Você está na página 1de 18


Agricultural Water Management 28 ( 1995) 95-112

Leaching of lime and fertilisers from a reseeded

upland pasture on a stagnogley soil in mid-Wales
S.P. Cuttle *, A.R. James
Institute ofGrassland and Environmental Research, Plas Gogerddan, Aberystwyth, Dyfed SY23 3EB, UK

Accepted 17 November 1994


Concentrations of nutrients were measured in drainage water from a drained but otherwise unim-
proved area of Molinia grassland and from a similar area following pasture improvement in Year 1
of the 4-year study. Application of lime and fertilisers to the improved area and reseeding with a
grass/ white clover mixture had little effect on concentrations of nitrate and ammonium-N in drainage
water except for one year when nitrate concentrations were increased to a maximum of 24 mg N l- ’
for a brief period following application of nitrogen fertiliser. Concentrations of organic-N, potassium,
phosphorus and calcium and the pH of water samples increased following pasture improvement and
were consistently greater than corresponding values for the Moliniu area. By Year 4 almost all of the
potassium fertiliser applied to the reseeded area in Years 1 and 2 had been leached. In the same period,
about 12% of the phosphorus fertiliser and 24% of the lime application had been leached.

Keywords: Fertilisers; Leaching: Lime; Pasture; Upland; Water quality

1. Introduction

The productivity of native hill pastures in the UK is limited by climate, low productivity
of the native species, soil acidity and shortages of available nutrients (Munro and Davies,
1973; Munro et al., 1973). The development of suitable land improvement techniques
involving additions of lime and fertilisers and the introduction of improved grass species
and white clover, together with the provision of financial assistance to encourage farmers
to increase the productivity of their hill land, led to an expansion in the area of pasture
improvement in the uplands during the 1970s and early 1980s.
In Wales, much of the land remaining available for improvement was represented by
native grassland on peaty soils overlying impermeable subsoils. In these high rainfall areas,

* Corresponding author. Tel.: 01970 828255; fax: 01970 828357

0378-3774/95/$09.50 0 1995 Elsevier Science B.V. All rights reserved

SSDIO378-3774(95)01 174-9
96 S.P. Cuttle, A.R. James/Agricultural Water Management 28 (1995) 95-112

these soils are waterlogged for much of the year and excess water is an important factor
influencing their management after improvement. Artificial drainage would be expected to
reduce the risk of poaching by stock and increase the productivity and persistency of sown
species. However, these sites, even after improvement, are of limited agricultural potential
and information is required for assessing the cost-effectiveness of any drainage measures.
The changes in agricultural management associated with pasture improvement are also
of significance in the wider context of land use and water resources. The uplands of Wales
are an important source area for the water supply industry, traditionally providing high
quality water, largely free of contamination from industry and intensive agriculture. The
expansion in the area of pasture improvement in these upland catchments raised concerns
that intensification of agricultural activity would lead to increased leaching of nutrients,
particularly nitrogen and phosphorus, and reductions in water quality. Conversely, pasture
improvement may in some circumstances have beneficial effects on water quality where the
application of lime results in increased leaching of bases and neutralisation of the acidity
arising from atmospheric acid deposition.
By the end of the 1980s this period of agricultural intensification had been replaced by
one of increased concern about overproduction of agricultural commodities and the loss of
semi-natural habitats. If future policies lead to a reduction in agricultural activity in the
uplands, it is probable that inputs of lime and fertilisers to pastures on these relatively
difficult-to-manage soils will be reduced. Predictions of how land-use changes may influence
water quality require an understanding of rates of depletion of fertiliser nutrients from
improved pastures and the contribution that leaching currently makes to nutrient loadings
to surface waters in these upland areas.
This paper describes an experiment that was started in 1983 to compare the cost-effect-
iveness of alternative drainage techniques and to measure leaching of nutrients following
reseeding of an area of Molinia grassland. Results of the leaching study are presented and
interpreted in terms of the effects of pasture improvement on water quality, the likely
duration of these effects and fertiliser inputs required for the maintenance of reseeded
swards on peaty soils. Results of the drainage comparisons are not considered in detail.

2. Materials and methods

2.1. Site

The experiment was established on a relatively uniform area of unimproved moorland at

Pwllpeiran Experimental Husbandry Farm, near Aberystwyth (SN 789756) in mid-Wales.
The site was at an altitude of 400 m with a slope of about 4” and a southerly aspect. Soils
were cambic stagnohumic gleys of the Wilcocks series (Rudeforth et al., 1984) with a 0. l-
0.3 m depth of humified peat overlying a clay loam subsoil developed from compacted drift
material. At the start of the experiment, the vegetation was dominated by tussocks ofMolinia
caerulea L. with less abundant Eriophorum uaginatum L., Vaccinium myrtillus L., Juncus
squarrosus L., Festuca spp. and mosses.
A drainage trial had been conducted at the site between 1960 and 1968 and the open
ditches which survived from this earlier work defined the layout of the present experiment
S.P. Cuttle, A.R. James /Agricultural Water Management 28 (1995) 95-112 97


0 0

0 D2 0

0 0

0 D3 ’
0 0
0 0

0 0
0 0
0 0
0 0

0 0
50 m
Dl I
0 0

Fig. I. Plan of the leaching experiment at Pwllpeiran showing the reseeded area (shaded) and Molinia area with
cut-off ditch (ABC), collection ditches leading to weirs/sampling points (Wl and W2) and position of the
automatic weather station ( AWS). Dipwell positions are indicated by open circles. See Table 1 for a description
of the drainage treatments (DlLD4) applied to subplots.

(Fig. 1) . This provided two hydrologically isolated areas, each of approximately 1 ha.
Water draining from each area was collected by separate 1 m-deep ditches along their lower
sides while a 0.6-1.0 m-deep cut-off ditch along the upper boundary prevented entry of
water from outside of the experimental area.
At the start of the investigation the experimental area was drained as part of an associated
study by the Agricultural Development and Advisory Service to compare the cost-effect-
iveness of alternative drainage treatments. The two halves of the area were divided into
subplots with the same four drainage treatments applied to both halves (Table 1). Although

Table I
Drainage treatments installed in the reseeded and Mdiniu areas in 1983-1984

Drainage treatment Drain spacing (m) Drain depth (mm)

Dl Drains (plastic pipe) 11 750

D2 Drains with permeable till 22 650
D3 Drains with permeable fill 22 650
+ mole drains 1 450
D4 Mole drains alone, with gravel in mole channel and slot 5 450
98 S.P. Cuttle, A.R. James/Agricultural Water Munagement 28 (1995) 95-112

this subdivision introduced undesirable complications into the interpretation of the leaching
data, both of the hydrologically separated areas contained the same treatments, differing
only in their relative positions, and for the purpose of the nutrient leaching studies the
overall hydrological properties of the two areas were assumed to be similar. Plastic under-
drains were installed in November 1983 and moling operations were carried out in October
An automatic weather station adjacent to the main plots provided hourly values of rainfall
which were supplemented by weekly values from a standard bulk collection gauge. The
automatic station was removed in April 1987 and only the weekly rainfall measurements
were continued after this date. The automatic weather station also provided data for the
calculation of potential evapotranspiration using the Penman-Monteith equation for a short
grass surface (Monteith, 1965). Soil moisture deficits in the rooting zone (to about 700
mm depth) in Years 1 to 3 were estimated from the difference between rainfall and evap-
oration, assuming that the first 50 mm of available water in the soil profile would be
evaporated at the potential rate and the next 50 mm at half this rate (Smith, 1976). Water
table depths were measured at weekly intervals in a series of dipwells installed in the
drainage subplots and in adjoining undrained areas following the completion of moling
operations in autumn 1984.

2.2. Pasture establishment and management

The treatments included in the earlier drainage trial involved the excavation of shallow
surface ditches and installation of tile drains but no other pasture improvement measures.
For the present experiment, one of the areas was left as Molinia grassland and the other
reseeded using a minimal cultivation technique appropriate to wet hill land conditions. In
December 1983, a rotary mower was used to remove Moliniu tussocks from the area that
was to be improved. Lime was applied in April 1984 (Table 2) and two weeks later the
area was sown with a seed mixture of S.23 and Perma perennial ryegrass (Lolium perenne
L.), S.59 red fescue (Festuca rubra L.), S.48 timothy (Phleum pratense L.) and S.184
white clover (Trifolizdm repens L.). The strip seeder that was used limited soil disturbance
to the cultivation of 25 mm-deep strips affecting about 30% of the ground surface. A
Rhizobium inoculum was applied to the sward in June 1984. A compound NPK fertiliser
and triple superphosphate were applied at the time of sowing and further dressings of
compound fertiliser were applied in June and August 1984. Details of the quantities of
nutrients supplied as fertiliser are given in Table 2. In each subsequent year, ammonium
nitrate fertiliser was applied to the pasture in April. A further dressing of PK fertiliser,
corresponding to the recommended maintenance requirements for a 3-year period (Ministry
of Agriculture, Fisheries and Food, 1973; Ministry of Agriculture, Fisheries and Food,
198 I ), was applied in August 1985. No lime or fertilisers were applied to the Molinia area.
The abundance of native and sown species was assessed in autumn each year by recording
the ground cover of each species within randomly positioned quadrats.
The two areas were fenced and grazed as separate units but without further subdivision
between drainage subplots. Both areas were grazed continuously by Welsh Mountain sheep
between April and October with sheep numbers adjusted on the basis of visual assessments
of the herbage available. Mature sheep were used throughout the experiment except for part
S.P. Cuttle, A.R. James /Agricultural Water Management 28 (1995) 95-112 99

Table 2
Quantities of lime and fertiliser nutrients applied to the reseeded area in each year of the investigation

Application date Quantity applied

Lime (t ha-‘) N (kg ha-‘) P (kg ha-‘) K (kg ha-‘)

9 April 8.8 _
24 April 29 41 12
25 June 49 11 20
29 August 25 5 10
22 April 65 _
I August _ _ 40 15
30 April _ 65
28 April _ 65 _
12 April 50

of the 1986 season when the reseeded area was stocked with ewes and lambs. Average
stocking rates were 15 mature sheep ha-’ on the reseeded pasture and five sheep ha-’ on
the Molinia area.

2.3. Flow measurements and water sampling

The volume of water draining from each area was measured by separate V-notch weirs
and water level recorders installed on the main outlet ditches (Wl and W2 in Fig. 1).
Recorder traces of water level at the weir were digitised and converted to discharges using
a standard equation (British Standards Institution, 1965). The weirs did not operate satis-
factorily when the water in the weir channels was frozen and discharges in these cases were
estimated from the water balance for the affected period.
Water samples for chemical analysis were collected from the outflow at each weir. Bulk
samples of water were collected on a weekly basis via a sampling tube inserted through the
weir plate 5 mm below the level of the vertex of the V-notch and leading to a collecting
bin. Initially, the how through the sampling tube was controlled by a constriction which
limited the volume collected during the weekly sampling periods. In September 1985 this
arrangement was replaced on both weirs by a flow-proportional sampler providing a com-
posite sample made up of sample increments collected at 30 min intervals (Cuttle and
Mason, 1988). The collecting bins were protected from sunlight but no preservatives were
added to their contents. Samples of rainfall were also obtained from two open-funnel
collectors at the site.

2.4. Analysis of water samples

The pH of drainage water and rain-water samples was measured immediately on return
to the laboratory using a glass combination electrode. Samples were filtered through a glass
100 S.P. Cuttle. A.R. James /Agricultural Water Management 28 (1995) 95-112

microfibre filter ( 1.2 pm particle retention) before storage at 4°C. Concentrations of

nitrate + nitrite-N, ammonium-N and orthophosphate-P were measured calorimetrically
(Murphy and Riley, 1962; Henriksen and Selmer-Olsen, 1970; Crooke and Simpson, 1971) .
Concentrations of nitrite were assumed to be negligible and results of the nitrate + nitrite-
N analyses are reported as nitrate-N. Potassium and calcium concentrations were measured
by llame emission and atomic absorption spectroscopy, respectively. Analyses were nor-
mally completed within 3 days of sample collection. From November 1986, additional
analyses were included to determine ‘total’ contents of nitrogen, phosphorus and potassium
in unfiltered water samples after digestion with a sulphuric acid/hydrogen peroxide mixture
(Allen, 1989).

3. Results

3.1. Vegetation changes and establishment of the reseeded pasture

There was little change in the botanical composition of the vegetation on the unimproved
area during the experiment. Molinia remained the dominant species, representing 57-64%
of the ground-cover at the time of the botanical surveys in the autumn of each year. Although
there was an apparent increase in the proportion of ground occupied by mosses, from 6%
prior to drainage to 19% in the final year of the investigation, this may have been a result
of increased grazing pressure exposing more of the basal vegetation.
The sown species established successfully on the reseeded area. Although the total
proportion of the ground surface occupied by sown species remained relatively constant at
between 83 and 90%, the relative proportion of individual species changed during the study.
Perennial ryegrass was the most abundant initially (5 1% ground-cover) but was replaced
by red fescue as the dominant species in 1986 and 1987 (45 and 36% ground-cover,
respectively). Timothy remained a minor component of the sward throughout the investi-
gation. The proportion of clover increased from 3% initially to 13% in autumn 1987.

3.2. Hydrology of the site and water-flows

Results were analysed on the basis of hydrological years from April to March, referred
to as Years l-4. Annual rainfall, evaporation and drainage from the two areas are summar-
ised in Table 3. Maximum soil moisture deficits in Years l-3 were 70, 21 and 46 mm,
respectively. Moisture deficits could not be calculated for Year 4 because of the absence of
evaporation data.
There was considerable difference in the effectiveness of the drains in lowering water
tables within individual drainage subplots and poor agreement between duplicate treatments
in the reseeded and Molinia blocks. In the worst cases, water tables were little different
from those in adjoining undrained areas. In spite of the variation between subplots, mean
water table levels over the reseeded area as a whole were similar to those for the Molinia
area. In view of this similarity and the apparent independence of water tables levels and
drainage treatments, it was considered valid to regard the overall hydrology of the two
drained areas as being broadly similar. Fig. 2 shows the weekly variation in water table
S.P. Cuttle, A.R. James /Agricultural Water Management 28 (1995) 95-112 101

Table 3
Annual rainfall, evaporation and measured discharge from the Molinia and reseeded areas (April-March) with
discharges expressed as a percentage of the flow predicted from the water balance. Evaporation was not measured
in Year 4

Period Rainfall (mm) Evaporation (mm) Drainage (mm) (and as % of value predicted from the water

Molinia area Reseeded area

Year 1 1364 423 991 (105) 1020 (108)

Year2 1950 407 1635 (106) 1727 (111)
Year3 1788 365 1465 (103) 1574 (110)
Year4 2076 1783 (-) 1768 (-)

level during the experiment expressed as means of the values for all dipwells within the
drainage subplots. Over the full period of the experiment, the mean depth to the water table
in these drained areas was 233 mm, compared with a mean of 178 mm for the adjoining
undrained areas.
Patterns of water flow recorded at the weirs were also similar for the Molinia and reseeded
areas, Weekly discharges are shown in Fig. 2 as means of the values for the two areas.
Hydrographs were characterised by rapid responses to rainfall events, generally within less
than an hour of the onset or cessation of rainfall. These rapid fluctuations were superimposed
on a base flow that continued through all but the driest periods. Even though Year 1 was
abnormally dry, there were only 6 weeks during the summer when no flow occurred.
Measured values of drainage per unit area were slightly higher for the reseeded area than
for the Molinia area in Years 1 to 3 (Table 3). The relatively lower discharge from the
reseeded area in Year 4 was due, at least in part, to a slight leakage of water around the weir

2000 -0
mL 1500 -200
b 1000~ -400 $
5 P
6 0
>\ 500. -600 ,’
Sz E
0 : I-
Year 3 Year 4
Fig. 2. Weekly drain discharge (bars) and level of the water table below the ground surface (dashed line) in
Years l-4 (April-March). Drain discharges are means of values for the Molinia and reseeded areas. Water table
levels are means of values from all dipwells within the drainage subplots.
102 S.P. Cut& A.R. James /Agriculturul Water Management 28 (1995) 95-112

during the summer. Both weirs and recorders performed satisfactorily for about 90% of the
time. Except during freezing weather, there were few occasions when both weirs were non-
operational at the same time and data from one weir could generally be used to substitute
for missing values from the other.
The hydrological integrity of the areas was examined by comparing measured discharges
for Years 1-3 with estimates of flow obtained from the water balance, determined as the
difference between annual rainfall (April-March) and evaporation, assuming that differ-
ences in soil water storage between the start and end of the year were negligible (Table 3).
Measured discharges from the Molinia and reseeded areas in Years l-3 were, respectively,
5 and 9% greater than those predicted from the water balance. These discrepancies are
within the range of errors expected from measurements of rainfall and stream flow and
provide little evidence of significant entry of water from outside the areas or of loss other
than via the weirs. Fluctuations in the nutrient content of drainage water from the Molinia
area were independent of fertiliser applications to the reseeded area and neighbouring areas,
further indicating that the two areas were hydrologically isolated.

3.3. Solute concentrations andpH of drainage water

Concentrations of inorganic solutes and the pH of water samples from the Molinia and
reseeded areas are shown in Fig. 3 < FIGR > 4 < /FIGR > < FIGR > 5 < /FIGR > . Reseed-
ing and the application of lime and fertilisers increased the concentrations of all the nutrients
studied. Increases in nitrate- and ammonium-N concentrations were confined to short periods
following fertiliser applications (Fig. 3). These increases were generally small ( < 3 mg N
1- ’), except in Year 2 when the nitrate concentration in the bulk sample reached a peak of
23.6 mg N I ~ ’ two weeks after the fertiliser had been applied. Although this was consid-
erably in excess of the European Community limit of 11.3 mg N 1-l for drinking water
supplies (European Economic Community, 1980)) subsequent samples contained less than
4 mg N 1- ’ The nitrate peak was accompanied by a corresponding peak concentration of
3.7 mg ammonium-N I- ‘. During the course of the study, water samples from the Molinia
area contained between 0.01 and 1.8 mg nitrate-N l- ’and a similar range of ammonium-N
Concentrations of soluble orthophosphate in water samples from the Molinia area and in
rainfall were frequently below the analytical limit of detection ( < 0.002 mg P 1~ ‘) and at
no time exceeded 0.006 mg P 1-l. Much larger concentrations were measured in samples
from the reseeded area following the applications of phosphorus fertiliser, particularly after
the maintenance dressing of PK fertiliser in Year 2 (Fig. 4). The sample collected in the
week following this application contained 1.3 mg P 1- ‘, indicating that the moderately high
rainfall that week was sufficient to provide rapid transport of dissolved fertiliser to the
drains. Care was taken to avoid spreading fertiliser directly into the open ditches. During
the following month, concentrations declined to less than 0.4 mg P 1-l but remained well
above those from the Molinia area for the remainder of the investigation.
Potassium fertiliser was applied to the reseeded area on the same dates as phosphorus
and patterns of loss were similar to those described above (Fig. 4). The initial applications
in Year 1 increased concentrations slightly compared to the Molinia area but the application
in August of Year 2 had a much greater effect and was followed by a peak concentration of
S.P. Cuttle, A.R. James /Agricultural Water Management 28 (1995) 95-112 103

(i) Ammonium-N


Year 2 Year 3 Year 4

(ii) Nitrate-N

Year 1 Year 2 Year 3 Year 4

Fig. 3. Concentrations of (i) ammonium-N and (ii) nitrate-N determined in non-digested water samples from the
Molinia area (solid line) and reseeded area (dotted line) in Years l-4 (April-March).

26 mg K 1-l. Over the next 3 weeks, concentrations fell to less than 6 mg 1-r and then
declined more gradually until by the end of Year 4 they were approaching those from the
unfertilised Molinia area. Water samples from the Molinia area contained between 0.1 and
2.3 mg K l- ‘.
Liming increased both the calcium content and pH of drainage water from the reseeded
area. Calcium concentrations and pH values increased during the summer following the
lime application and were consistently greater than those from the Molinia area for the
remainder of the investigation (Fig. 5). Water samples collected from September of Year
1 onwards contained between 5.8 and 41.9 mg Ca 1-l and pH varied between 5.7 and 7.8.
104 S.P. Curtle. A.R. James/Agricultural Water Management 28 (1995) 95-112

(i) Phosphorus

.;.. :;.. i

Year 1 Year 2 Year 3 Year 4


E 1 (ii) Potassium

Year 1 Year 2 Year 3 Year 4

Fig. 4. Concentrations of (i) orthophosphate-P and (ii) potassium determined in non-digested water samples
from the Mdinia area (solid line) and reseeded area (dotted line) in Years 1-4 (April-March). Concentrations
of phosphorus in samples from the Mdinia area were all less than 0.01 mg I-’ and do not appear on the graph.

In contrast, water samples from the Moliniu area contained between 0.4 and 5.0 mg Ca 1~ ’
and varied in pH from 3.8 to 6.8. In the case of the unlimed area, pH values greater than 5
occurred more frequently from the middle of Year 2 onwards, after installation of the flow-
proportional samplers, although moderately high pHs also occurred in samples collected
from both areas prior to liming. In the case of samples collected from Year 2 onwards, and
as observed elsewhere (e.g. Jenkins et al., 1991)) there was an inverse relationship between
pH and discharge.
Quantities of nutrients leached from the areas were calculated from solute concentrations
in water samples and the corresponding weekly drainage volumes. Annual losses are sum-
S.P. Cuttle, A.R. James /Agricultural Water Management 28 (1995) 95-112 105

(i) Calcium

Year 1 Year 2 Year 3 Year 4

Year 1 Year 2 Year 3 Year 4

Fig. 5. (i) Concentrations of calcium and (ii) pH of non-digested water samples from the Molinia area (solid
line) and reseeded area (dotted line) in Years l-4 (April-March).

marked in Table 4, together with similarly calculated values for the input of nutrients in
rainfall. As phosphorus concentrations were unavailable for part of Years 3 and 4, the annual
totals for these two years were estimated using the average of the concentrations from before
and after this period in place of the missing values. Greater quantities of phosphorus,
potassium and calcium were leached from the reseeded area than from the unfertilised
Moliltia area. Only in Year 2 was the loss of mineral-N (nitrate + ammonium-N) from the
reseeded area appreciably greater than that from the Moliniu area. Differences in the quan-
tities leached each year largely reflected variations in nutrient concentration rather than
differences in drainage volumes.
106 S. P. Cuttle, A.R. James /Agricultural Water Management 28 (1995) 95-112

Table 4
Quantities of nutrients supplied in rainfall and leached from the Molinia and reseeded areas annually (Apri-
March), as derived from the analysis of non-digested water samples. Phosphorus losses in Years 3 and 4 include
some estimated values

Input in rain and quantity leached (kg ha-’ yr- ‘)

Year 1 Year 2 Year 3 Year 4 Total

Mineral-N (NO, f NH,)

Rainfall 1.9 1.2 1.7 7.8 24.5
Molinia area 3.8 4.1 4.1 3.0 15.0
Reseeded area 3.5 10.0 4.6 2.7 20.7
Rainfall < 0.01 < 0.02 < 0.02 < 0.02 < 0.07
Molinia area < 0.02 < 0.02 ( < 0.03) ( < 0.04) <O.ll
Reseeded area 0.33 3.15 ( 1.22) (0.75) 5.45
Rainfall 0.3 1.4 1.8 2.7 6.2
Molinia area 4.6 5.0 6.5 3.9 20.0
Reseeded arca 8.6 63.6 19.3 12.2 103.7
Rainfall 0.4 2 4 5 I1
Molinia area 14 12 13 14 53
Reseeded area 117 324 258 246 946

3.4. Concentrations of ‘total’ nitrogen, phosphorus and potassium in water samples

Water from both areas, but particularly the reseeded area, was often strongly coloured,
indicating that these samples contained an appreciable content of dissolved organic matter.
Organic-N and -P, if present in the samples, would not be detected by the calorimetric
methods used in the study although the flame spectroscopy determinations of potassium
and calcium would be expected to include some organic components. Analysis of acid
digests of unfiltered samples collected in the later part of the study provided a measure of
‘total’ nutrient content. As well as simple inorganic solutes, these determinations would
include organic species, condensed phosphates and nutrients displaced from suspended
mineral material.
Concentrations of ‘total’ nitrogen, phosphorus and potassium determined after acid diges-
tion are summarised in Table 5. In almost all cases, greater concentrations were measured
in digested than in non-digested samples. However, there was no consistent pattern of
variation between increases in the concentration of one nutrient and those of another,
indicating that the digestion process released nutrients from different sources in each case.
Concentrations of ‘total’ nitrogen in water samples from the reseeded area were appre-
ciably greater than those from the Molinia treatment. This is in contrast to concentrations
of mineral-N which were generally similar for both areas except for brief periods following
application of nitrogen fertiliser. On the basis of ‘total’ contents, the quantities of nitrogen
leached from the Molinia and reseeded areas in Year 4 would be approximately 15 and 30
kg N ha-‘, respectively. The increased concentrations of nitrogen measured in samples
after digestion are assumed to be mainly due to organic-N. The greater ‘total’ nitrogen
S.P. Cuttle, A.R. James/Agricultural Water Management 28 (1995) 95-112 107

Table 5
Mean concentrations of mineral-N, orthophosphate and potassium in filtered, non-digested water samples collected
between November 1986 and March 1988 and ‘total’ contents determined after digestion of unfiltered samples

Nutrient contents of non-digested and digested water samples

Mean (and range) (mg 1-l)

Molinia area Reseeded area

N content
- non-digested 0.23 (0.06-0.73) 0.25 (0.05-1.10)
- digested 0.70 (0.24-3.13) 1.77 (0.41-6.86)
P content
- non-digested < 0.002 ( < 0.002-0.004) 0.056 (0.032-0.108)
- digested 0.020 ( < 0.002~.048) 0.146 (0.032-0.563)
K content
- non-digested 0.26 (0.10-0.60) 0.86 (0.1 l-1.50)
- digested 0.74 (0.05-4.34) 1.49 (0.374.61)

content of water samples from the reseeded area, together with their stronger coloration,
suggests that mineralisation of soil organic matter was increased by the lime and fertiliser
applications. Hornung et al. ( 1986) reported that soil solutions from limed upland pastures
contained greater concentrations of dissolved organic carbon than did samples from unim-
proved moorland.
The largest proportional increase as a result of digestion occurred in the case of phos-
phorus concentrations in samples from the Moliniu area. ‘Total’ concentrations were gen-
erally more than an order of magnitude greater than concentrations measured in non-digested
samples. Soluble organic phosphorus compounds may represent a significant proportion of
the total dissolved phosphorus in soil solution, particularly in the case of organic soils and
those of low nutrient status (Harrison, 1987; Ron Vaz et al., 1993). Digestion may have
also released adsorbed phosphorus from the surface of silt and clay particles present in the
unfiltered samples. The transport of phosphorus sorbed on suspended sediments has been
identified as an important contributor to the loss of phosphorus from catchments (e.g.
Bargh, 1978). In the case of water samples from the reseeded area, relative differences
between concentrations of soluble orthophosphate and ‘total’ phosphorus were smaller but
are of greater significance in terms of absolute concentrations. Estimates of fertiliser loss
based on ‘total’ contents would be approximately double those in Table 4.
In the case of potassium concentrations, the increases brought about by digestion were
similar for samples from both areas and estimates of fertiliser loss based on ‘total’ concen-
trations are therefore little different from those calculated using values for non-digested

4. Discussion

The drained Molinia area represents a situation that is unlikely to occur in agricultural
practice as the cost of draining this type of land would not be considered worthwhile except
108 S.P. Cuttle. A.R. Jumes/Agricultural Water Management 28 (1995) 95-112

as part of a comprehensive pasture improvement scheme involving reseeding and the

application of fertilisers. As drainage would be expected to improve soil aeration and modify
the hydrological pathways by which solutes are transported, leaching data from the Molinia
treatment cannot be considered to be representative of unimproved upland pasture. The
absence of equivalent undrained areas prevented a separate assessment of the effects of
drainage; however, the quantities and concentrations of solutes leached from this area are
similar to those reported from studies on unimproved moorland elsewhere in the UK (Crisp,
1966; Edwards et al., 1985; Hornung et al., 1985; Reynolds et al., 1989)) indicating that in
this instance the effects of drainage were small. This may be a reflection of the limited
disturbance of the site and the ineffectiveness of the drainage treatments over much of the
Quantities of mineral nitrogen and orthophosphate leached from the Molinia area were
similar to the input of these nutrients in rainfall (Table 4). Although this indicates that
nutrient balances had not been disturbed by drainage, estimates of nitrogen and phosphorus
losses based on ‘total’ concentrations are greater and, particularly in the case of phosphorus,
indicate a net depletion of these elements that may be associated with increased mineralis-
ation of soil organic matter. Quantities of potassium and calcium leached from the Molinia
area both exceeded the input in rainfall, irrespective of whether losses were calculated from
solute or ‘total’ concentrations. The Lower Palaeozoic rocks from which the local soils are
derived are relatively rich in potassium but contain little calcium (Adams and Evans, 1990)
and the net depletion of this element is less readily explained. Reynolds et al. ( 1987) also
reported a net depletion of calcium from an undrained, unimproved moorland catchment
on a similar lithology elsewhere in mid-Wales.
The Molinia and reseeded areas both included similar drainage treatments and differences
in the quantities of nutrients leached from the two areas could therefore be attributed to
effects of reseeding and the applications of lime and fertilisers. The greater loss of ‘total’
nitrogen from the reseeded pasture compared with the Molinia area may be a reflection of
the greater input of nitrogen or of increased mineralisation of soil organic matter as a result
of liming and application of other nutrients (Shah et al., 1990). The only significant loss
that could be directly attributed to the application of nitrogen fertiliser occurred in Year 2
of the study when nitrate concentrations increased following the application. In this instance,
a maximum rainfall intensity of 15 mm in 10 h was recorded in the week during which the
nitrate peak occurred. Though relatively modest, this was greater than the maxima recorded
following nitrogen applications in other years and appeared to have been sufficient to
promote rapid transport of dissolved fertiliser to the drains. The loss of nitrogen in the 5
weeks following application was equivalent to 9% of the fertiliser applied and accounted
for almost half of the total quantity of mineral-N leached that year. In addition to the nitrogen
applied as fertiliser, clover in the reseeded pasture would also have supplied nitrogen through
symbiotic fixation of atmospheric nitrogen. Goodman ( 1988) estimated that a similar sward
atthesamesitefixed 32 kgNhaa’ year-‘. Other estimates of fixation for improved pastures
on peaty soils indicate larger inputs of 92-125 kg N ha- ’year- ’ (Newbould, 1982). These
values indicate that in the present investigation, the use of fertilisers and introduction of
clover would have increased the input of nitrogen to the reseeded area by at least 100 kg N
ha-’ year- ‘; however, leaching appeared to account for only a small proportion of this
increased input. On a wet site such as this, with a plentiful supply of organic carbon,
S.P. Cuttle, A.R. James /Agricultural Water Management 28 (1995) 95-112 109

denitrification may be the dominant form of nitrogen loss. Net immobilisation may also
contribute to the apparent loss of nitrogen following application of fertiliser to improved
peat soils (Williams and Wheatley, 1992).
A much greater proportion of the potassium fertiliser applied to the reseeded area was
leached. Estimates derived from solute concentrations in undigested water samples indicate
that 72% of the fertiliser applied in Years 1 and 2 had been leached by the end of the
experiment. Although relatively little of the fertiliser applied in Year 1 was leached, the
maintenance dressing in Year 2 appeared to exceed the capacity of the soil to retain potassium
and subsequent losses were greater. This later application was intended to provide sufficient
potassium to maintain the nutrient status of the soil for 3 years. In the 2.5 years between the
application and the end of the investigation, the increased loss from the reseeded area
accounted for almost all of the potassium supplied in the maintenance dressing. In this time,
potassium concentrations had fallen to values similar to those in drainage from the Moliniu
area. Although these results indicate that the effects of potassium fertiliser are likely to
persist for 2-3 years and broadly confirm the accuracy of the recommendations, they also
demonstrate the poor utilisation of the applied fertiliser: 64% of the potassium applied in
the summer of Year 2 had been leached before the start of the following growing season.
Although the application of fertiliser had a marked effect on concentrations of ortho-
phosphate in drainage water, the quantities involved were relatively small. The greater loss
from the reseeded pasture in Years l-4 compared with the Moliniu area was equivalent to
only 6% of the fertiliser applied. Revised estimates based on concentrations of ‘total’
phosphorus determined in digested water samples in Years 3 and 4 indicate that the total
loss was unlikely to have exceeded 12% of the phosphorus applied. The effects of phosphate
applications are therefore likely to persist for much longer than those of potassium fertiliser.
As little phosphorus is exported from these grazed pastures as livestock or by other means,
repeating the maintenance dressing every 3 years as recommended would be expected to
lead to a considerable net accumulation of phosphorus in the soil and the possibility of an
increased rate of loss.
Broadly similar patterns of fertiliser loss were observed by Roberts et al. ( 1989) following
improvement of an upland pasture on an area of stagnohumic gleys and ironpan stagnopod-
zols at Plynlimon, 11 km north of Pwllpeiran. The pasture was reseeded using a minimal
cultivation technique similar to that in the present study but no artificial drains were installed.
As at Pwllpeiran, nitrate concentrations in drainage remained small with only occasional
peaks greater than 1 mg N l- ‘. Concentrations of orthophosphate increased following
reseeding and reached a peak value of 2 mg P l- ’4 months after fertiliser had been applied.
The loss of phosphorus in the establishment year was estimated as 5 kg P ha-‘, similar to
that measured in the present investigation. The same authors reported contrasting results
from a second site at Plynlimon. In this case, pasture improvement on a drained peat soil
had little effect on phosphorus and potassium concentrations in drainage but increased
nitrate concentrations considerably with annual leaching losses of 14-61 kg N ha-‘. The
greater loss of nitrogen appeared to result from increased mineralisation of soil organic
matter following drainage and cultivation rather than from fertiliser applications. The use
of disc harrows to cultivate the site would have disturbed the soil far more than the minimal
cultivation technique used at Pwllpeiran. Losses of phosphorus and potassium may have
been less than at Pwllpeiran because the fertilisers were worked into the soil and much of
110 S.P. Cuttle. A.R. James/Agriculturul Water Management 28 (1995) 95-112

the phosphorus was applied as relatively insoluble basic slag. It should also be noted that
losses of phosphorus and potassium were small at both sites in the reseeding year: at
Pwllpeiran, it was only after the additional maintenance dressing in Year 2 of the study that
appreciable losses occurred.
The pH and calcium content of water samples from the reseeded area at Pwllpeiran
increased following liming and showed little tendency to decline during the course of the
experiment. Soil pH was increased by liming and also remained relatively constant during
this period. The pH of the O-50 mm soil depth, which increased from 3.5 prior to liming to
5.6 at the start of Year 2, remained at 5.5 when measurements were repeated in Year 4. As
observed in other studies on peat soils (Dampney, 1985)) effects of liming were confined
to a relatively shallow depth. Soil pH between 50-100 mm depth increased from 3.8 in
Year 2 to 4.5 in Year 4 but was little changed below this.
Differences between the quantities of calcium leached from the two areas indicated that
24% of the lime application had been leached by the end of Year 4. This corresponds to an
annual loss of about 0.5 t lime ha-‘. At this rate, the lime applied at reseeding would be
fully leached in about 17 years. However, it is more likely that the leaching rate would
decline as the reservoir of calcium in the soil was depleted with the result that the effects of
liming would persist for longer but at a lower intensity. In a study of upland pastures at
Plynlimon reported by Hornung et al. ( 1986), concentrations of calcium in soil water
samples from improved pasture were greater than those from unimproved moorland, even
though lime had last been applied 40 years earlier. Adams and Evans (1989) modelled
rates of calcium depletion using data from a study of soil properties and drainage water
chemistry from an upland pasture in mid-Wales and similarly concluded that the effects of
liming would persist for 3040 years.
Comparative values for rates of calcium depletion in the initial 2-3 years after liming are
available from an investigation in which different liming strategies for ameliorating surface
water acidification were applied to stream catchments draining into the Llyn Brianne res-
ervoir in mid-Wales (Jenkins et al., 1991). Blanket liming of a moorland catchment at a
rate of 9 t lime ha- ’, but without other pasture improvement measures, resulted in an annual
loss equivalent to about 1% of the lime applied. Lime applied at 16 t ha-’ to the 5 ha
hydrological source area within a 54 ha catchment resulted in an annual loss of about 12%
of the total applied. This latter treatment equates most closely to the reseeded area at
Pwllpeiran, both hydrologically and in terms of the rate of lime depletion, which at
Pwllpeiran was equivalent to 6% per year following an application of 8.8 t lime ha-‘.
Agricultural improvement schemes in the uplands often involve only part of the total
catchment area and the overall impact on water quality will be less than that indicated by
studies of nutrient leaching from relatively small areas of uniformly improved land. Roberts
et al. ( 1989) stressed the importance of extrapolating the results from isolated areas to the
catchment scale and also the difficulties associated with this process. In the simplest case,
all areas of a catchment may be assumed to contribute equally to streamflow so that leaching
data from improved areas may be extrapolated to the larger scale on the basis of the
proportion of the total catchment that has been subjected to improvement. However, the
hydrology of upland catchments is frequently complex and non-uniform. For example, in
the Llyn Brianne study, a further treatment involving standard agricultural improvement of
S.P. Cuttle, A.R. James/Agricultural Water Management 28 (1995) 95-112 111

23% of the total catchment area had no effect on stream water chemistry, possibly because
this part of the catchment did not contribute to streamflow generation (Jenkins et al., 199 1).

5. Conclusions

Drainage alone, in the absence of other pasture improvement measures, appeared to have
little effect on nutrient losses. Pasture improvement had little effect on concentrations of
nitrate and ammonium-N in drainage water samples except for brief periods following the
application of nitrogen fertiliser. Increases in mineral-N concentrations at these times were
usually small and were only of significance in one year of the study. However, concentrations
of organic-N in water from the reseeded area were consistently greater than those from the
Moliniu area.
Concentrations of phosphorus, potassium and calcium and the pH of drainage water
increased following pasture improvement and were consistently greater than those in water
from the Moliniu area.
Almost all of the potassium fertiliser applied to the reseeded area in Years 1 and 2 had
been leached by the end of Year 4 indicating that effects of potassium fertiliser applications
on drainage water chemistry are unlikely to persist for more than about 3 years. In contrast,
about 12% of the phosphorus fertiliser and 24% of the lime had been leached by the end of
the experiment and effects of liming and phosphorus fertilisation would be expected to
continue well beyond the 4-year study period.
Much of the phosphorus in non-digested water samples from this peaty soil, and a
particularly high proportion in the case of the unimproved area, was present in forms that
were not detected by the molybdate calorimetric method used for the analysis of these


The study was conducted as part of work commissioned by the Ministry of Agriculture,
Fisheries and Food. We are grateful to Pwllpeiran Experimental Husbandry Farm for
provision of the site and to B.M.S. Davies, G. Daniel and D. Jones for technical assistance
during the experiment.


Allen, SE. (Editor), 1989. Chemical Analysis of Ecological Materials. 2nd edn. Blackwell Scientific, Oxford,
368 pp.
Adams, W.A. and Evans, G.M., 1989. Effects of lime applications to parts of an upland catchment on soil properties
and the chemistry of drainage waters. J. Soil Sci., 40: 585-597.
Adams, W.A. and Evans, G.M., 1990. Input/export relationships of major ions in west Wales catchments. Agric.
Ecosyst. Environ., 32: 13-24.
Bargh, B.J., 1978. Output of water, suspended sediment, and phosphorus and nitrogen forms from a small
agricultural catchment. N. 2. J. Agric. Res., 21: 29-38.
112 S.P. C&e, A.R. James /Agricultural Water Management 28 (1995) 95-112

British Standards Institution, 1965. Methods of measurement of liquid flow in open channels. B.S.3680, Part 4A.
Her Majesty’s Stationery Office, London, 91 pp.
Crisp, D.T., 1966 Input and output of minerals for an area of Pennine moorland: the importance of precipitation,
drainage, peat erosion and animals. J. Appl. Ecol., 3: 327-348.
Crooke, W.M. and Simpson, W.E., 197 1. Determination of ammonium in Kjeldahl digests of crops by an automated
procedure. J. Sci. Food Agric., 22: 9-10.
Cuttle, S.P. and Mason, D.J., 1988. A flow-proportional water sampler for use in conjunction with a V-notch weir
in small catchment studies. Agric. Water Manage., 13: 93-99.
Dampney, P.M.R., 1985. A trial to determine the lime requirement for reseeded grassland on a peaty hill soil. Soil
Use Manage., 1: 95-100.
Edwards, A.C.. Creasey, .I. and Cresser, M.S., 1985. Factors influencing nitrogen inputs and outputs in two Scottish
upland catchments. Soil Use Manage., 1: 83-87.
European Economic Community, 1980. Council directive on the quality of water for human consumption. Official
Journal No. 80/778, EEC L229.
Goodman, P.J., 1988. Nitrogen fixation, transfer and turnover in upland and lowland grass-clover swards using
15N isotope dilution. Plant Soil, 112: 247-254.
Harrison, A.F., 1987. Soil Organic Phosphorus. A Review of World Literature. C.A.B. International, Wallingford,
UK, 257 pp.
Henriksen, A. and Selmer-Olsen, A.R., 1970. Automatic methods of determining nitrate and nitrite in water and
soil extracts. Analyst, 95: 5 14-5 18.
Homung, M., Reynolds, B. and Hatton, A.A., 1985. Land management, geological and soil effects on streamwater
chemistry in upland mid-Wales. Appl. Geog., 5: 71-80.
Hornung, M., Stevens, P.A. and Reynolds, B., 1986. The impact of pasture improvement on the soil solution
chemistry of some stagnopodzols in mid-Wales. Soil Use Manage., 2: 18-26.
Jenkins, A., Waters, D. and Donald, A., 1991. An assessment of terrestrial liming strategies in upland Wales, J.
Hydrol., 124: 243-261.
Ministry of Agriculture, Fisheries and Food, 1973. Fertilizer Recommendations. MAFF Bull. 209, 106 pp.
Ministry of Agriculture, Fisheries and Food, 198 1. Systems for Welsh mountain sheep. Sheep Husbandry 3. MAFF
Booklet 2323, 26 pp.
Monteith, J.L., 1965. Evaporation and environment. In: G.E. Fogg (Editor), The State and Movement of Water
in Living Organisms. 19th Symp. Sot. Exp. Biol., Academic Press, New York, pp. 205-234.
Munro, J.M.M. and Davies, D.A., 1973. Potential pasture production in the uplands of Wales. 2. Climatic
limitations on production. J. Br. Grassld. Sot., 28: 161-169.
Munro, J.M.M., Davies, D.A. and Thomas, T.A., 1973. Potential pasture production in the uplands of Wales. 3.
Soil nutrient resources and limitations. J. Br. Grassl. Sot., 28: 247-255.
Murphy, J. and Riley, J.P., 1962. A modified single solution method for the determination of phosphate in natural
waters. Anal. Chim. Acta, 27: 31-36.
Newbould, P., 1982. Biological nitrogen fixation in upland and marginal areas of the U.K. Phil. Trans. R. Sot.
Lond., B 296: 4054 17.
Reynolds, B., Homung, M. and Hughes, S., 1989. Chemistry of streams draining grassland and forest catchments
at Plynlimon, mid-Wales. Hydrol. Sci. J., 34: 667-686.
Reynolds, B., Homung, M. and Stevens, P.A., 1987. Solute budgets and denudation rate estimates for a mid-
Wales catchment. Catena, 14: 13-23.
Roberts, A.M., Hudson, J.A. and Roberts, G., 1989. A comparison of nutrient losses following grassland improve-
ment using two different techniques in an upland area of mid-Wales. Soil Use Manage., 5: 174-179.
Ron Vaz, M.D., Edwards, A.C., Shand, CA. and Cresser, M.A., 1993. Phosphorus fractions in soil solution:
influence of soil acidity and fertilizer additions. Plant Soil, 148: 175-183.
Rudeforth, CC., Hartnup, R., Lea, J.W., Thompson, T.R.E. and Wright, P.S., 1984. Soils and their Use in Wales.
Bull. 11, Soil Survey of England and Wales, Harpenden, 336 pp.
Shah, Z., Adams, W.A. and Haven, C.D.V., 1990. Composition and activity of the microbial population in an
acidic upland soil and effects of liming. Soil Biol. Biochem., 22: 257-263.
Smith, L.P., 1976. The Agricultural Climate of England and Wales. Tech. Bull. 35, Ministry of Agriculture,
Fisheries and Food. Her Majesty’s Stationery Office, London, 147 pp.
Williams, B.L. and Wheatley, R.E., 1992. Mineral nitrogen dynamics in poorly drained blanket peat. Biol. Fe&l.
Soils. 13: 96-101.