Http Www.sciencedirect.com Science Ob=MImg& Imagekey=B6T3Y-4XT2CHD-2-9& Cdi=4959& User=126036&...

8/7/2019 Http Www.sciencedirect.com Science Ob=MImg& Imagekey=B6T3Y-4XT2CHD-2-9& Cdi=4959& User=126036& Pii=…

http://slidepdf.com/reader/full/http-wwwsciencedirectcom-science-obmimg-imagekeyb6t3y-4xt2chd-2-9-cdi4959 1/15

Streamwater phosphorus and nitrogen across a gradient in rural–agricultural

land use intensity

H.P. Jarvie a,*, P.J.A. Withers b, M.J. Bowes a, E.J. Palmer-Felgate a, D.M. Harper c, K. Wasiak c, P. Wasiak c,R.A. Hodgkinson b, A. Bates b, C. Stoate d, M. Neal a, H.D. Wickham a, S.A. Harman a, L.K. Armstrong a

a Centre for Ecology and Hydrology, Wallingford, Oxfordshire OX10 8BB, UK b ADAS Environment Group, ADAS Gleadthorpe, Meden Vale, Mansfield, Notts NG20 9PF, UK cDepartment of Biology, University of Leicester, UK d The Game & Wildlife Trust, Allerton Project, Loddington, Leics LE7 9XE, UK

1. Introduction

River eutrophication resulting fromphosphorus (P) and nitrogen

(N) enrichment is an issue of global significance, leading to

deterioration of water quality for potable supply and amenity, as

well as reduced aquatic biodiversity (Carpenter et al., 1998a; Cheng

andChi,2003; Smith,2003; Neal andJarvie, 2005). Across the world,

intensive agricultural practices, such as cereal production, forage

maize, potatoes, intensive dairy and outdoor pigs are generally

regarded as ‘high risk’ for P and N loss to rivers, because they are

either regularly over-fertilised, recycle large amounts of manure, or

are highly vulnerable to soil erosion (Carpenter et al., 1998b;

Chambers et al., 2000; Wilcock et al., 2006). These effects may be

exacerbated where ‘high-risk’ agricultural practices are located in

close proximity to watercourses, on steep slopes or on under-

drained land, which increase hydrological connectivity, resulting in

greater efficiency of delivery of P and N to surface waters (Gburek

and Sharpley, 1998; Sharpley et al., 2008). Increasingly, other rural

sources, such as sewage treatment works (STW), septic tank

overflows and runoff from impervious surfaces (roads and farm-

Agriculture, Ecosystems and Environment 135 (2010) 238–252

A R T I C L E I N F O

Article history:

Received 27 February 2009

Received in revised form 27 September 2009

Accepted 2 October 2009

Available online 27 November 2009

Keywords:

Rural population

Nitrogen

Phosphorus

Agriculture

Stream

River

Septic tank

SewageLoad apportionment model

A B S T R A C T

This paper provides an overview of theimpacts of rural land useon lowland streamwater phosphorus (P)

and nitrogen (N) concentrations and P loads and sources in lowland streams. Based on weekly water

quality monitoring, the impacts of agriculture on streamwater P and N hydrochemistry were examined

along a gradient of rural–agricultural land use, by monitoring three sets of ‘paired’ (near-adjacent) rural

headwater streams, draining catchments which are representative of the major geology, soil types and

rural/agricultural land use types of large areas of lowland Britain. The magnitude and timing of P and N

inputs were assessed and the load apportionment model (LAM) was applied to quantify ‘continuous’

(point) source and ‘flow-dependent’ (diffuse) source contributions of P to these headwater streams. The

results show that intensive arable farming had only a comparatively small impact on streamwater total

phosphorus (TP loads), with highly consistent stream diffuse-source TP yields of ca. 0.5 kg-P haÀ1 yearÀ1

for the predominantly arable catchments with both clay and loam soils, compared with 0.4 kg-

P haÀ1 yearÀ1 for low agricultural intensity grassland/woodland on similar soil types. In contrast,

intensive livestock farming on heavy clay soils resulted in dramatically higher stream diffuse-source TP

yields of 2 kg-P haÀ1

yearÀ1

. The streamwater hydrochemistry of the livestock-dominated catchmentwas characterised by high concentrations of organic P, C and N fractions, associated with manure and

slurry sources. Across the study sites, the impacts of human settlement were clearly identifiable with

effluent inputs from septic tanks and sewage treatmentworks resulting in large-scale increasesin soluble

reactive phosphorus (SRP) loads and concentrations. At sites heavily impacted by rural settlements, SRP

concentrations under baseflow conditions reached several hundred mg-P L À1. Load apportionment

modelling demonstrated significant ‘point-source’ P inputs to the streams even where there were no

sewage treatment works within the upstream catchment. This indicates that, even in sparsely populated

rural headwater catchments, small settlements andeven isolated groupsof housesare sufficientto cause

significant nutrient pollution and that septic tank systems serving these rural communities are actually

operating as multiple point sources, rather than a diffuse input.

ß 2009 Elsevier B.V. All rights reserved.

* Corresponding author at: Centre for Ecology and Hydrology, Maclean Building,

Crowmarsh Gifford, Wallingford, Oxfordshire OX10 8BB, UK.

Tel.: +44 01491692260; fax: +44 01491692424.

E-mail address: [email protected] (H.P. Jarvie).

Contents lists available at ScienceDirect

Agriculture, Ecosystems and Environment

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / a g e e

0167-8809/$ – see front matter ß 2009 Elsevier B.V. All rights reserved.

doi:10.1016/j.agee.2009.10.002

mailto:[email protected]

http://www.sciencedirect.com/science/journal/01678809

http://dx.doi.org/10.1016/j.agee.2009.10.002

http://dx.doi.org/10.1016/j.agee.2009.10.002

http://www.sciencedirect.com/science/journal/01678809

mailto:[email protected]



yards) have also been identified as significant sources of P and N to

streams draining rural watersheds (Edwards and Withers, 2008;

Withers et al., 2009). The synchronicity between P and N delivery

and periods of biological nutrient demand is of key importance for

eutrophication impacts of P and N sources, which contribute at

different times of year and under varying flow conditions ( Jarvie et

al., 2006; Edwards and Withers, 2007). Appropriate understanding

of the relative contributions and timing of P and N inputs to rivers

and streams is therefore of central importance for targeting

mitigation options most effectively.

In this study, the impacts of agriculture on streamwater P

and N hydrochemistry were examined along a gradient of rural–

agricultural land use intensities, by means of three sets of

‘paired’ rural headwater streams. This formed part of a wider

study, PARIS (Phosphorus from Agriculture: Riverine Impacts

Study; Palmer-Felgate et al., 2009; Withers et al., 2009) and a

companion paper examines streambed sediments and the

impact of agricultural land use and other geochemical controls

on bed-sediment P concentrations (Palmer-Felgate et al., 2009).

In this study, we focus on the streamwater component: we

examine the magnitude and timing of P and N inputs and

quantify the ‘point’ and ‘diffuse’ source inputs of P to these

headwater streams, using the load apportionment modelling

approach (Bowes et al., 2008, 2009). Load apportionmentmodelling was undertaken for P, which is often the key limiting

nutrient in flowing freshwaters and therefore the main target for

remediation to reduce eutrophication in rivers and streams

(Reynolds and Davies, 2001; Bowes et al., 2007). A full

description of the ‘paired catchment’ approach is provided in

Palmer-Felgate et al. (2009): in essence, each set of ‘paired’

streams included both a stream draining low agricultural

intensity rural land use and one or more streams draining

higher intensity agricultural land use. ‘Paired’ streams were

chosen with similar baseline catchment characteristics, includ-

ing catchment area, soils, underlying geology and rainfall

patterns. Headwater streams were chosen because they provide

a clearer hydrochemical ‘fingerprint’ of the land use and

background geology, whereas the water quality signals in larger

rivers are integrated from multiple land use and effluent

discharges, which tends to obscure the impacts from individual

agricultural land use types.

2. Study sites

The three sets of paired catchments were located in three

lowland river basins: the Herefordshire Wye, Hampshire Avon and

Leicestershire Welland (Palmer-Felgate et al., 2009). These three

river basins were chosen because they are representative of the

major geology, soil types and agricultural and rural land use of

large areas of lowland Britain. The paired lowagricultural intensity

and high agricultural intensity streams within each of the

catchments were as follows:

Table 1

Summary of catchment characteristics for each of the streamwater monitoring sites in the Wye, Welland and Avon river basins; for more detailed information, see Palmer-

Felgate et al. (2009). L denotes low agricultural intensity catchments; H denotes high agricultural intensity catchments.

Basin Catchment Area (km2) General soils description Dominant land use Population and point sources

Wye Whitchurch (L) 6.4 Sandy silt loam and silty

clay loam on variably

sloping terrain; soils are

highly dispersive and erosive

Low intensity beef and

sheep farming and grass

and ley-arable crops on

the perimeter plateau land

No major wastewater discharges;

ca. 90 residents (24% of the rural

population of 375) rely on septic tanks

Dinedor (H) 8.7 Mixed beef/sheep and

cereal/potato farms

No major wastewater discharges; the

rural population of 290 residents relies

on septic tanks, including a village hall

Kivernoll (H) 9.9 Intensive arable cultivation

(winter cereal, oilseed rape,

sugar beet, potato); also

poultry farming

Small village sewage treatment works

(discharges directly to the stream).

190 residents (ca. 27% of the population

of 709) rely on septic tanks

Wellan d Digby Fa rm (L) 0.44 Chalky bou lder clay soils;

seasonally waterlogged

clayey and fine loamy over

clayey soils; most fields

have under-drainage and

ditches are common

Permanent pasture (beef,

sheep, silage production)


all 4 residents (1 farm) are served by

a septic tank which is not in direct

connectivity with stream

Belton Bridge (H) 1.5 Intensive arable production

(cereals, oilseed rape and beans)

and mixed agriculture

(ley grassland grazed by sheep,

spring cereals, stubble turnips)

No major wastewater discharges; ca. 15

residents within 6 houses are served by

septic tanks. Septic tank effluent

discharges to a network of ditches which

ultimately drain into the stream

Lone Pine (H) 1.2 Sheep and spring cereals and

stubble turnips

Sampling site is dominated by direct

discharge of effluent from septic tanks

in Loddington Village which serve a

residential population of ca. 30.

Avon Cools Cottage (L) 1.6 Seasonally waterlogged

fine loamy over clay soil;

under-drained to varying

levels of efficiency

Largely woodland and permanent

pasture; grazed by beef cattle

and calves in summer


10 residents, all on septic tanks

Priors Farm (H) 4.7 Grassland for intensive dairy

production; more recent reversion

to beef farming and forage maize.

Stream receives direct runoff from

farmyards, cattle crossings and

buildings housing beef cattle during

winter are located close to the stream

Resident catchment population of 515, of

which 115 (22%) are served by septic tanks;

the remainder of the population are on mains

sewerage to a sewage treatment works in the

northern part of the catchment. However, the

treatment works discharges to a constructed

wetland and not directly into the stream

H.P. Jarvie et al. / Agriculture, Ecosystems and Environment 135 (2010) 238–252 239



Wye: Whitchurch (low agricultural intensity); Dinedor and

Kivernoll (high agricultural intensity).

Welland: Digby Farm (low agricultural intensity); Belton Bridge

and Lone Pine (high agricultural intensity).

Avon: Cools Cottage (low agricultural intensity); Priors Farm

(high agricultural intensity).

Details of the paired catchments are provided in Palmer-Felgate et al.

(2009), including soils, dominant farming systems, aerial land use

percentages, P inputs (as fertiliser and manure) and Olsen P

concentrations. A summary table outlining the major characteristics

of thepaired‘lowagricultural intensity’ and‘highagricultural intensity’

streams for each of the three catchments is provided in Table 1 and a

summary of the rainfall data forthe water quality monitoring period is

providedin Table 2. Therainfall data were obtainedfrommonthly1 km

rainfall grids which were derived from daily and monthly rain gauge

data using Voronoi interpolation (British Standards Institution, 1996).

The rainfall data show that the period coveringthe monitoring for this

study (October 2004–October2007) was characterised by a widerange

of hydrological conditions, as shown by the monthly rainfall totals for

the Wye, Avon and Welland (compared with average long-term data

from 1961 to 2007). A period of protracted drought began in the

autumn of 2004, which continued until the autumn of 2006. This was

followed by above average rainfall during the winter/spring of 2006–2007 and then extremely heavy rainfall leading to summer flooding in

summer 2007.

In addition to the contrast between high and low agricultural

intensity, four of the monitored streams which drain predomi-

nantly arable land (defined here as having >45% arable land

coverage, Palmer-Felgate et al., 2009) show a gradation of

increasing levels of human settlement (Table 1):

Belton Bridge in the Welland had minimal human settlement;

the six dwellings were served by septic tanks which did not

discharge directly to the stream, but via a network of ditches.

Dinedor in the Wye had no sewage works, but a relatively large

rural population served by septic tanks, including a village hall.

Lone Pine in theWelland received direct discharges of septictankeffluent immediately upstream of the sampling point.

Kivernoll had a larger human population with effluent inputs

from a village STW and septic tanks.

This gradation provided an opportunity to examine the relative

varying levels of human influence on streamwater P and N

concentrations and loads in the predominantly arable catchments.

This is an important consideration since, in many cases, it is difficult

to isolate the impacts of farming alone, because in most UK rural

catchments, intensive arable areas are closely interspersed with

villages and market towns (Neal and Jarvie, 2005).

3. Methods

3.1. Field sampling and flow measurement

At each site, weekly streamwater quality samples were

collected on the same day each week. Streams in the Avon were

sampled between October 2004 and October 2007; in the Wye and

Welland, streams were sampled between November 2004 and

October 2006. The samples were filtered in the field (using

0.45mm cellulose nitrate filter membranes) and sent in coolboxes

by overnight courier to the CEH Wallingford laboratories for

chemical analysis. The samples were refrigerated and the analysis

forP fractions was undertaken within ca. 24 h of sampling, to avoid

significant sample deterioration on storage, according to estab-

lished protocols ( Jarvie et al., 2002).

Flows were measured on the Avon and Wye streams using

Starflow 6526B sensors, which areincoherent Doppler devices thatT

a b l e

2

M o n t h l y r a i n f a l l t o t a l s f o r t h e W y e , W e l l a n d a n d A

v o n , d u r i n g t h e f u l l m o n i t o r i n g p e r i o d , w

i t h a n n u

a l r a i n f a l l t o t a l s ( J a n u a r y 2 0 0 4 – D e c e m b e r 2 0 0 7 ) ,

l o n g - t e r m ( 1 9 6 1 – 2 0 0 7 ) m o n t h l y a n d a n n u a l r a i n f a l l m e a n s , r a n g e s a n d s t a n d a r d

d e v i a t i o n s .

B a s i n

Y e a r

J a n u a r y

F e b r u a r y

M a r c h

A p r i l

M a y

J u n e

J u l y

A u g u s t

S e p t e m b e r

O c t o b e r

N o v e m b e r

D e

c e m b e r

A n n u a l

W y e

2 0 0 4

9 9

4 3

4 6

7 9

5 0

4 1

3 8

7 3

5 7

1 3 5

3 3

4

5

7 3 9

2 0 0 5

4 0

3 2

7 0

5 5

3 4

3 9

8 0

2 5

4 3

1 2 6

7 3

6

5

6 8 2

2 0 0 6

2 4

3 6

8 3

2 0

1 1 4

2 3

2 5

6 0

5 5

9 6

1 1 3

1 1

9

7 6 8

2 0 0 7

8 4

1 0 1

7 9

9

1 0 7

1 4 2

1 6 2

2 9

3 2

3 8

5 5

8

3

9 2 1

L o n g - t e r m

m e a n ( r a n g e )

8 0 ( 1 2 – 1 9 0 )

6 0 ( 4 – 1 5 8 )

5 9 ( 1 – 1 2 7 ) 5 4 ( 4 – 1 5 4 )

5 9 ( 9 – 1 5

0 )

5 2 ( 1 0 – 1 5 1 )

4 6 ( 7 – 1 6 2 )

5 6 ( 5 – 1 4 9 )

6 4 ( 9 – 2 1 1 )

7 7 ( 8 – 1 6 8 )

7 1 ( 2 6 – 1 6 2 ) 8

2 ( 1 4 – 1 7 9 ) 7 5 7 ( 4 9 1 – 9 9 9 )

S t a n d a r d d e v i a t i o n

4 2

3 8

2 8

3 3

3 4

3 4

3 1

3 2

4 3

4 2

3 5

4

2

1 0 6

W e l l a n d 2 0 0 4

8 2

3 1

2 7

8 7

5 7

4 8

1 0 2

1 6 6

3 8

1 2 7

3 9

2

1

8 2 5

2 0 0 5

4 1

3 9

2 5

4 5

4 2

4 9

6 5

7 3

6 9

6 3

3 7

4

4

5 9 2

2 0 0 6

1 9

3 6

4 9

5 5

8 9

1 5

7 4

8 0

6 6

8 6

8 3

6

2

7 1 4

2 0 0 7

6 1

6 3

5 5

6

1 0 6

1 4 6

1 2 5

4 6

4 1

4 3

6 1

5

3

8 0 6

L o n g - t e r m


5 7 ( 1 3 – 1 0 8 )

4 7 ( 1 2 – 1 5 1 ) 5 0 ( 5 – 1 0 1 ) 5 6 ( 6 – 1 2 5 )

5 6 ( 1 0 – 1

3 0 )

6 0 ( 6 – 1 5 2 )

6 1 ( 1 4 – 1 2 9 )

6 7 ( 6 – 1 6 6 )

6 0 ( 1 3 – 1 2 9 )

6 3 ( 9 – 1 2 7 )

6 3 ( 2 9 – 1 4 7 ) 6

3 ( 1 1 – 1 4 5 ) 7 0 1 ( 4 6 8 – 8 6 4 )


2 3

2 7

2 6

3 0

2 9

4 0

3 1

3 6

3 3

3 5

2 5

3

0

1 0 4

A v o n

2 0 0 4

1 1 2

4 8

5 0

9 0

4 4

5 1

6 6

1 0 8

4 5

1 7 4

3 7

7

4

8 9 9

2 0 0 5

5 2

2 9

6 2

7 2

4 2

5 6

6 6

6 2

3 4

1 1 2

7 6

1 0

5

7 6 8

2 0 0 6

2 6

6 6

7 0

3 3

1 1 1

2 6

3 1

5 8

8 2

1 2 4

1 5 1

1 1

5

8 9 3

2 0 0 7

9 8

1 1 1

6 0

8

1 2 2

1 2 6

1 1 9

4 7

3 1

4 6

9 9

9

4

9 6 1

L o n g - t e r m


9 1 ( 1 5 – 1 8 5 )

7 1 ( 7 – 1 9 8 )

7 1 ( 6 – 1 5 2 ) 6 2 ( 2 – 1 8 6 )

6 3 ( 1 0 – 1

5 8 )

6 0 ( 6 – 1 4 9 )

5 4 ( 1 5 – 1 1 9 )

7 0 ( 1 3 – 1 5 2 ) 7 4 ( 9 – 2 0 5 )

9 2 ( 1 0 – 1 7 9 )

9 6 ( 3 5 – 2 5 3 ) 1 0

2 ( 2 2 – 1 9 0 ) 9 0 5 ( 6 5 4 – 1 2 1 9 )


4 5

4 3

3 3

3 5

3 5

3 6

2 7

3 7

4 7

5 0

4 9

4

6

1 3 0

H.P. Jarvie et al. / Agriculture, Ecosystems and Environment 135 (2010) 238–252240



use an ultrasonic beam to measure mean water velocity, combined

with a pressure transducer to measure water depth. Discharge was

then calculated from the product of channel cross section and

velocity. Flow data were recorded at 15 min intervals and the

water chemistry data were matched with instantaneous flow

readings closest to the time of water quality sampling. For the

Welland streams, flows were estimated using Environment Agency

flow data from a gauging station on a nearby first order stream

(GS31025; the River Gwash South Arm at Manton), and scaled

according to catchment area.

3.2. Chemical analysis

For phosphorus analysis, three operationally defined measure-

ments were made:

Soluble reactive P (SRP): a measure of the largely inorganic

monomeric and easily hydrolysable P in the less than 0.45mm

fraction.

Total dissolved P (TDP): the combination of SRP and dissolved

hydrolysable P (DHP). DHP is the polymeric/organic P in the

less than 0.45 mm fraction, released by potassium peroxodi-

sulphate digestion on a filtered sample (Eisenreich et al.,

1975). Total phosphorus (TP) is the combination of TDP and particulate

P (PP). Particulate P is the fraction released from the sediment by

potassium peroxodisulphate digestion on an unfilteredsample. It

probably represents the less refractory part of the total

suspended sediment—the more labile organic and inorganic

part of the solid phase (e.g. plant debris andacid soluble minerals

such as the less crystalline iron oxides/hydroxides and calcium

carbonate that sorb and co-precipitate P).

The SRP fraction was determined colorimetrically, by the

method of Murphy and Riley (1962), as modified by Neal et al.

(2000), and the TDP and TP fractions by the method of Eisenreich

et al. (1975). Major ion concentrations (including nitrate,NO3À-N

and nitrite, NO2À

-N) were measured using ion chromatography.Ammonium, NH4

+-N concentrations were determined color-

imetrically using an indophenol blue method (Leeks et al.,

1997). Total inorganic nitrogen (TIN) was calculated as the

concentrations of nitrate as N (NO3À-N), nitrite as N (NO2

À-N) and

ammonium as N (NH4+-N). Total dissolved nitrogen (TDN) was

measuredby an automated thermaloxidation process,by whichN

in the sample is converted to its oxide (NO) and measured using a

chemoluminescent detector. Dissolved organic nitrogen was

calculated as the difference between TDN and DIN concentrations

(DON = TDN À DIN). Dissolved organic carbon (DOC) was mea-

sured by automated thermal oxidation to CO2 and nondispersive

infrared gas analysis. Suspended sediment (SS) was determined

gravimetrically (filtration and drying at 105 8C). Suspended

sediment P concentration (SS-P) was calculated as PP/SS (mg-P mg-SSÀ1). Fe concentrations weremeasuredon samples filtered

through 0.45mm membranes using an Inductively Coupled

Plasma Optical Emission Spectrometer (ICP-OES) to provide

supplementary information on P-cycling linked to redox-related

dissolution of Fe minerals.

3.3. Load apportionment modelling

Concentrations of P as TP, SRP, DHP and PP were modelled

individually as a function of river flow, using the load apportion-

ment model, which is described in detail by Bowes et al. (2008,

2009) and a brief summary of the modelling approach is provided

here. The loads of P from ‘continuous’ or ‘flow independent’ inputs

(typically ‘point’ sources) (F p) and ‘flow-dependent’ inputs

(typically ‘diffuse’sources) (F d) were modelled as a power-law

function of river flow (Q ; m3 sÀ1):

F p ¼ AÂQ B and F d ¼ C Â Q D (1)

The total load (F t; mg-P sÀ1) was then calculated as a linear

combination of the loads from continuous and flow-dependent

sources:

F t¼ F

pþ F

d¼ AÂQ B þ C ÂQ D (2)

where A, B, C and D are parameters determined empirically. The P

concentration at any given sampling point (C p; mg mÀ3) is equal to

the load divided by the volumetric flow rate, expressed as:

C p ¼ AÂQ BÀ1þ C ÂQ DÀ1 (3)

Eq. (3) was fitted to the data using non-linear least squares

regression in the Solver function in Microsoft EXCEL ß. Eq. (3) was

evaluated by varying the four fitting parameters against P

concentrations and flow from the monitoring programme. The B

term assumes that continuous (point) source derived P is not

chemically conservative within the stream. Where B > 0, this is

indicative that point-source derived P undergoes net in-stream

retention during low flows, as a result of uptake by sediments or

biota and/or deposition of particulate P. In contrast B < 0 isindicative of net in-streamrelease of SRP, which could occur due to

development of anoxia in the surface sediments of organic

enriched streams (Bowes et al., 2008; Jarvie et al., 2008b). A

simpler model in which B was set to zero was also tested. In some

cases, inclusion of the B coefficient did not improve the model

solution. However, for other datasets the B parameter was required

to substantially improve the model fit.

The load apportionment model requires that the dataset of P

concentrations and flow for each site cover the full range of river

flow conditions at the site. The P concentration datasets used in

this study are based on regular weekly sampling, which may

under-represent high-flow events and their associated P export.

However,in this study,the weeklysampling over a 2–3year period

provided good representation of the flow range during themonitoring period. Stream flow recorded at the time of water

quality sampling covered 82–98% of the range in daily mean flows

for all sites, apart from Priors Farm (where a lower coverage of 65%

reflects the flashier flow regime on the clay). The load apportion-

ment model applies the relationship between P concentration and

flow to model daily P concentrations and loads, which minimises

bias towards low flows.

Examples of how the model apportions the relative contribu-

tions of TP, SRP, DHP and PP from different sources are given in

Fig. 1. The modelled fit is produced from the sum of the

continuous’ (‘point’ source) and flow-dependent (‘diffuse’ source)

input estimates. The intersection of these two lines is the

estimated flow at which the flow-dependent and continuous load

inputsare equal (Q e; m3

sÀ1

).At flowslessthan the Q e, continuoussources dominate the load. Since mean daily flow data were

available for each of the sampling sites throughout the weekly

monitoring period, the percentage of time that continuous

sources were dominant was estimated by calculating the

percentage of mean daily flow values which fell below the Q evalue at each site. Although continuous sources are typically

equated with point sources, they can refer to any source which

reaches a maximum concentration below the Q e, so could also

include ‘background’ sources such as groundwater or in-stream

sources. Continuous inputs may not be uniformly constant in

concentration (for example, effluent discharges may show

diurnal and seasonal variability (Palmer-Felgate et al., 2008)),

however they contribute consistently under varying flow condi-

tions and thus have greatest impact on in-stream concentrations




under low-flow conditions. Although flow-dependent sources are

typically equated with diffuse sources mobilised from catchment

and within-channel sources during high-flow events, it is

recognised that some in-channel sources of P may have been

ultimately derived from point sources and stored in streambed

sediments (Dorioz et al., 1998)

The model was unable to find a satisfactory solution forWhitchurch (the low agricultural intensity catchment in the

Wye), as concentrations of P were very low at this site, and there

was no clear relationship between P concentrations and flow.

Similar observations have been made by Bowes et al. (2008)

where inputs to the river were neither continuous nor clearly

flow related.

4. Results

4.1. Gradients in phosphorus and nitrogen concentrations across the

study sites and contrasts between low and high agricultural intensity

catchments

4.1.1. Phosphorus fractionsThestreams chosenfor this studycovereda wide gradation ofP

concentrations, ranging from very clean ‘near-pristine’ low

agricultural intensity streams of the Welland and Wye, where

SRP concentrations were just a few mg-P L À1, to heavily nutrient-

impacted streams where TP concentrations were several hundred

mg-P L À1 (Fig. 2). For each of the three study basins (Wye, Avon

and Welland), average concentrations of each of the major P

fractions (TP, SRP, DHP and PP) were consistently greater in

streams draining the high agricultural intensity catchments,

compared with their corresponding low agricultural intensity

catchments (Fig. 2).

These results show that the speciation of P varied consider-

ably between catchments (Fig. 2; Table 3). Priors Farm, which

drained more intensive livestock farming, exhibited high

concentrations of all P fractions and the highest concentrations

of DHP and PP. In contrast, Kivernoll, which received effluent

from a village STW was characterised by very high SRP, but

relatively low PP concentrations. Lone Pine, which also received

effluent (albeit from septic tank rather than STW discharges),

had lower SRP concentrations than Kivernoll, but higher DHP, PP

and SS-P concentrations. Indeed, at Lone Pine, DHP and PPexpressed as a percentage of TP (%DHP and %PP) were twice as

high as those at Kivernoll (Table 3). Belton Bridge, which drains

intensive arable farmland, had lower TP concentrations than the

other high agricultural intensity catchments, but the highest

%PP. The low agricultural intensity catchments (Whitchurch,

Digby Farm and Cools Cottage) tended to have lower concen-

trations of most P fractions and lower %DHP and %PP values than

the predominantly arable catchments. For the predominantly

arable catchments, as the influence of human settlement

increased, median TP concentrations increased (Fig. 2). How-

ever, median %PP values decreased with increasing influence of

human population, which were directly linked with increases in

SRP concentrations (Table 3).

4.1.2. Nitrogen fractions

N fractions did not show such a well-defined contrast between

low and high agricultural intensity catchments, compared with P

fractions. For example, in the Welland, the median TDN concentra-

tions at the low agricultural intensity site (Digby Farm) exceeded

thatof the highagricultural intensity site(Belton Bridge)and, for the

Avon, median TDN concentrations were identical at both the low

agricultural intensity site (Cools Cottage) and high agricultural

intensity site (Priors Farm) (Fig. 2). TDN concentrations were

consistentlyhigherin theWye thanthe Avon andWelland. Across all

the catchments, TDN concentrations ranged from 4 mg-N L À1 at

BeltonBridgeto 13.4 mg-N L À1at Dinedor. Nitrate(NO3À-N) was the

dominant N fraction(Table 3), accounting forbetween 71%of TDNat

Priors Farm and 95% of TDN at Whitchurch. Ammonium (NH4

+

-N)

Fig. 1. Example load apportionment solutions for different P fractions in selected catchments of the Avon, Wye and Welland.




accounted for<1% of TDN at Whitchurch,Dinedor andKivernoll,ca.

1% of TDN at Cools Cottage,Priors Farm and Belton Bridge and 6% of

TDN at Lone Pine. Median NH4+-N concentrations ranged from

0.014 mg-N L

À1

at Digby Farmto 0.45 mg-N L

À1

atLonePine(Fig. 2).

DON was consistently lower in the low agricultural intensity

catchments compared with the higher intensity catchments (Table

3), and DON accounted for between 5% of TDN at Whitchurch and

26% of TDN at Priors Farm.

Fig. 2. Boxplots showing concentration gradients (medians and inter-quartile ranges) across the Avon, Wye and Welland catchments, for total phosphorus (TP), soluble

reactivephosphorus(SRP), dissolvedhydrolysablephosphorus (DHP), particulate phosphorus (PP), particulate phosphorus normalised to suspended sediment concentration

(SS-P), total dissolved nitrogen (TDN), ammonium (NH4+-N) and dissolved organic carbon (DOC). Notethe catchment order varies with each determinand,as catchments are

plotted in order of increasing concentration. H in parentheses denotes high agricultural intensity catchments; L denotes low agricultural intensity catchments.




4.2. Seasonal variability and flow dependence of phosphorus and

nitrogen concentrations

4.2.1. Phosphorus fractions

4.2.1.1. Soluble reactive phosphorus. Strong seasonality in SRP

concentrations was seen at Kivernoll, Lone Pine and Priors Farm,

which were the catchments with highest average SRP concentra-

tions (Fig. 3; Kivernoll is shown as an example). In these

catchments, maximum SRP concentrations (in the order of several

hundred to over 1000 mg-P L À1) occurred during between July and

September and minimum concentrations (typically <200mg-

P L À1) occurred between December and February. A reversed

pattern in SRP seasonality was observed at Digby Farm ( Fig. 3),

Table 3

Median concentrations of P fractions expressed as a percentage of total phosphorus (soluble reactive phosphorus (%SRP), dissolved hydrolysable phosphorus (%DHP) and

particulatephosphorus(%PP));median concentrations of suspended sediment (SS),suspendedsediment P (SS-P: particulateP normalisedto suspended sediment concentration),

nitrate(NO3À-N), dissolved organic nitrogen (DON);medianvalue of DONexpressed as a percentage of total dissolved nitrogen (%DON)and median iron concentration(Fe), for

each of the catchments of the Wye, Welland and Avon. L denotes l ow agricultural intensity catchments; H denotes high agricultural intensity catchments.

Basin Catchment %SRP %DHP %PP SS (mg L À1) SS-P (mg-Pmg-SSÀ1) NO3À-N (mg-NL À1) DON (mg- N L À1) %DON Fe (mg L À1)

Wye Whitchurch (L) 50 12 36 10 1.6 7.9 0.41 5 7.6

Dinedor (H) 65 11 21 10 2.2 11.7 1.25 9 14.9

Kivernoll (H) 86 4 9 6 4.0 11.6 1.15 9 20.7

Welland Digby Farm (L) 35 28 25 5 0.8 4.1 0.29 8 7.0Belton Bridge (H) 28 11 57 6 9.1 3.7 0.44 13 35.4

Lone Pine (H) 69 8 18 4 12.5 5.9 0.92 12 26.4

Avon Cools Cottage (L) 47 12 38 14 3.1 3.8 0.61 13 38.5

Priors Farm (H) 49 13 36 15 8.1 3.2 1.26 26 206

Fig. 3. Example concentration time series for streamwater concentrations of soluble reactive phosphorus (SRP), dissolved hydrolysable phosphorus (DHP), nitrate (NO3À-N),

ammonium(NH4+-N),dissolved organic nitrogen(DON) anddissolved organic carbon (DOC), in selectedcatchmentsof the Avon, Wye andWelland.The datahave been fitted

with a loess smoothing curve. H in parentheses denotes high agricultural intensity catchments; L denotes low agricultural intensity catchments.




with lowest SRP concentrations (falling below analytical detection

limits of <7mg-P L À1) between April and September, with highest

concentrations (typically 10–25mg-P L À1) during the winter

months. Similarly at Belton Bridge, there were dips in SRP

concentrations during the spring and summer months (Fig. 3).

Whitchurch, Dinedor and Cools Cottage showed no well-defined

seasonality in SRP concentrations.

Fig. 4 shows the flow dependence of SRP concentrations forselected catchments representative of the three seasonal patterns

identified above:

(a) Sites with highest SRP concentrations during the summer

(Kivernoll, Lone Pine and Priors Farm) showed a strong dilution

effect, with maximum SRP concentrations occurring under

summer/early autumn low-flow conditions and rapid reduc-

tions in SRP concentrations with increasing flow. This was

followed by a subsequent increase in SRP concentrations at

higher flows during the winter, although the winter high-flow

concentrations typically did not exceed the summer low-flow

concentrations.

(b) Sites with lowest SRP concentrations during summer and

autumn low flows (Belton Bridge and Digby Farm) showed

increasing (but highly variable) SRP concentrations as flows

increase.

(c) Sites with no well-defined seasonality (Dinedor, Whitchurch

and Cools Cottage) showed a more mixed pattern of flow

dependence. There was some evidence of dilution with small

increases in flow during the spring and summer, but this

dilution effect was relatively small (in comparison with the

dilution effects at Kivernoll, Lone Pine and Priors Farm), andwas counterbalanced by increases in SRP concentrations at

higher flows, suggesting a mixture of sources.

4.2.1.2. Dissolved hydrolysable phosphorus. Seasonality in DHP

concentrations was seen at each of the stream monitoring sites

(Fig. 3; where BeltonBridge and Lone Pine are shown as examples),

with the exception of Priors Farm. DHP concentrations were highly

variable, but typically peaked between August and November. The

greatest amplitude in seasonal variability of DHP concentrations

was at Lone Pine, Kivernoll and Dinedor, where DHP concentra-

tions typically ranged from negligible to 100–200 mg L À1.

In terms of flow dependence, at most sites there was a large

variability in DHP concentrations at low flows, particularly in

summer and autumn, combined with reductions in DHP concen-

Fig. 4. Graphs showing effects of flow (m3 sÀ1) on streamwater concentrations of soluble reactive phosphorus (SRP), dissolved hydrolysable phosphorus (DHP), particulate

phosphorus (PP), suspended sediment phosphorus (SS-P), nitrate (NO3À-N), ammonium (NH4

+-N) and dissolved organic carbon (DOC), in selected catchments of the Wye,

Avon and Welland.




trations at slightly higher to intermediate flows, and moderate

increases DHP concentrations under higher winter flows (Fig. 4;

Belton Bridge is shown as an example). The exception was Priors

Farm, where there were greater DHP concentrations at both

intermediate and high flows (Fig. 4).

4.2.1.3. Particulate phosphorus. There was no well-defined season-

ality in either PP or SS-P. In most cases, PP showed a positive, but

highly variable relationship with flow (Fig. 4; Belton Bridge is

shown as an example). At all sites, there were elevated SS-P

concentrations at low flows, with a reduction in SS-P with small

increases in flow (Fig. 4; Kivernoll and Lone Pine are shown as

examples). Baseflow SS-P concentrations were particularly high at

Lone Pine (reaching several hundred mg-P mg-SSÀ1).

4.2.2. Nitrogen fractions

4.2.2.1. Nitrate. Strong seasonality in NO3À-N concentrations was

observed for Belton Bridge, Lone Pine and Priors Farm (Fig. 3). At

these three sites, NO3À-N concentrations peaked at >10 mg-N L À1

between November and March, but fell to below detection limits

(<0.1 mg-N L À1) between July and September. A more damped

pattern in NO3À-N seasonal variability was seen at Digby Farm, but

NO3À-N concentrations remained above analytical detection levelsduring the summer. Thethree streamswithin the Wye(Whitchurch,

Dinedor and Kivernoll) all had higher NO3À-N concentrations than

theWelland andAvon streams;minimum NO3À-N concentrationsat

all three Wye streams didnot fall below approximately 6 mg-N L À1.

In all three Wye streams, highest NO3À-N concentrations occurred

during the late autumn and winter months. At Whitchurch and

Dinedor, peak NO3À-N concentrationsin autumn/winter 2005–2006

were substantially higher than peak NO3À-N concentrations during

autumn/winter 2004–2005. At Whitchurch, peak NO3À-N concen-

trations in autumn/winter 2005/2006 exceeded 9 mg-N L À1,

whereas at Dinedor, peak NO3À-N concentrations exceeded

20 mg-N L À1 in autumn/winter 2005/2006. At Kivernoll and

Dinedor, lowest NO3À-N concentrations occurred during the late

summerand early autumn, but forWhitchurch, sporadic low NO3À

-N concentrations occurred throughout the year.

In terms of flow dependence, Belton Bridge, Lone Pine and

Priors Farm showed strongest positive relationships between

NO3À-N and flow, with a rapid reduction in NO3

À-N concentrations

approaching the lowest baseflows in summer and autumn (Fig. 4;

Lone Pine is shown as an example). The other streams showed

more variable and complex relationships with flow.

4.2.2.2. Ammonium. Only Lone Pine showed pronounced season-

ality in NH4+-N concentrations, with large increases in concentra-

tions during the summer (June–September), when peak NH4+-N

concentrations exceeded 6 mg-N L À1 (Fig. 3). The seasonal pattern

ofNH4+-N concentrations at Lone Pine produced a reverse signal to

that of the NO3-N concentrations: NH4

+

-N concentration peakedwhen NO3

À-N was depleted to below analytical detection levels.

This increase in NH4+-N compensated for loss of NO3

À-N, resulting

in no large-scale reductions in summerTDN concentrations at Lone

Pine, unlike Priors Farm and Belton Bridge, where TDN showed

strong seasonality linked to summer depletion of NO3À-N.

Lone Pine showed a strong dilution in NH4+-N concentrations

with flow (Fig. 4). At Priors Farm and Cools Cottage highest NH4+-N

concentrations also occurred under low flows, although, at Priors

Farm, the highest NH4+-N concentrations (up to ca. 3 mg-N L À1)

occurred during winter and spring low flows (Fig. 4). Dinedor,

Kivernoll and Digby Farm showed substantially lower NH4+-N

concentrations, but broadly positive (albeit highly variable)

increases in NH4+-N with flow (Fig. 4; Kivernoll is shown as an

example).

4.2.2.3. Dissolved organic nitrogen and dissolved organic car-

bon. There was little evidence of any well-defined seasonality in

DON or DOC concentrations. At Lone Pine and Priors Farm, the

lowest DON concentrations tended to occur during spring and

early summer (when values approached zero) and highest DON

concentrations occurred in autumn and winter (when maximum

DON concentrations reached ca. 4–5 mg-N L À1) (Fig. 3; LonePineis

shown as an example). There was a high degree of variability in the

magnitude of autumn and winter DON concentrations between

years for any given stream.

At all sites, DON typically showed poor flow dependence, with

considerable variability in response to changing flow conditions.

Highest DOC concentrations occurred during the winter months at

Digby Farm, Cools Cottage and Belton Bridge, with lowest DOC

concentrations at Belton Bridge and Cools Cottage during the

summer (Fig. 3; Cools Cottage is shown as an example). At most

sites DOC showed broadly positive flow dependence (Fig. 4; Belton

Bridge is shown as an example).

At Priors Farm, there were strong positive inter-correlations

(P < 0.00001) between DOC and DON, and DOC and DHP. Lone

Pine, Belton Bridge, Dinedor and Cools Cottage also showed

significant positive correlations (P < 0.00001) between DOC and

DON although, at Kivernoll, P < 0.01, and correlations between

DOC and DON were statistically insignificant for Digby Farm andWhitchurch. Correlations between DOC and DHP were consider-

ably weaker than at Priors Farm for all other sites.

4.3. Load apportionment modelling of phosphorus sources

Flow-dependent P loads (typically ‘diffuse’ sources) and

continuous P loads (typically ‘point’ sources’), estimated from

the load apportionment model are presented in Table 4, together

with the contribution of continuous and flow-dependent sources

to the annual loads. To compare flow-dependent (diffuse) loads

between catchments (and address the effects of variable

watershed area and flow), P yields were calculated by normalising

the flow-dependent loads to catchment area. The proportion of

time that the river load was continuous (point) source dominated(i.e. when Q < Q e) is also presented in Table 4.

Continuous (point) source TP loads varied substantially, from

<1 kg-P yearÀ1 at Cools Cottage at Digby Farm to 667 kg-P yearÀ1

at Kivernoll. In terms of contribution to annual TP loads,

continuous (point) sources accounted for between 0.3% at Cools

Cottage and 58% at Kivernoll. Kivernoll also had the highest

continuous (point) sourcecontributions to annualSRP andPP loads

(67% and 7%, respectively).

Flow-dependent (diffuse) annual TP,SRP, DHP and PP loads were

highest at Priors Farm, Dinedor and Kivernoll. Flow-dependent

annual TP loads ranged from 16.3 kg-P yearÀ1 at Digby Farm to

976 kg-P yearÀ1 at Priors Farm. However, when flow-dependent

loads were normalised to catchment area (to give a catchment

‘yield’), TP, SRP, DHP and PP yields were overwhelmingly greatestatPriors Farm. The flow-dependent TP yield at Priors Farm was

1.98 kg-P haÀ1 yearÀ1, compared with 0.50–0.56 kg-P haÀ1 yearÀ1

for the other high agricultural intensity catchments and 0.36 and

0.37 kg-P haÀ1 yearÀ1 for the low agricultural intensity catchments.

Flow-dependent sources contributed more than 95% of annual TP

loads at Cools Cottage, Priors Farm, Digby Farm and Belton Bridge.

Indeed, flow-dependent sources dominated the annual TP loads at

all sites apartfrom Kivernoll and accounted for more than 90% of PP

loads at all the sites.

Despite the dominance of flow-dependent (diffuse) sources

for annual TP loads, the percentage of the time that TP loads

were dominated by continuous (point) sources was often

disproportionately high. For example, at Dinedor, although

continuous sources accounted for only 33% of the annual TP




Table 4

Estimated continuous (‘point’) and flow-dependent (‘diffuse’) contributions toannual P loads, P yields and the proportion of timethat P loads are point source dominated, der

agricultural intensity catchments; H denotes high agricultural intensity catchments.

P fraction Basin Catchment Continuous

(‘point’)

source load

(kg-PyearÀ1)

Flow-dependent

(‘diffuse’)

source load

(kg-PyearÀ1)

Flow-dependent

(‘diffuse’)

source yield

(kg-P haÀ1yearÀ1)

Total yield

(flow dependent +

continuous)

(kg-P haÀ1yearÀ1)

% of annual load

contributed by

continuous

(‘point’) sources

TP Wye Dinedor (H) 234 483 0.56 0.82 32.6

Kivernoll (H) 667 490 0.50 1.17 57.7

Welland Digby Farm (L) 0.82 16.3 0.37 0.39 4.8

Belton Bridge (H) 1.11 77.6 0.52 0.53 1.4

Lone Pine (H) 25.3 57.4 0.47 0.67 30.6

Avon Cools Cottage (L) 0.19 58.2 0.36 0.37 0.3

Priors Farm (H) 10.4 976 1.98 2.00 1.1

SRP Wye Dinedor (H) 163 171 0.20 0.38 48.8

Kivernoll (H) 542 266 0.27 0.82 67.1

Welland Digby Farm (L) NA 1.28 0.03 NA NA

Belton Bridge (H) 0.05 17.6 0.12 0.12 0.3 Lone Pine (H) 19.0 14.5 0.12 0.27 56.6

Avon Cools Cottage (L) 0.12 22.4 0.14 0.14 0.5

Priors Farm (H) 5.17 379 0.77 0.78 1.3

DHP Wye Dinedor (H) 51.2 24.5 0.03 0.09 67.7

Kivernoll (H) 53.5 35.5 0.04 0.09 60.1


Belton Bridge (H) 1.07 4.91 0.03 0.04 18

Lone Pine (H) 0.68 10.6 0.09 0.09 6.1

Avon Cools Cottage (L) 0 7.6 0.05 0.05 0.0

Priors Farm (H) 1.0 161 0.33 0.33 0.6

PP Wye Dinedor (H) 4.17 303 0.35 0.35 1.4

Kivernoll (H) 13.7 183 0.19 0.20 7.0


Belton Bridge (H) 0.90 57.8 0.39 0.39 1.5

Lone Pine (H) 3.54 36.8 0.30 0.33 8.8

Avon Cools Cottage (L) 0.07 28.4 0.18 0.18 0.3 Priors Farm (H) 3.07 361 0.73 0.74 0.8



load, they provided the dominant contribution to TP loads in the

stream for 75% of the time.

Within the predominantly arable catchments, continuous

(point) source SRP inputs increased with increasing levels of

human settlement influence from <1 kg-P yearÀ1 at Belton Bridge

to 542 kg-P yearÀ

1 at Kivernoll, whilstthe proportion of annualSRPloads accounted for by continuous sources increased from 0.3% at

Belton Bridge, 49% at Dinedor, 57% at Lone Pine to 67% at Kivernoll.

Flow-dependent (diffuse) source TP yields were near constant

across the predominantly arable catchments, irrespective of

human settlement influence (ca. 0.5 kg-P haÀ1 yearÀ1). Similarly,

flow-dependent (diffuse) source PP yields in the predominantly

arable catchments did not increase in line with increasing human

population influence. In contrast, flow-dependent SRP yields in the

predominantly arable catchments increased with greater popula-

tion influence: flow-dependent SRP yields were highest at Dinedor

and Kivernoll (0.20 and 0.27 kg-P haÀ1 yearÀ1, respectively),

compared with Lone Pine and Belton Bridge (both 0.12 kg-

P haÀ1 yearÀ1) and Digby Farm (0.03 kg-P haÀ1 yearÀ1).

Whereas flow-dependent (diffuse) source annual TP yields inthe predominantly arable catchments were largely unaffected

by population, the total annual TP yields (which includes both

flow-dependent and continuous TP sources normalised to

catchment area) increased dramatically with increasing popula-

tion influence from 0.53 kg-P haÀ1 yearÀ1 at Belton Bridge to

0.67 kg-P haÀ1 yearÀ1 at Lone Pine, 0.82 kg-P haÀ1 yearÀ1 at

Dinedor to 1.17 kg-P haÀ1 yearÀ1 at Kivernoll. In contrast, the

low intensity agriculture catchments had total annual TP yields

of 0.37–0.39 kg-P haÀ1 yearÀ1 and Priors Farm (intensive live-

stock farming) had a total annual TP yield of 2 kg-P haÀ1 yearÀ1,

all of which were almost identical to their flow-dependent

(diffuse) source annual TP yields.

Parameter values from the load apportionment modelling are

shown in Table 5. In the majority of cases, B%

0, indicating

continuous inputs were behaving largely conservatively. However,

positive B values were required for the model fits for SRP at

Dinedor (0.89), Kivernoll (0.45) and Lone Pine (0.34), which are

indicative of varying levels of net in-streamSRP retention, with the

highest SRP retention rate at Dinedor. For DHP model fits, positive

B values were required for Dinedor (0.82), Kivernoll (0.78) andBelton Bridge (0.42), indicative of net in-stream DHP retention,

with the highest DHP retention rate also at Dinedor. For PP, a

positive B value of 0.46 was required for the model fit at Lone Pine,

which is indicative of net in-stream PP retention. In contrast, a

negative B value of À1.06 was required for the SRP model fit at

Priors Farm, which is indicative of net in-stream SRP release at this

site.

5. Discussion

5.1. Overview

Total P, DON and DOC concentrations provided the best

indicators of agricultural/anthropogenic ‘impact’ across thestudy sites, with consistently higher concentrations in catch-

ments draining high intensity agricultural land. The most

heavily ‘impacted’ catchments had higher TP concentrations

and typically higher percentages of SRP, particularly for

catchments receiving inputs of sewage effluent (from STW

(Kivernoll) and septic tanks (Lone Pine and Dinedor)). In

contrast, NO3À-N, DHP and PP showed no systematic differences

in concentration between high and low agricultural intensity

streams. Higher median NO3À-N concentrations in the low and

high intensity Wye catchments can be attributed to ground-

water sources which are highly enriched with NO3À-N across the

wider lowland Herefordshire Wye basin, linked to historically

high fertiliser and manure application rates and long ground-

water residence times ( Jarvie et al., 2003).

Table 5

Parameter values from the load apportionment model fits. For catchments, L denotes low agricultural intensity catchments; H denotes high agricultural intensity

catchments. Please see text for explanation of parameters.

P fraction Basin Catchment Model parameter values Q e (m3 sÀ1)

A B C D

TP Wye Dinedor (H) 52.5 0.79 3747 2.74 0.113

Kivernoll (H) 72.5 0.49 550 2.70 0.400

Welland Digby Farm (L) 2.32 0.72 471706 2.87 0.003

Belton bridge (H) 0.04 – 2480 1.61 0.001

Lone Pine (H) 6.51 0.38 9737 2.06 0.013Avon Cools Cottage (L) 0.006 – 175 1.11 0.0001

Priors Farm (H) 0.058 –0.34 483 1.16 0.002

SRP Wye Dinedor (H) 45.6 0.89 888 2.46 0.152

Kivernoll (H) 53.8 0.45 289 2.57 0.453

Welland Digby Farm (L) 0.00 – 70 1.30 0

Belton Bridge (H) 0.0015 – 316 1.43 0.0002

Lone Pine (H) 4.10 0.34 3842 2.21 0.026

Avon Cools Cottage (L) 0.004 – 61 1.09 0.0001

Priors Farm (H) 0.0004 À1.06 178 1.10 0.003

DHP Wye Dinedor (H) 12.47 0.82 154 2.59 0.242

Kivernoll (H) 10.63 0.78 29 1.96 0.423

Welland Digby Farm (L) 0.00 – 471603 3.37 0.003

Belton Bridge (H) 0.32 0.42 218 1.72 0.007

Lone Pine (H) 0.02 – 552 1.67 0.002

Avon Cools Cottage (L) 0.00 – 25 1.13 0

Priors Farm (H) 0.03 – 95 1.21 0.001

PP Wye Dinedor (H) 0.13 – 2676 2.84 0.030

Kivernoll (H) 0.44 – 233 3.81 0.192

Welland Digby Farm (L) 0.002 – 471597 2.94 0.001

Belton Bridge (H) 0.03 – 3102 1.78 0.001

Lone Pine (H) 1.43 0.46 5455 2.01 0.005

Avon Cools Cottage (L) 0.002 – 89 1.12 0.0001

Priors Farm (H) 0.10 – 200 1.15 0.001




The load apportionment modelling results support our initial

choice of low and high agricultural intensity catchments, by

demonstrating (a) minimal continuous (point) source P loads in

the low agricultural intensity catchments, and (b) that flow-

dependent (diffuse) source TP yields in the low agricultural

intensity catchments were consistently lower than in the high

intensity arable catchments (by ca. 25%). Belton Bridge (the

arable catchmentin theWelland) was also confirmedas receiving

minimal continuous (point) source P inputs, compared with the

other high agricultural intensity catchments, despite the loca-

lised influence of septic tank discharges from some domestic

dwellings in this catchment. This probably reflects the low

numbers of residents within the catchment and the indirect

direct discharge to the stream channel via a ditch network which

is likely to retain P under low-flow conditions. Indeed, Arnscheidt

et al. (2007) and Sharpley et al. (2007) highlight the importance

of drainage ditches as ‘reactive conduits’ which can attenuate P

inputs to receiving waters. In contrast, at Dinedor, higher

population pressure and more direct discharge of septic tank

overflows increased the continuous (point) source contribution.

The load apportionment modelling also clearly showed that

intensive livestock farming on heavy clay soils in the Priors Farm

catchment (Avon) produced very high-flow-dependent (diffuse-

source) P yields, which were aroundfour times higherper hectarethan for the high intensity arable catchments.

5.2. Arable land use and influence of increasing levels of human

population on streamwater phosphorus and nitrogen concentrations

The gradation of increasing human settlement across arable-

dominated catchments (Belton Bridge< Dinedor< Lone Pine -

< Kivernoll) provides new information about the relative influence

of population pressures in rural catchments. The increase in

concentrationsof SRP and %SRPwith increasinghuman population,

from BeltonBridge to Kivernoll, linked directly with increasing SRP

loads from point-source inputs and increasing percentage con-

tributions to annual P loads from continuous (point) sources. The

magnitude of flow-dependent (diffuse) P loads was heavilyinfluenced by catchment area and, to factor out the effects of

variable catchment size, flow-dependent loads were normalised to

catchment area (termed ‘yield’). The increasing influence of human

population in the predominantly arable catchments had no

significant impact on flow-dependent (diffuse) source annual TP

yields, with all arable catchments producing very similar diffuse-

source TP yields of ca. 0.5 kg-P haÀ1 yearÀ1, compared with

0.37 kg-P haÀ1 yearÀ1 for the low intensity land use catchments.

In contrast, diffuse-source SRP yields increased in line with

increasing population. This may result from near-river effluent

sources, such as septic tank soakaways being intercepted as water

levels rise, a phenomenon observed within permeable Chalk

catchments of the Hampshire Avon and upper River Thames (Neal

et al., 2005; Jarvie et al., 2008c). Like TP, PP diffuse-source yieldsdid not increase with increasing humanpopulation, demonstrating

that PP yields were not related to effluent inputs, but were directly

linked to diffuse sources from agricultural land, impermeable

surfaces (e.g. roads) and/or in-stream sources (Withers et al.,

2009). However, total yields of TP (which include both flow-

dependent (diffuse) and continuous (point) source annual loads

normalised to catchment area) increased dramatically with

increasing population, by 120% at Kivernoll relative to Belton

Bridge, demonstrating the large impact of point sources which

resulted in a more than doubling of total annual TP yields within

the intensively farmed arable catchments.

There was little change in DHP concentrations or loads with

increasing population in the predominantly arable catchments of

theWye andWelland (the small increase in continuous DHP inputs

at Dinedor and Kivernoll being largely counterbalanced by higher

flow and thus dilution capacity). There was also little change in

DOC concentrations with increasing human population in the

predominantly arable catchments of the Wye and Welland,

although DON concentrations were substantially higher in the

Dinedor and Kivernoll catchments, which receives inputs of

poultry manure containing very high N concentrations (MAFF,

2000).

5.3. Effluent inputs: sewage treatment works versus septic tanks

Lone Pine, which received direct septic tank inputs, had very

high NH4-N and SS-P concentrations compared with other sites

(and strong negative NH4-N and SS-P flow dependence, with

summer concentration maxima). Lone Pine also had higher PP,

DHP and DOC concentrations than Kivernoll (which received

effluent from a village STW). These differences in N and P

hydrochemistry between Kivernoll and Lone Pine reflect differ-

ences in levels of wastewater treatment within septic tanks and

sewage treatment works. At Kivernoll, the biological trickle filter

sewage treatment processes (and subsequent in-stream proces-

sing) appear to be more successful in organic matter oxidation,

aerobic microbial NH4-N nitrification and settling fine particu-

lates than the septic tanks at Lone Pine. Lower water residencetimes within septic tanks, togetherwith lackof soakawaycapacity

in heavy clay soils and consequent rapid runoff and direct

dischargeof septic tankeffluentto the stream arelikely to account

for lower levels of particulate matter retention at Lone Pine. High

NH4-N (and DOC) concentrations at Lone Pine are indicative of

lower rates of aerobic digestion during wastewater treatment in

the septic tanks (Wilhelm et al., 1994). These effects were also

compounded by the relatively low dilution capacity within the

stream at Lone Pine, resulting in higher ammonia, organic and

particulate concentrations and approximately double the %DHP

and %PP, compared with Kivernoll. At most of the sites, PP was

mobilised under high flows typically in autumn and winter, and

the relationship between PP and flow was highly variable (as a

result of multiple in-channel, near-channel, catchment sourcesand exhaustion effects). In contrast, at Lone Pine, PP and SS-P

concentrations were highest during low flows in summer and

autumn. This suggests that a major source of fine P-rich sediment

(as exemplifiedby SS-P)at Lone Pine came directly from theseptic

tanks.

5.4. Influence of livestock farming on streamwater quality and

phosphorus and nitrogen loads

Livestock farming at Priors Farm resulted in significantly higher

concentrations of DHP, PP, DON and DOC (approximately double

the concentration than the other high agricultural intensity

catchments), demonstrating that intensive cattle farming located

in close proximity to the stream channel is a major source of organic and particulate N and P. Indeed, the load apportionment

modelling for Priors Farm showed that diffuse sources accounted

for>98% of TPand SRP annualloads and>99% of DHP and PPloads,

with annual diffuse-source TP yields of 1.98 kg haÀ1 yearÀ1, which

was approximately four times higher than the annual diffuse-

source TP yield at Kivernoll. Although there are known STW and

septictank discharges in thePriors Farm catchmentresponsiblefor

localised very large SRP concentrations in the stream, these did not

appearto have any major significance with respect to annual loads,

although there was a strong continuous (point-source) influence

on SRP concentrations.

Thehigh organic N andP concentrations andstrong correlations

between DOC, DON and DHP at Priors Farm suggest that these

three fractionswere alllinked to a dominant manure/slurry source.




At most of the other stream monitoring sites, highest concentra-

tions of DOC and DON occurred during autumn and winter, due to

mobilisation from the catchment surface during higher flows,

whereas at Priors Farm, the time series showed a more complex

pattern. The DOC, DON and DHP time series at Priors Farm may

well be linkedto near-river farmyardsources of manureand slurry,

which had both flow-dependent and continuous source compo-

nents. In the other high agricultural intensity catchments, there

was no single dominant manure/slurry source and, whilst there

was a correlation between DOC and DON, the relationship with

DHP was more complex, which may relate to multiple DHP

sources, including variable organic and polymeric P contributions

from sewage effluent as well as microbial breakdown of organic

matter (Monbett et al., 2007). Indeed, the poorer correlations

between organic N and P concentrations at the low agricultural

intensity catchments (Whitchurch and Digby Farm) may reflect

high variability in DON and DOC concentrations during breakdown

of natural organic matter and a wide range of allochthonous and

autochthonous organic matter sources (Kalbitz et al., 2000).

These results suggest thatintensive livestockfarming on heavy

clay soils provides the greatest risk of diffuse-source P loss of all

the agricultural land use types investigated during this study.

Livestock farming produced substantially higher diffuse-source

yields for all P fractions compared with arable farming, andlivestock farming was characterised by particularly high organic

and particulate contributions, linked to manure sources which

were largely flow dependent. The location of cattle farming

enterprises with high stocking densities in close proximity to the

stream channel, and grazing land on poorly drained heavy clay

soils close the stream channel, provided direct hydrological

connectivity for flow-dependent delivery of livestock manure.

This provided a ‘critical source area’ for P loss whereby high P

inputs occur in areas where there is high potential for surface

runoff to the stream (Sharpley et al., 1994, 2001). Additionally,

continuous inputs such as leakage from slurry stores and dirty

water from farmyards/cattle housing, as well as direct access for

cattle to the stream for drinking, provide direct inputs and

contribute to relatively high baseline streamwater P and Nconcentrations. Similar results have been observed for faecal

indicator organisms (FIOs) in lowland UK grassland catchments,

which aredirectlyattributable to livestocksources (Aitken,2003).

Analogous to P inputs at Priors Farm, diffuse-source FIOs in

lowland grassland systems are rapidly flushed into streams

during high-flow events, but elevated concentrations under low

flows are associated with ‘point’ sources derived from poor slurry

containment and dirty water management (Crowther et al.,

2001).

5.5. Timing and delivery of phosphorus and nitrogen inputs: flow-

dependent versus continuous sources

Flow-dependent (‘diffuse’) sources accounted for the highestproportion of annual TP loads, relative to point sources, in all

catchments, apart from at Kivernoll (the site with the highest

human population pressure and greatest effluent inputs). How-

ever, theproportionof timethat daily loads werepoint- or diffuse-

source dominated provides a different perspective: although

periodic high-flow events generated the majority of annual TP

loads, continuous (point) sources dominated daily loads for a

higher proportion of time than flow-dependent (diffuse) sources

at all sites except for Cools Cottage,Prior’s Farmand Belton Bridge.

This indicates that, even in the high intensity arable catchments,

where diffuse sources dominate annual loads, point sources have

a much more significant day-to-day impact on streamwater TP

loadings. Similar patterns were also seen for other P fractions.

However, Belton Bridge has a very low residential population and

so the continuous (point) source inputs probablyarise fromeither

the sparse septic tanks, background baseflow inputs from

subsurface runoff or mobilisation from in-stream sources, at

flows below the Q e. This provides an important reminder that

continuous sourcesare notonly ‘traditional’ point sources,but can

also be derived from background sources such as groundwater,

which have greatest proportional inputs under low flows. In most

cases, however, groundwater contributions of P to streamwater

are very low, as a result of high ‘self-purification capacity of

carbonate-rich aquifers, by CaCO3-P co-precipitation (Neal, 2001)

and therefore groundwater sources are typically swamped by

point-source inputs, even in rural areas ( Jarvie et al., 2008c; Neal

et al., 2005).

Load apportionment modelling of P sources and the proportion

of time that daily P loads are predominantly flow dependent or

continuous provides a starting point for prioritisation of the major

contributing sources, for mitigation purposes. However, the

magnitude of P loads do not per se provide a good indicator of

ecological exposure risk. Concentrations of P, rather than P loads,

are of greater significance for ecosystem response in flowing

freshwaters (Edwards et al., 2000), particularly the magnitude,

duration and timing of elevated P concentrations and their

synchronicity in relation to water retention times and periods of

biological demand and ecosystem sensitivity. Continuous sourcesmay therefore be of greater eutrophication risk because they

produce highest concentrations under the lowest flow conditions

during summer (when dilution effects are at a minimum) and

when water residence times and eutrophication risk are highest.

This is illustrated by the time series of SRP at Kivernoll, Lone Pine

and Priors Farm that all show strong flow dilution effects and

maximum SRPconcentrationsunderlow flows duringsummer and

early autumn. In contrast diffuse sources are flow dependent, with

highest concentrations generated during high flows with low

water residence times, typically during autumn and winter, which

do not usually correspond with times of greatest eutrophication

risk in flowing freshwaters. This is illustrated by the time series for

nitrate, where maximum stream concentrations occurred between

November and March.

5.6. In-stream phosphorus and nitrogen processing

SRP, DHP, NO3À-N and NH4

+-N all showed evidence of in-

stream processing. For SRP, catchments with no significant point-

sourceinputs (Belton Bridgead Digby Farm), andwhich hadmuch

lower SRP concentrations than the sewage impacted catchments,

showed seasonal reductions in SRP concentrations (to below

analytical detection limits) during the summer. The timing of SRP

loss suggests that this may be linked to biological P retention,

reducing downstream SRP concentrations to levels which may

limit further primary productivity, similar to that described by

Mulhollandet al. (1985) and Svendsenet al. (1995). However,the

summer reductions in SRP were, to some extent, partiallycompensated by increases in DHP during late summer and

autumn (observed at all sites apart from Priors Farm). Indeed, a

companion study showed enhanced rates of phosphatase activity

at Digby Farmand Belton Bridge (unpublisheddata).This suggests

that organic P sources may be utilised as a P source to sustain

primary productivity, when the more readily bioavailable SRP

concentrations weredepleted (Bowman et al., 2005). Catchments

influenced by point-source inputs (both septic tank and sewage

effluent) (i.e. Priors Farm, Lone Pine and Kivernoll) showed large

seasonal variability in SRP concentrations, but with a reversed

pattern to that observed at Digby Farm and Belton Bridge. Priors

Farm, Lone Pine and Kivernoll showed pronounced summer SRP

maxima and winter minima, with strong negative flow depen-

dency, linked to seasonal changes in flow and point-source




dilution ( Jarvie et al., 2006). During the summer, increased

concentrations resulting from lower baseflow dilution would

have completely masked any in-stream biological SRP uptake.

The load apportionment model provides new insight into in-

stream processing of P fractions, through model-fit values of the B

parameter, which give an indication of the degree to which P

fractions behave conservatively or non-conservatively. For the SRP

model fits, positive B parameter values at Dinedor, Kivernoll and

Lone Pine (catchments with the greatest ‘point’ source influence)

indicate net in-stream SRP retention at low flows. These results

correspond with laboratory sorption studies (equilibrium P

concentrations) of streambed sediments at the same sites

(Palmer-Felgate et al., 2009), which showed capacity for SRP

uptake from the streamwater. However, streamwater SRP time

series at Dinedor, Kivernoll and Lone Pine demonstrated maximum

concentrations during summer baseflow conditions, when point-

source dilution is lowest. These results therefore suggest that these

dominant hydrological controls mask any in-stream SRP retention

at Dinedor, Kivernoll and Lone Pine. Indeed, even higher summer

baseflow SRP concentrations might be expected if SRP was

behaving conservatively (no in-stream SRP uptake) and if

hydrology alone was controlling streamwater SRP concentrations

in these three catchments.

In contrast, Priors Farm, which receives large inputs of organicmatter from livestock farming, had a negative B parameter value

for the SRP model fit, indicating net in-stream SRP release.Parallel

studies have shown high organic matter content in streambed

sedimentsat Priors Farm: ca.20% loss on ignition ( Jarvie and King,

2007). The link between high organic enrichment of bed

sediments and SRP release at low flows, particularly during the

summer has been observed in other recent studies of lowland

agricultural catchments ( Jarvie et al., 2005, 2008b). SRPrelease is

likely to arise from microbial mineralization of organic matter,

microbial Fe reduction and reductive dissolution of Fe(III)

phosphate and Fe(III)(oxyhydr)oxides, releasing phosphate.

Indeed, the high Fe concentrations in the water column at Priors

Farm indicate dissolution of Fe minerals and redox-regulated Fe

release at the sediment–water interface, with subsequentproduction of Fe–organic complexes or precipitation of Fe oxide

nanocolloidsunder more oxidisedconditions in the water column

( Jarvie et al., 2008a).

For PP model fits, positive B parameter values at Lone Pine

suggest that PP settled out and was retained within the stream

channel during low flows. This may also account for the high SS-P

values observed at Lone Pine, since, as the coarser material settles

out, finer and/or less dense (organic-rich) particles are retained in

suspension, and these tend to have higher PP concentration

relative to their mass.

Nitrate showed a classic flow-dependent diffuse-source

mobilisation during the winter, but in-stream loss (denitrifica-

tion) during the summer. At Lone Pine, Belton Bridge and Priors

Farm NO3À

-N concentrations fell below analytical detectionlimits during the summer. At Belton Bridge and Priors Farm, TDN

concentrations also approached zero, suggesting that, at these

two sites, primary production may have become N-limited

during the summer months (Dodds and Welch, 2000; Marcarelli

et al., 2009). At Lone Pine, the loss of NO3À-N during the

summer was largely compensated by increases in NH4+-N,

resulting in no significant summer-time reduction in TDN

concentrations. The summer increase in NH4+-N at Lone Pine

resulted from high rates of anaerobic decomposition of organic

matter either within the streambed sediments (Garban et al.,

1995) or within the septic tank system itself (Wilhelm et al.,

1994). The build-up of ammonium and loss of nitrate suggests

that rates of anaerobic decomposition/ammonification > rates of

aerobic nitrification<

rates of anaerobic denitrification (Wil-

helm et al. , 1994). At Priors Farm, the highest NH4+-N

concentrations occurred under winter low flows, suggesting

that, under cooler winter conditions and high organic matter

availability, rates of decomposition/ammonification> rates of

nitrification, resulting in a build-up of relatively high concen-

trations of NH4+-N. At all the other sites, highest NH4

+-N

concentrations occurred sporadically under higher flows, which

may be related to mobilisation of near-river/catchment sources

and/or reduced efficiency of sewage treatment works or septic

tank treatment as a result of higher rates of water throughput

and reduced residence times.

6. Conclusions

This study provides an overview of theimpacts of rural land use,

characteristic of large areas of lowland Britain, on streamwater P

and N concentrations and loads and sources of P. The results show

that:

1. Intensive arable farming, by itself, had only a relatively small

impact on streamwater P loads, resulting in highly consistent

diffuse-source TP yields of ca. 0.5 kg-P haÀ1 yearÀ1 for both clay

and loam soils, compared with 0.4 kg-P haÀ1 yearÀ1 for low

agricultural intensity grassland/woodland on similar soils types.In contrast, intensive livestock farming on heavy clay soils

resulted in the much higher streamwater P loadings compared

with all other land uses, with a diffuse-source TP yield of 2 kg-

P haÀ1 yearÀ1.

2. Streamwater draining the livestock-dominated catchment was

characterised by high concentrations of organic P, C and N

fractions associated with manure and slurry sources. High

winter stocking densities in cattle barns and farmyards close to

the stream channel, together with summer grazing on poorly

drained heavy soils with direct access to the stream channel,

resulted in highly efficient delivery of livestock manure,

farmyard waste and slurry to the stream channel throughout

the year.

3. There were significant continuous point-source inputs tostreams, even where there were no sewage treatment works

within the catchment. This indicates that, even in rural

headwater catchments in Britain, the human population within

farming communities is great enough to cause significant P

pollution, and that the septic tank systems used by these rural

communities are actually operating as multiple point sources,

rather than a diffuse-source input.

4. Streams which were most heavily impacted by sewage effluent

and septic tank discharges also showed evidence, from load

apportionment modelling, of greatest in-stream uptake of SRP.

This indicates considerable ‘self-purification’ capacity of

streams by sorption of SRP to bed sediments.

Theapproach adopted in this study of combining detailed monitoringof P and N concentration dynamics under changing flow conditions

through the year, with load apportionment modelling, has wider

relevance because it provides a first step for identifying and

prioritising source types which contribute most at particular times

of the year and under conditions where stream ecology may be at

most risk of eutrophication.

Acknowledgements

This work was carried out with funding from the UK

Department for the Environment, Food and Rural Affairs, the

Environment Agency and English Nature (project PE0116: Linking

agricultural land use and practices with a high risk of phosphorus

loss to chemical and ecological impacts in rivers).




References

Arnscheidt, J., Jordan, P., McCormick, S., McFaul, R., McGrogan, H.J., Neal, M., Sims, J.T., 2007. Defining the sources of low-flow P transfers in complex catchments.Science of the Total Environment 382, 1–13.

Aitken, M.N., 2003. Impact of agricultural practices and river catchment character-istics on river and bathing water quality. Water Science and Technology 48,217–224.

Bowes, M.J., Smith, J.T., Hilton, J., Sturt, M.M., Armitage, P.D., 2007. Periphytonbiomass response to changing phosphorus concentrations in a nutrient-impacted river: a new methodology for phosphorus target setting. Canadian

Journal of Fisheries and Aquatic Sciences 64, 227–238.Bowes,M.J., Smith, J.T.,Jarvie, H.P.,Neal,C., 2008. Modellingof phosphorus inputs torivers from diffuse and point sources. Science of the Total Environment 395,125–138.

Bowes, M.J., Smith, J.T., Jarvie, H.P., Neal, C., Barden, R., 2009. Changes in point anddiffuse source phosphorus inputs to the River Frome (Dorset, UK) from 1966 to2006. Science of the Total Environment 407, 1954–1966.

Bowman, M.F., Chambers, P.A., Schindler, D.W., 2005. Epilithic algal abundance inrelation to anthropogenic changes in phosphorus bioavailability and limitationin mountain rivers. Canadian Journal of Fisheries and Aquatic Sciences 62, 174–184.

British Standards Institution, 1996. Guide to the Acquisition and Management of Meteorological Precipitation Data. BSI 743. .

Carpenter, S.R., Bolgrien, D., Lathrop, R.C., Stow, C.A., Reed, T., Wilson, M.A., 1998.Ecological and economic analysis of lake eutrophication by nonpoint pollution.Australian Journal of Ecology 23, 68–79.

Carpenter, S.R., Caraco, N.F., Correll, D.L., Howarth, R.W., Sharpley, A.N., Smith, V.H.,1998. Nonpoint pollution of surface waters with phosphorus and nitrogen.Ecological Applications 8, 559–568.

Chambers, B.J., Garwood, T.W.D., Unwin, R.J., 2000. Controlling soil water erosionand phosphorus losses from arable land in England and Wales. Journal of Environmental Quality 29, 145–150.

Cheng, W.P., Chi, F.H., 2003. Influence of eutrophication on the coagulation effi-ciency in reservoir water. Chemosphere 53, 773–778.

Crowther, J., Kay, D., Wyer, M.D., 2001. Relationships between microbial waterquality and environmental conditions in coastal recreational waters: the Fyldecoast, UK. Water Research 35, 4029–4038.

Dodds, W.K., Welch, E.B., 2000. Establishing nutrient criteria in streams. Journal of the North American Benthological Society 19, 196.

Dorioz, J.M., Cassell, E.M., Orand, A., Eisenman, K.G., 1998. Phosphorus storage,transport and export dynamics in the Foron River watershed. HydrologicalProcesses 12, 285–309.

Edwards, A.C., Twist, H., Codd, G.A., 2000. Assessing the impact of terrestriallyderived phosphorus on flowing water systems. Journal of Environmental Qual-ity 29, 117–124.

Edwards, A.C., Withers, P.J.A., 2007. Linking phosphorus sources to impacts indifferent types of water body. Soil Use and Management 23, 133–143.

Edwards, A.C., Withers, P.J.A., 2008. Transport and delivery of suspended solids,

nitrogenand phosphorus fromvarious sourcesto freshwatersin the UK. Journalof Hydrology 350, 144–153.

Eisenreich, S.J, Bannerman, R.T., Armstrong, D.E., 1975. Simplified phosphorusanalysis technique. Environmental Letters 9, 43–53.

Garban,B., Ollivon,D., Poulin,M., Gautier,V., Chesterikoff,A., 1995. Exchanges at thesediment–water interface in the River Seine, downstream of Paris. WaterResearch 29, 473–481.

Gburek, W.J., Sharpley, A.N., 1998. Hydrologic controls on phosphorus loss fromupland agricultural watersheds. Journal of Environmental Quality 27, 267–277.

Jarvie, H.P., Withers, P.J.A., Neal, C., 2002. Review of robust measurement of phosphorus in river water: sampling, storage, fractionation and sensitivity.Hydrology and Earth System Sciences 6, 113–131.

Jarvie, H.P., Neal, C., Withers, P.J.A., Robinson, A., Salter, N., 2003. Nutrient waterquality of the Wye catchment, UK: exploring patterns and fluxes using theEnvironment Agency data archives. Hydrology and Earth System Sciences 7,722–743.

Jarvie, H.P., Jurgens, M.D., Williams, R.J., Neal, C., Davies, J.J.L., Barrett, C., White, J.,2005. Roleof river bedsediments as sourcesand sinksof phosphorus acrosstwo

major eutrophic UK river basins: the Hampshire Avon and Herefordshire Wye. Journal of Hydrology 304, 51–74. Jarvie, H.P., Neal, C., Withers, P.J.A., 2006. Sewage-effluent phosphorus: a greater

risk to river eutrophication than agricultural phosphorus? Science of the TotalEnvironment 360, 246–253.

Jarvie,H.P., King, S.M., 2007. Small-angleneutronscatteringstudy of naturalaquaticnanocolloids. Environmental Science & Technology 41, 2868–2873.

Jarvie, H.P., Haygarth, P.M., Neal, C.,Butler, P., Smith, B., Naden, P.S., Joynes, A., Neal,M., Wickham, H., Armstrong, L., Harman, S., Palmer-Felgate, E.J., 2008. Stream

water chemistry and quality along an upland-lowland rural land-use conti-nuum, south west England. Journal of Hydrology 350, 215–231.

Jarvie, H.P., Mortimer, R.J.G., Palmer-Felgate, E.J., Quinton, K.S., Harman, S.A., Carbo,P., 2008. Measurement of soluble reactive phosphorus concentration profilesand fluxes in river-bed sediments using DET get probes. Journal of Hydrology350, 261–273.

Jarvie, H.P., Withers, P.J.A., Hodgkinson, R., Bates, A., Neal, M., Wickham, H.D.,Harman, S.A., Armstrong, L., 2008. Influence of rural land use on streamwaternutrients and their ecological significance. Journal of Hydrology 350, 166–186.

Kalbitz, K., Solinger, S., Park, J.H., Michalzik, B., Matzner, E., 2000. Controls on thedynamics of dissolved organic matter in soils: a review. Soil Science 165, 277–

304.Leeks,G.J.L., Neal, C.,Jarvie,H.P., Casey,H., 1997. TheLOIS river monitoring network:strategy and implementation. Science of the Total Environment 194, 101–109.

MAFF (Ministry of Agriculture, Fisheries and Food), 2000. Fertiliser Recommenda-tions for Agricultural and Horticultural Crops (RB209). The Stationery Office,London.

Marcarelli, A.M., Bechtold, H.A., Rugenski, A.T., Inouye, R.S., 2009. Nutrient limita-tion of biofilm biomass and metabolism in the Upper Snake River basin,southeast Idaho, USA. Hydrobiologia 620, 63–76.

Monbett, P., McKelvie, I.D., Saefumillah, A., Worsfold, P.J., 2007. A protocol to assessthe enzymatic release of dissolved organic phosphorus species in waters underenvironmentally relevant conditions. Environmental Science & Technology 41,7479–7485.

Mulholland, P.J., Newbold, J.D., Elwood, J.W., Ferren, L.A., Webster, J.R., 1985.Phosphorus spiraling in a woodland stream: seasonal variations. Ecology 66,1012–1023.

Murphy,J., Riley, J.P.,1962.A modified single solutionmethodfor thedeterminationof phosphate in natural waters. Analytica Chimica Acta 27, 31–36.

Neal, C., Neal, M., Wickham, H., 2000. Phosphate measurement in natural waters:

two examples of analytical problems associated with silica interference usingphosphomolybdic acid methodologies. Science of the Total Environment 251,511–522.

Neal, C., 2001. The potential for phosphorus pollution remediation by calciteprecipitation in UK freshwaters. Hydrology and Earth System Sciences 5,119–131.

Neal, C., House, W.A., Jarvie, H.P., Neal, M., Hill, L., Wickham, H., 2005. Phosphorusconcentrations in the River Dun, the Kennet and Avon Canal and the RiverKennet, southern England. Science of the Total Environment 344, 107–128.

Neal, C., Jarvie, H.P., 2005. Agriculture, community, river eutrophication and thewater framework directive. Hydrological Processes 19, 1895–1901.

Palmer-Felgate, E.J., Jarvie, H.P., Williams, R.J., Mortimer, R.J.G., Loewenthal, M.,Neal, C., 2008. Phosphorus dynamics and productivity in a sewage-impactedlowland chalk stream. Journal of Hydrology 351, 87–97.

Palmer-Felgate, E.J., Jarvie, H.P., Withers, P.J.A., Mortimer, R.J.G., Krom, M., 2009.Stream-bed phosphorus in paired catchments with different agricultural land-use intensity. Agriculture, Ecosystems & Environment 134, 53–66.

Reynolds, C.S., Davies, P.S., 2001. Sources and bioavailability of phosphorus frac-tions in freshwaters: a British perspective. Biological Reviews 76, 27–64.

Sharpley, A.N., Chapra, S.C., Wedepol, R., Sims, J.T., Daniel, T.C., Reddy, K.R., 1994.Managing agricultural phosphorus for protection of surface waters: issues andoptions. Journal of Environmental Quality 23, 437–451.

Sharpley, A.N., McDowell, R.W., Weld, J.L., Kleinman, P.J.A., 2001. Assessing sitevulnerability to phosphorus loss in an agricultural watershed. Journal of Environmental Quality 30, 2026–2036.

Sharpley, A.N., Krogstad, T., Kleinman, P.J.A., Haggard, B., Shigaki, F., Saporito, L.S.,2007. Managing natural processes in drainage ditches for non-point sourcephosphorus control. Journal of Soil and Water Conservation 62, 197–206.

Sharpley, A.N., Kleinman, P.J.A., Peter, J., Heathwaite, A.L., Gburek, W.J., Weld, J.L.,Folmar, G.J., 2008. Integrating contributing areas and i ndexing phosphorus lossfromagricultural watersheds. Journal of Environmental Quality 37,1488–1496.

Smith, V.H., 2003. Eutrophication of freshwater and coastal marine ecosystems—aglobal problem. Environmental Science and Pollution Research 10, 126–139.

Svendsen, L.M., Kronvang, B., Kristensen, P., Graesbol, P., 1995. Dynamics of phos-phorus-compounds in a lowland river: importance of retention and non-pointsources. Hydrological Processes 9, 119–142.

Wilcock, R.J., Monaghan, R.M., Quinn, J.M., Campbell, A.M., Thorrold, B.S., Duncan,

M.J., McGowan, A.W., Betteridge, K., 2006. Land-use impacts and water qualitytargets in the intensive dairying catchment of the Toenepi Stream, NewZealand. New Zealand Journal of Marine and Freshwater Research 40, 123–140.

Wilhelm, S.R., Schiff, S.L., Cherry, J.A., 1994. Biogeochemical evolution of domesticwaste-waterin septicsystems. 1. Conceptual-model. GroundWater 32, 905–916.

Withers, P.J.A., Jarvie, H.P., Hodgkinson, R.A., Palmer-Felgate, E.J., Bates, A., Neal, M.,Howells, R., Withers, C.M., Wickham, H.D., 2009. Characterization of Phos-phorus Sources in Rural Watersheds. Journal of Environmental Quality 38,1998–2011.


Http Www.sciencedirect.com Science Ob=MImg& Imagekey=B6T3Y-4XT2CHD-2-9& Cdi=4959& User=126036&...

Documents

Transcript of Http Www.sciencedirect.com Science Ob=MImg& Imagekey=B6T3Y-4XT2CHD-2-9& Cdi=4959& User=126036&...