la plaine de la Birvre (Is&e). Thrse Doct. Univ. Joseph Fourier Grenoble I, 220 pp. Mariotti, A., 1982. Apport de la grochimie isotopique a la connaissance du cycle de l'azote. Thrse de Doct. Univ. P. et M. Curie, Paris, 476 pp. Martinez, J., 1989. Intervention d'une culture drrobre de Ray-Grass sur les transferts d'azote darts le syst&ne sol/plante/eau lors d'une succession culturale blr/mais: 6tude par tra~age isotopique 15N au cours d'essais au champ et en lysim&res. Thrse Doct., Univ. Perpignan, 183 pp. Nielsen, D.R., Biggar, J.W. and Erh, K.T., 1973. Spatial variability of field-measured soil-water properties. Hilgardia, 42: 215-260. Olson, R.V., 1980. Fate of tagged nitrogen fertilizer applied to irrigated corn. Soil Sci. Soc. Am. J., 44: 514517. Recous, S., 1988. Dynamique de l'azote en sol cultiv~: organisation microbienne des formes ammoniacale et nitrique, cons&tuences sur l'utilisation de l'azote des engrais par une culture de blr. Thrse Doct., Univ. Claude Bernard, Lyon 1, 189 pp. Recous, S., Machet, J.M. and Mary, B., 1992. The partitioning of fertilizer-N between soil and crop: comparison of ammonium and nitrate applications. Plant Soil, 144:101-111.

46

L. Kengni et al. / Journal of Hydrology 162 (1994) 23-46

Stbillotte, J., 1987. L'excts des nitrates dans les eaux: un probldme ~i rtsoudre, une lutte engagte. Perspect. Agric., 115: 12-17. Vachaud, G., Dancette, C., Sonko, M. and Thony, J.L., 1978. M&hodes de caracttrisation hydrodynamique in-situ d'un sol non saturt. Application 6 deux types de sol du Stntgal en vue de la dttermination des termes du bilan hydrique. Ann. Agron., 29: 1-36. Vachaud, G., Passerat de Sillans, A., Balabanis, P. and Vauclin, M., 1985. Temporal stability of spatially measured soil water probability density function. Soil Sci. Soc. Am. J., 49: 822-828. Vachaud, G., Vauclin, M. and Laty, R., 1991. Caracttrisation hydrodynamique in-situ de quelques sols soudano-sahtlien ~ l'aide d'un programme assist6 par ordinateur. IAHS Publ., 166:175-185 Van Genuchten, R.M.T., 1980. A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Sci. Soc. Am. J., 40: 892-898. Varvel, G.E. and Todd, A.P., 1992. Nitrogen fertilizer recovery by soybean in monoculture and rotation systems. Agron. J., 84: 215-218. Wagenct, R.J. and Hutson, J.L., 1989. LEACHM: leaching estimation and chemistry model. A process based model of water and solute movement, transformations, plant uptake and chemical reactions in the unsaturated zone. Continuum Vol. 2. Water Resources Institute, Comell University, Ithaca, NY. Walters, D.T. and Malzer, G.L., 1990. Nitrogen management and nitrification inhibitor effects on nitrogen15 urea: I. Yield and fertilizer use efficiency. Soil Sci. Soc. Am. J., 4:115-122.