B I O L O G I C A L C O N S E RVAT I O N 1 4 1 ( 2 0 0 8 ) 1 4 1 1 –1 4 2 7

available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/biocon

The impact of organic and inorganic fertilizers and lime

on the species-richness and plant functional characteristics
of hay meadow communities

Francis W. Kirkhama,*, Jerry R.B. Tallowinb, Roy A. Sandersonc, Anne Bhogald,

Brian J. Chambersd, David P. Stevense,z
a
Ecological Research & Consultancy, Far View, Nymet Rowland, Crediton, Devon EX17 6AL, UK
b
BBSRC Institute for Grassland and Environmental Research, North Wyke, Okehampton, Devon EX20 2SB, UK
c
Institute for Research on Environment and Sustainability, Devonshire Building, Newcastle University, Newcastle upon Tyne NE1 7RU, UK
d
ADAS Gleadthorpe, Meden Vale, Mansfield, Nottingham NG20 9PF, UK
e
Countryside Council for Wales, Plas Penrhos, Penrhos Road, Bangor, Gwynedd LL57 2LQ, UK

A R T I C L E I N F O A B S T R A C T

Article history: Vegetation responses to fertilizers and lime applied in a seven-year study at paired unim-
Received 1 September 2007 proved and semi-improved mesotrophic hay meadows in Cumbria and Monmouthshire,
Received in revised form UK, are described in terms of species-richness and several other plant community vari-
11 March 2008 ables. Treatments were farmyard manure (FYM) applied at rates of between 6 and
Accepted 17 March 2008 24 tonne ha1 annually or every third year, inorganic fertilizers giving equivalent amounts
Available online 1 May 2008 of N, P and K, and lime applied either alone or with FYM applied annually or three-yearly.
Annual FYM at 24 tonne ha1 reduced species-richness and the richness of positive indica-
Keywords: tor species at all sites and increased the proportional cover of nutrient-demanding species,
Farmyard manure whilst liming in conjunction with 12 tonne ha1 annual FYM application was equally det-
Vegetation composition rimental at the unimproved site in Wales but not in Cumbria. Inorganic fertilizers were
Agri-environment apparently no more detrimental to vegetation quality than equivalent FYM treatments,
although medium-term nutrient supply from FYM may have been underestimated. Spe-
cies-richness was maintained by FYM at 612 tonne ha1 year1 in the semi-natural north-
ern meadow, where such levels had been used in the past, but only by amounts equivalent
to 66 tonne ha1 year1 at the Welsh sites which had no recent history of fertilizer use. It is
unclear to what extent such differences are attributable to innate differences in plant com-
munity type, rather than to site-specific differences in past management. The implications
of our findings for defining sustainable fertilizer practices to maintain or enhance the nat-
ure conservation value of mesotrophic meadows are discussed.
 2008 Elsevier Ltd. All rights reserved.

1. Introduction berg, 1988; Bakker, 1989; Berendse et al., 1992; Myklestad

and Sætersdal, 2004). In the UK the loss of traditional hay
Increased use of artificial fertilizers is implicated in the wide meadows and other unimproved grasslands has been associ-
scale decline in species-rich grassland across Europe (Ellen- ated with severe declines in many once common plant and

* Corresponding author: Tel./fax: +44 1363 83187.

E-mail addresses: francis.kirkham@gmail.com (F.W. Kirkham), jerry.tallowin@bbsrc.ac.uk (J.R.B. Tallowin), r.a.sanderson@newcastle.
ac.uk (R.A. Sanderson), anne.bhogal@adas.co.uk (A. Bhogal), brian.chambers@adas.co.uk (B.J. Chambers).
z
Deceased.
0006-3207/$ - see front matter  2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.biocon.2008.03.010
1412 B I O L O G I CA L C O N S E RVAT I O N
animal species (Ratcliffe, 1977, 1984; Fuller, 1987; Rich and
Woodruff, 1996; Goulson et al., 2005).
Cynosurus cristatus–Centaurea nigra grassland, the MG5
community of the British National Vegetation Classification
(NVC – Rodwell, 1992), is the most widespread lowland exam-
ple of species-rich grazed hay meadow in the UK (Blackstock
et al., 1999).The MG3, Anthoxanthum odoratum–Geranium sylv-
aticum grassland, is the characteristic grazed hay meadow of
the upland fringes of northern England (Rodwell, 1992). MG5
grassland has close affinity to Atlantic and Sub-Atlantic
Cynosurion grassland in Europe (Rodwell et al., 2007). MG3
grassland has clear affinities with meadows growing at higher
altitudes throughout northern and central mainland Europe
(Rodwell et al., 2007). MG3 communities are reliant upon the
maintenance of traditional hay-making practices without
the use of inorganic fertilizers (Smith and Jones, 1991; Smith
and Rushton, 1994; Smith et al., 1996). The same management
regime, in combination with seed addition, is important in
the restoration of MG3 communities to semi-improved mead-
ows (Smith et al., 2000, 2002). A combination of hay cutting
and grazing is also the most common management of MG5
habitats, although some are managed by extensive grazing
only (Rodwell, 1992; Crofts and Jefferson, 1999).
Use of farmyard manure (FYM) and occasional liming are
traditional management practices for hay meadow communi-
ties (Smith, 1988; Simpson and Jefferson, 1996; Tallowin, 1998;
Crofts and Jefferson, 1999; Jefferson, 2005). However, what con-
stitutes sustainable practices to maintain the nature conserva-
tion value of some species-rich communities is ill defined.
Many studies have shown detrimental effects of inorganic
fertilizers on species-rich meadow vegetation (e.g. Berendse
et al., 1992; Mountford et al., 1993; Kirkham et al., 1996; Silver-
town et al., 2006), but there has been no study in which FYM
treatments have been matched with inorganic fertilizers sup-
plying equivalent amounts of inorganic nitrogen (N), phos-
phorus (P) and potassium (K). Since changes in farming
practice in the UK have reduced the availability of FYM, there
is a need to ascertain if or when inorganic fertilizers can be
sustainable alternatives.
Resolution of these questions is an important pre-requisite
for the refinement of management guidelines for the conser-
vation of biodiversity within existing species-rich, semi-natu-
ral plant communities within statutory sites and agri-
environment schemes. The importance of defining sustain-
able management for restoring such communities to agricul-
turally improved grasslands is also recognized, both in the UK
and in continental Europe (Ormerod, 2003). Improved grass-
lands (MG6 and MG7; Rodwell, 1992) are very common
throughout the UK with close equivalents widespread in
mainland Europe (Rodwell et al., 2007).
An experiment involving several FYM or matched inor-
ganic equivalent fertilizer treatments, and treatments incor-
porating initial liming, with or without annual or
intermittent (three-yearly) FYM applications, was established
in 1999 on paired, unimproved and semi-improved lowland
meadows in south-east Wales and in upland meadows in
north-west England. Treatment effects on botanical composi-
tion were monitored annually to identify: (a) optimum
amounts and frequencies of FYM for the maintenance or
enhancement of vegetation quality; (b) any difference be-
1 4 1 ( 2 0 0 8 ) 1 4 1 1 –1 4 2 7
tween the impact of FYM and equivalent amounts of N, P
and K applied as inorganic fertilizer; and (c) any interaction
between the use of FYM and lime inputs.
Preliminary agronomic and botanical results for 1999 and
2000 were reported elsewhere (Tallowin et al., 2002; Kirkham
et al., 2002). This paper describes botanical responses over
the period 1999–2005. Nomenclature of vascular plants fol-
lows Stace (1997), except for NVC community names which
follow Rodwell (1992).
2. Materials and methods
2.1. Experiment sites
Experimental plots were established on paired agriculturally
unimproved and semi-improved meadows at upland sites in
Cumbria, North West England, and a lowland site in Mon-
mouthshire, Wales. The Cumbrian unimproved site at Rais-
beck was located about 1.5 km from the semi-improved site
at Gaisgill in the Tebay-Orton area at 5427’ N and 234’ W.
The unimproved site in Wales, Pentwyn, was adjacent to
the semi-improved site, Bush, with both being close to Mon-
mouth at 5146 0 N and 241 0 W.
Soils at the Cumbrian sites were clay–loam in texture,
whilst those at the Welsh sites were a mixture of sandy silty
loam to silty clay–loam. The Cumbrian soils were higher in to-
tal N, carbon, organic matter, extractable P and exchangeable
K and sodium (Na), although the most striking difference was
the high level of extractable P at Gaisgill compared with all
the other sites (Table 1).
The vegetation of Raisbeck in 1999 (Kirkham et al., 2002)
corresponded to the MG3b (Anthoxanthum odoratum–Geranium
sylvaticum grassland, Briza media sub-community) of the NVC,
whilst that at Gaisgill corresponded to MG7 grassland, in par-
ticular the MG7e (Lolium perenne–Plantago lanceolata grassland)
(Rodwell, 1992). The vegetation of Pentwyn in 1999 corre-
sponded to the MG5a (Cynosurus cristatus–Centaurea nigra
grassland, Lathyrus pratensis sub-community) but with some
similarities to the MG5c (Danthonia decumbens sub-commu-
nity). The vegetation of Bush was closer to the MG5a and
MG6 (Lolium perenne–cynosurus cristatus grassland) communi-
ties (Rodwell, 1992).
2.2. Past management
Raisbeck, Pentwyn and Bush meadows had previously been
managed over a long period by cutting for hay. Gaisgill had
also been cut for hay and occasionally for silage. Cutting oc-
curred after 1 July at both Raisbeck and Gaisgill, and after
the second week in July at Pentwyn and Bush in compliance
with conditions imposed by the owner, Gwent Wildlife Trust.
Raisbeck had received about 12 tonne FYM ha1 each year,
usually in late April, but occasionally in mid summer follow-
ing the hay harvest, and had received periodic applications of
lime (amounts not known), the last one in about 1993. Gaisgill
had received inorganic fertilizers in March to late April,
whereas FYM was generally applied in February to March.
Prior to about 1979, Gaisgill had received much lower nutrient
inputs.
 B I O L O G I C A L C O N S E RVAT I O N 1 4 1 ( 2 0 0 8 ) 1 4 1 1 –1 4 2 7 1413

Table 1 – Soil chemical properties at experimental sites in Cumbria (Raisbeck and Gaisgill) and Wales (Pentwyn and Bush)
in 1999

pH %N %C %OM Mg kg1
P K Ca Mg Na

Raisbeck 5.2 0.65 6.8 17.2 7.0 250 1879 197 74.4
Gaisgill 4.9 0.45 4.1 11.6 29.5 228 1049 198 64.2
Pentwyn 5.0 0.34 3.2 9.0 4.5 154 934 178 44.0
Bush 5.2 0.32 3.0 8.4 4.9 113 1075 197 36.6
Values relate to air-dried soil from samples taken to a depth of 7.5 cm: pH; total nitrogen (N), carbon (C), and organic matter (OM); Olsen
extractable phosphorus (P) and exchangeable potassium (K), calcium (Ca), magnesium (Mg) and sodium (Na).

Details of past fertilizer application at the Welsh sites are
uncertain. Neither site had received any form of fertilizer or
lime in at least the preceding 10 years, considerably longer
in the case of the unimproved site (>20 years). It was not
known if or when lime was last applied to these sites,
although there is some evidence to suggest that basic slag
and lime may have been applied at some time prior to 1980.
2.3. Experimental design
All experimental treatments (see Table 2), were applied by
hand to individual 7 m · 5 m plots laid out in a randomized
block design with three replicate blocks at each site. Treat-
ments were applied between March and late April. Due to ac-
cess restrictions resulting from the national outbreak of Foot
and Mouth Disease (FMD), no treatment was applied in 2001.
This meant that treatments requiring annual applications
only received a total of 6/7 (85.7%) of the intended amounts
over the seven year period 1999–2005, as reflected in the mean
amounts of N, P and K shown in Table 2. Three-yearly treat-
ments were applied in 1999, 2002 and 2005.
The FYM used in the experiment was sourced locally from
the same farms each year, and was sampled at the time of
application for subsequent chemical analysis. The ADAS
manure nitrogen evaluation routine (MANNER; Chambers
et al., 1999) was used to predict the amount of plant available
N supplied by the FYM. On the basis of previous studies
(Anon, 2000), it was assumed that 60% and 90% of the total
P and K content of the FYM, respectively, was plant available.
The inorganic fertilizer treatments were based upon these
estimates of N, P and K supply by the FYM. Inorganic N, P
and K were applied as ammonium nitrate, triple super-phos-
phate and muriate of potash, respectively, to the relevant
treatments (see Table 2). The FYM treatments used in this
study were designed to encompass the range of inputs tradi-
tionally applied to species-rich hay meadows upon which cur-
rent agri-environment guidelines are based (Simpson and
Jefferson, 1996; Crofts and Jefferson, 1999).
Lime was applied to treatments 13, 14 and 15 (Table 2) at
each site in March–April 1999. Based on soil pH tests con-
ducted in February–March 1999 and an assessment of soil tex-
ture, the amounts of lime were calculated to be sufficient to
raise the soil pH of the experimental plots to 6.0 (Anon,
2000). The lime treatment plots at Raisbeck and Gaisgill re-
ceived, respectively, 2 and 4 tonne of ground limestone ha1
in a single application in September 1999. At Pentwyn and
Bush, totals of 9.4 and 6.8 tonne ha1 were applied in two
installments, in September 1999 and March 2000.
At each site, one plot per replicate block received no fertil-
izer or lime (Treatment 1, nil input control) whilst plots

Table 2 – Treatments applied 1999–2005

Tr. no. Treatment Cumbria Wales

N P K N P K

1# Nil input control 0.0 0.0 0.0 0.0 0.0 0.0

2 Continuation of past inputs (UIa) 6.9 7.2 57.8 0.0 0.0 0.0
2 Continuation of past inputs (SIa) 57.0 22.4 115.7 0.0 0.0 0.0
3 FYM at 6 tonne ha1 annual 3.6 3.5 27.4 2.9 2.8 30.7
4 +, # FYM at 12 tonne ha1 annual 7.1 7.0 54.8 5.8 5.5 61.4
5+ FYM at 24 tonne ha1 annual 14.2 14.0 109.6 11.5 11.1 122.7
6 FYM at 6 tonne ha1 3-yearly 1.9 2.3 12.2 1.6 1.3 15.8
7 +, # FYM at 12 tonne ha1 3-yearly 3.8 4.5 24.4 3.3 2.7 31.6
8+ FYM at 24 tonne ha1 3-yearly 7.6 9.1 48.8 6.5 5.3 63.3
9–12 + inorganic equivalents to Trs. 4,5, 7 and 8, respectively
13 # Lime applied in year 1 0.0 0.0 0.0 0.0 0.0 0.0
14 # As Tr. 13 + FYM as Tr 4 7.1 7.0 54.8 5.8 5.5 61.4
15 # As Tr. 13 + FYM as Tr 7 3.8 4.5 24.4 3.3 2.7 31.6
Values shown are the mean amounts (kg ha1 year1) of N, P and K applied either as FYM (estimated) or in inorganic form (actual), averaged
over seven years 1999–2005. Treatments comprising factorial series for fertilizer Form · Rate · Frequency or Lime · FYM Frequency are indi-
cated by ‘+’ and #, respectively.
a UI = unimproved sites (Raisbeck and Pentwyn); SI = semi-improved sites (Gaisgill and Bush).
1414 B I O L O G I CA L C O N S E RVAT I O N 1 4 1 ( 2 0 0 8 ) 1 4 1 1 –1 4 2 7

designated as Treatment 2 received inputs representing a con-

Table 3 – Positive indicator species for MG3 and MG5
tinuation of the past inputs described in Table 2. At Raisbeck, National Vegetation plant communities (after Robertson
Treatment 2 corresponded exactly with Treatment 4, and re- and Jefferson, 2000)
ceived 12 tonne FYM ha1 each year. The lack of any recent his-
NVC Community
tory of fertilizer or lime application at Pentwyn and Bush meant
that Treatments 1 and 2 were identical at these sites. MG3 MG5 MG5 (contd.)
At all sites, the experimental plots were cut for hay after 1 Alchemilla glabra Agrimonia eupatoria Polygala spp
July each year, the actual cutting date being dependent on Alchemilla spp. Alchemilla glabra Potentilla erecta
weather conditions. The hay aftermath growth was grazed Anemone nemorosa Alchemilla spp Polygala vulgaris
each year, with experimental plots being grazed with the Centaurea nigra Anemone nemorosa Primula veris
remainder of the field. At Raisbeck and Gaisgill mature sheep Cirsium heterophyllum Carex flacca Rhinanthus minor
Conopodium majus Carex nigra Sanguisorba minor
were used, normally commencing in September–October
Euphrasia officinalis Carex panicea Sanguisorba
continuing until March or late April. Pentwyn and Bush were agg. officinalis
grazed by beef store cattle, from mid October/early November Euphrasia spp. Centaurea nigra Serratula tinctoria
to late February. Filipendula ulmaria Euphrasia officinalis Silaum silaus
Botanical assessments were carried out during May each agg.
year, except in 2001 at the Cumbrian sites due to FMD restric- Geranium sylvaticum Euphrasia spp. Stachys officinalis
Geum rivale Filipendula ulmaria Succisa pratensis
tions. The percentage cover of each species present was visu-
Lathyrus pratensis Filipendula vulgaris Tragopogon pratensis
ally estimated in three 1m2 quadrats that were positioned at
Leontodon autumnalis Galium verum
random within each plot in 1999, and then fixed for the pro- Leontodon hispidus Genista tinctoria
ject duration. Leontodon spp. Lathyrus linifolius
Lotus corniculatus Lathyrus pratensis
2.4. Derived variables Persicaria bistorta Leontodon hispidus
Rhinanthus minor Leontodon saxatilis
Sanguisorba officinalis Leucanthemum vulgare
In order to minimize the effects of year-to-year variation and
Succisa pratensis Lotus corniculatus
variation between treatments in vegetation density and total
Trollius europaeus Pimpinella saxifraga
cover (which substantially exceeded 100% cover in most
Species highlighted in bold in the MG3 and MG5 lists were recorded
cases), estimates of cover for individual species were con-
in at least one year at the Cumbrian and Welsh sites, respectively.
verted to a percentage of the total live vegetation cover pres-
ent in each quadrat. Detailed analyses of treatment effects on
individual species are not reported here. Instead the data
were used to calculate several composite variables (see below) from extensive survey data from mainland Europe (Ellenberg,
to characterize the vegetation as a whole. These were then 1988) and modified for British conditions by Hill et al. (1999).
averaged across the three quadrats in each plot to give plot The Ellenberg N value should be regarded as an indicator of
mean values for each variable which were then used as the general soil fertility (Hill et al., 1999). The values are based
basic units for subsequent analyses of treatment effects. upon a nine point scale: species with an N index of 1 are asso-
ciated with very nutrient-poor habitats, with nine indicating
2.4.1. Key variables extremely rich conditions. Species recorded in this study fell
Particular attention is focused on three key variables: species- within the range 2–9.
richness, i.e. the total number of vascular plant species per
m2; the number per m2 of MG3 and MG5 positive indicator 2.4.2. Additional variables
species (Cumbrian and Welsh sites, respectively); and a The following composite variables were also calculated. The
weighted British Ellenberg nitrogen index. Positive mesotro- proportions (%) of total vegetation accounted for by herbs,
phic indicator species (Table 3) were identified from lists pro- grasses, legumes and MG3 and MG5 positive indicator species
duced by Robertson and Jefferson (2000) for use in monitoring (i.e. five variables); the ratio of herbs to grasses in vegetation
the condition of grassland Sites of Special Scientific Interest cover; a weighted British Ellenberg R index (R-score) (Hill
(SSSIs) in MG3 (Anthoxanthum odoratum–Geranium sylvaticum et al., 1999); and competitor, stress-tolerator and ruderal ra-
grassland) and MG5 (Cynosurus cristatus–Centaurea nigra grass- dius scores (Hodgson et al., 1995). Statistical analyses for
land) communities of the NVC (Rodwell, 1992). These two treatment effects were carried out on these additional vari-
communities were considered to be the most relevant types ables. However, only results that aid the interpretation of
for the unimproved sites in Cumbria and Monmouthshire, the treatment effects on the key variables are presented.
respectively, based both upon their location and soil charac- The R-score was tested specifically for the effects of lime.
teristics (Rodwell, 1992) and the vegetation already present The Ellenberg R index has a nine-point scale and is based
in 1999, and are therefore suitable target communities for upon the affinity of a species for a particular level of soil reac-
the corresponding semi-improved meadows. tion (pH). Species associated with very acid conditions have
Weighted Ellenberg N indices (N-scores) were calculated an R index of 1 whilst those associated with very basic condi-
for each quadrat as the mean indicator value averaged over tions score 9. Species recorded here fell within the range 3–7.
the plant species present, weighted according to the propor- The competitor, stress-tolerator and ruderal indices for
tional contribution of each species to total vegetation cover. each species were based upon the three primary plant strate-
The individual species Ellenberg N values were those derived gies proposed by Grime (1977) and used information from a
 B I O L O G I C A L C O N S E RVAT I O N
database in which each species has been allocated a score be-
tween 1 and 5 for each of these three characteristics (Hodgson
et al., 1995), referred to as a ‘radius’ score by reference to the
triangular ordination of the three characteristics described by
Grime (1977). As with the Ellenberg scores, the values used in
analyses (Cr-, Sr- and Rr-scores) were weighted according to
the proportional contribution of each species to total vegeta-
tion cover.
2.5. Data analysis
Data from 2005 for the above variables were tested for treat-
ment effects by analysis of variance (ANOVA) and analysis
of covariance (ANCOVA), the latter using 1999 data as the
covariate (Genstat 5 Committee, 1997). Where an ANCOVA
showed significant treatment covariance with 1999 data, the
adjusted results were used. The 2005 data were analyzed sep-
arately for each site, using three separate ANOVA models in
each case. In each model, variation between replicate blocks
was accounted for by treating replicate as a blocking factor.
The first model included all the treatments applied, whilst
the other two models used balanced factorial subsets (series)
of treatments (see Table 2): the Form · Rate · Frequency (FRF)
series and the Lime · FYM frequency (LFF) series. The FRF ser-
ies tested the effects of form of fertilizer (i.e. FYM or inor-
ganic), the rate at which fertilizers were applied (i.e. 12 or
24 tonne FYM ha1 and inorganic treatments equivalent to
these), the frequency at which treatments were applied (i.e.
annual or 3-yearly), and all two- and three-way interactions
between these factors. The LFF series tested the effects of
lime (i.e. +lime versus lime), the frequency of FYM applica-
tion (i.e. nil, annual or 3-yearly) and the Lime · FYM fre-
quency interaction. All FYM treatments within this series
were applied at 12 tonne ha1. Only data for the three key
variables were tested within both the FRF and LFF series,
whilst the R-score was also tested within the latter series.
Data for the key variables for all treatments and years
1999–2005 at each site (excluding 2001, except for the Welsh
sites) were analyzed together by ANOVAs expanded to include
Year as a repeated measures factor. Whilst these analyses in-
cluded the overall effect of treatments (i.e. averaged over all
years), the primary interest was in identifying significant tem-
poral variation and any interaction between treatment and
Year to highlight temporal trends in response to individual
treatments. Degrees of freedom for Year and interactions with
Year were adjusted to account for departure from homogene-
ity of the population covariance matrix, using e-coefficients
calculated by the Greenhouse-Geisser method (Genstat 5
Committee, 1997).
In order to satisfy the conditions for ANOVA, all percent-
p
age data were transformed before analysis by arcsine( p),
where p is the percentage value expressed as a proportion.
The response, in terms of species-richness in 2005, to the
mean amount of FYM or inorganic equivalent fertilizer ap-
plied at each site averaged over the seven years of the study
was tested by linear regression. Two sets of analysis were car-
ried out for each site, the first restricted to FYM treatments
(Treatments 3–8, i.e. excluding those applied in conjunction
with liming, T13-15) and the second restricted to inorganic
equivalent treatments (T9-12), with nil input control plots in-
1 4 1 ( 2 0 0 8 ) 1 4 1 1 –1 4 2 7 1415
cluded in each case. In both sets, the same units were used for
the explanatory (x) variable, i.e. tonne ha1ha1 year1. A
non-linear response was tested by adding a quadratic func-
tion (x2) to the model and this was retained where its inclu-
sion significantly (P < 0.05) increased the variance accounted
for compared with the linear model. Where there was little
difference in response between the FYM compared with the
inorganic model, as at Bush, an additional analysis was car-
ried out using combined data.
The inter-relationships between all key and additional re-
sponse variables in 2005 were also tested by correlation anal-
ysis in order to amplify the description of the vegetation
provided by the key variables alone.
3. Results
3.1. Effects of FYM, inorganic equivalents and lime
treatments after seven years of application
Analyses revealed significant treatment effects on at least one
key variable in 2005 – all three at Pentwyn (Welsh unim-
proved) – after seven years of fertilizer application (Table 4).
3.1.1. Treatment effects at unimproved sites in 2005
Treatment 5 plots at Raisbeck (Cumbrian unimproved) were
significantly less species-rich in 2005 than those of all other
treatments, with the difference reaching a high level of signif-
icance (P < 0.001) compared with three-yearly FYM treatments
at 12 and 24 tonne ha1 and their inorganic equivalents and
compared with lime-only plots (Table 4). Annual FYM treat-
ments were less species-rich than their inorganic equivalents,
although the difference was significant only between Treat-
ment 5 and its inorganic equivalent (P < 0.01). Treatment ef-
fects in 2005 were less clear-cut in terms of the number per
m2 of MG3 PI species (P = 0.079) at Raisbeck, but the response
was significant (P = 0.017) when these species were expressed
as a proportion of vegetation cover (data not shown). The MG3
PI cover of 28% on Treatment 5 was lower (P < 0.01) than on
the nil input control, where it was 48%. The MG3 PI cover of
51% on the lime-only (Treatment 13) plots was higher
(P < 0.01) compared with lime plus annual FYM application
(Treatment 14) with a cover of 30%. N-scores were higher for
Treatment 5 than all others (P < 0.001), except Treatments 10
and 12 (i.e. the inorganic equivalent to Treatment 5 and the
same rate of inorganic fertilizer applied three-yearly, for both
of which P < 0.01). N-scores were also significantly higher for
liming with annual FYM applications compared with either
liming alone or lime plus three-yearly FYM (Treatment 15)
(both P < 0.05).
The treatment effects described above resulted in signifi-
cant overall effects of rate and frequency of fertilizer applica-
tion within the form · rate · frequency (FRF) factorial subset
of treatments (see Table 2) at this site, both on species-
richness (i.e. high rate < medium rate, P = 0.004, and three-
yearly > annual, P = 0.002) and, mirroring these effects, on
N-scores (i.e. medium rate < high rate, P = 0.001 and three-
yearly < annual, P = 0.008). Form of fertilizer (i.e. FYM or inor-
ganic) had no overall effect on total species-richness in this
analysis, but MG3 PI species were more numerous overall
1416 B I O L O G I CA L C O N S E RVAT I O N 1 4 1 ( 2 0 0 8 ) 1 4 1 1 –1 4 2 7

Table 4 – Effect of treatments on three key variables in 2005Values for positive indicator (PI) species are for MG3 PI species
at Raisbeck and Gaisgill and MG5 PI species at Pentwyn and Bush

Treatment Raisbeck Pentwyn Gaisgill Bush

2 2 2
Species m Species m Species m Species m2
a a
Total PI N-score n Total PI N-score n Total PI N-score n Total PIa N-score n

1. Nil input control 25.6 6.11 4.26 3 23.9 6.11 4.04 6 15.6 0.33 4.75 3 17.0 1.86 4.22 6
2. Continued past inputs = Treatment 4 = Treatment 1 10.8 0.33 5.55 3 = Treatment 1
3. FYM at 6 tonne ha1 26.6 7.10 4.35 3 23.4 5.44 4.25 3 16.0 0.44 4.60 3 17.3 1.30 4.22 3
annual
4. FYM at 12 tonne ha1 25.4 6.44 4.46 6 20.6 5.00 4.38 3 14.6 0.22 4.54 3 16.1 0.94 4.46 3
annual
5. FYM at 24 tonne ha1 21.6 5.42 4.83 3 18.9 4.00 4.46 3 14.0 0.33 4.67 3 15.3 0.83 4.64 3
annual
6. FYM at 6 tonne ha1 26.2 6.58 4.35 3 23.6 6.11 3.96 3 17.2 0.56 4.70 3 17.1 1.41 4.20 3
3-yearly
7. FYM at 12 tonne ha1 28.7 6.76 4.40 3 22.2 5.44 4.16 3 16.6 0.56 4.69 3 16.5 1.94 4.15 3
3-yearly
8. FYM at 24 tonne ha1 27.4 6.63 4.39 3 20.2 5.11 4.31 3 14.4 0.11 4.86 3 17.1 1.39 4.27 3
3-yearly
9. Inorg. equivalent 26.9 7.12 4.40 3 21.8 5.11 4.30 3 15.9 0.33 4.58 3 15.7 1.60 4.22 3
to Tr. 4
10. Inorg. equivalent 26.8 6.83 4.54 3 21.2 4.89 4.43 3 15.9 0.22 4.82 3 14.8 1.62 4.24 3
to Tr. 5
11. Inorg. equivalent 27.1 7.72 4.39 3 25.9 6.11 4.10 3 16.9 0.44 4.40 3 17.3 1.84 4.10 3
to Tr. 7
12. Inorg. equivalent 27.2 7.57 4.53 3 26.2 6.22 4.06 3 15.8 0.11 4.58 3 17.2 1.54 4.19 3
to Tr. 8
13. Lime applied in year 1 27.9 6.87 4.23 3 22.7 4.89 4.23 3 16.1 0.00 4.63 3 20.1 1.94 4.28 3
14. As Tr. 13 + FYM as Tr 4 25.1 6.25 4.45 3 17.3 3.33 4.52 3 15.8 0.44 4.85 3 19.5 1.44 4.53 3
15. As Tr. 13 + FYM as Tr 7 26.8 6.04 4.27 3 23.0 6.11 4.25 3 15.8 0.11 5.02 3 17.9 1.07 4.40 3

SEDs
n=6 v n=3 1.38 0.55 0.076 1.63 0.447 0.125 – – – 1.59 0.395 0.130
n=3 1.59 0.63 0.088 1.88 0.516 0.144 1.14 0.240 0.252 1.83 0.456 0.150
* P = 0.079 *** ** **** ** ** NS * NS NS *

Asterisks denote confidence levels for ANOVA treatments effects: P < 0.05; * **P
< 0.01; P < 0.001; NS, non-significant. Actual values are given for
**

levels between P < 0.1 and P < 0.05. The numbers of values in each mean are indicated by n. Standard errors for the difference between means
(SEDs) are calculated on the basis of 29 residual degrees of freedom (df), except where analyses were adjusted for covariance with 1999 data,
where df = 28. Values in bold type differ significantly (P < 0.05) from nil input control.
a Values adjusted for covariance with 1999 data.

with inorganic fertilizers compared with FYM (P = 0.021).
N-scores showed a significant interaction between form and
frequency of application within the FRF series (P = 0.020),
and a significant three-way interaction between form, rate
and frequency (P = 0.023). The latter effect was heavily attrib-
utable to Treatment 5, which produced significantly higher
scores than all other treatments in the series (P < 0.001, all ex-
cept for Treatments 10 and 12, for which P < 0.01). There was
no significant effect of liming or lime · FYM interaction on
any of the three key variables or the weighted Ellenberg reac-
tion index (R-score) at Raisbeck.
At Pentwyn (Welsh unimproved), Treatment 14 was signif-
icantly different from other treatments in all three key vari-
ables in 2005 (Table 4), being less species-rich than the nil
input controls and three-yearly inorganic fertilizer treatments
(Treatments 11 and 12, all P < 0.001), the lime-only treatment,
the lime plus three-yearly FYM and both annual and three-
yearly FYM at the lowest rate (Treatments 3 and 6, respec-
tively, all P < 0.01), three-yearly FYM at 12 tonne ha1 (Treat-
ment 7) and the inorganic equivalents to annual FYM at 12
and 24 tonne ha1 (i.e. Treatments 9 and 10, P < 0.05). Treat-
ment 14 plots had significantly fewer MG5 PI species per m2
compared with all treatments except Treatment 5 at Pentwyn
(P < 0.001 compared with Treatments 3, 6, 7, 11 and 12, P < 0.01
the remainder). Treatment 5 plots were less species-rich and
with fewer MG5 PI species per m2 than nil input controls
(P < 0.01 and P < 0.001, respectively), than three-yearly inor-
ganic treatments (P < 0.001 for both variables) and less than
lime plus three-yearly FYM (Treatment 15, P < 0.05 for spe-
cies-richness, P < 0.001 for PI species per m2). Differences be-
tween individual FYM treatments and their inorganic
counterparts were not significant for either of these variables,
but in both cases the overall effect of form of fertilizer was
significant within the FRF series (FYM < inorganic, P = 0.004
and P = 0.005 for total species-richness and PI species, respec-
tively). The effect of frequency of application (three-year-
ly > annual) was also significant in each case (P = 0.007 and
P = 0.002, respectively).
The highest mean N-score at Pentwyn was that for Treat-
ment 14, with scores for Treatment 5 and its inorganic equiv-
 B I O L O G I C A L C O N S E RVAT I O N
alent (Treatment 10) marginally lower (Table 4). Differences
compared with nil input controls were significant for each
of these three treatments (P < 0.01, P < 0.001 for Treatment
5), whilst the mean score for nil input control was also lower
(P < 0.05) than those for Treatments 4, 8 and 9. There was no
overall effect of form or rate of fertilizer application within
the FRF series, but scores for three-yearly treatments were
significantly lower overall than those for annual treatments
within both the FRF (P = 0.004) and LFF (P = 0.048) series.
There was no overall effect of liming within the LFF series
on any of the three key variables, but limed plots produced a
significantly higher (P = 0.012) R-score overall than un-limed
ones within this series, i.e. 5.85 compared with 5.72, from
within a range of 5.53–6.05 over all plots.
3.1.2. Treatment effects at semi-improved sites in 2005
At Gaisgill (Cumbrian semi-improved), differences between
Treatment 2 (continuation of past inputs) and all others were
highly significant: P < 0.01 for comparison with annual FYM at
12 tonne ha1 (Treatment 4) and with both annual and three-
yearly FYM at 24 tonne ha1 (Treatments 5 and 8) and
P < 0.001 for comparison with the remainder. Plots receiving
three-yearly fertilizers at the low and medium rates of FYM
(Treatments 6 and 7) and the inorganic equivalent of the latter
(Treatment 11) were the most species-rich in 2005, with sig-
nificant differences between both Treatments 6 and Treat-
ments 11 and 4 (P < 0.05) and between Treatments 6 and 5
(P < 0.01). Plots receiving inorganic treatments tended to be
marginally more species-rich than their FYM counterparts,
and the overall effect of form of fertilizer was significant in
the FRF factorial analysis (P = 0.029). PI species were too
sparse to provide meaningful statistical analysis at Gaisgill,
and the only significant treatment effect shown by N-scores
at this site was the difference between Treatment 2 and all
others (P < 0.01 or greater for all except Treatments 8 and 15,
for which P < 0.05).
There was no statistically significant treatment effect on
either total species-richness or on the richness of MG5 PI spe-
cies at Bush (Welsh semi-improved) in 2005 when all treat-
ments were included (Table 4). However, within the FRF
subset at this site, three-yearly treatments were more spe-
cies-rich than annual treatments averaged over both forms of
fertilizer and application rates, albeit only marginally so
(P = 0.044). Within the LFF series, limed plots were more spe-
cies-rich than un-limed ones, averaged over nil, annual and
three-yearly FYM (P = 0.026). There was a significant (P = 0.024)
lime · FYM interaction within this series for MG5 PI species
per m2, after adjustment for covariance with 1999 data. The
mean number per m2 of these species was lower on un-limed
plots receiving annual FYM application (Treatment 4) than on
all others within the series except Treatment 15 (lime plus
three-yearly FYM) (P < 0.05 compared with both nil input con-
trol and lime plus annual FYM (T14), P < 0.01 the remainder),
and lower on Treatment 15 plots than three-yearly FYM without
lime (T7) or than on lime-only plots (T13) (both P < 0.01).
In contrast to Pentwyn, therefore, liming in combination
with annual FYM application (Treatment 14) was not particu-
larly damaging to species-richness at Bush, nor to the rich-
ness of PI species. However, N-scores were relatively high on
these plots, second only to Treatment 5 (Table 4). Mean
1 4 1 ( 2 0 0 8 ) 1 4 1 1 –1 4 2 7 1417
N-scores for both these treatments were significantly higher
than for most others, including nil input controls (P < 0.01
and P < 0.05, respectively for Treatments 5 and 14 compared
with control). Only Treatments 4, 14 and 15 did not differ from
Treatment 5, whilst Treatment 14 did not differ from these or
from Treatments 8, 10 or 13 (i.e. P > 0.05 for these differences).
Treatment effects on N-score were marginally non-significant
within the LFF series (P = 0.070), although mean scores were
higher for annual FYM treatments than for either nil or
three-yearly FYM, averaged over plus and minus lime. The
overall effect of application frequency was significant
within the FRF treatment series (i.e. three-yearly < annual,
P = 0.020), as was the overall effect of form of fertilizer
(FYM > inorganic, P = 0.029). R-scores showed no overall effect
of lime within the LFF series at this site, but the effect of fre-
quency of FYM application was significant, averaged over plus
and minus lime: annual FYM treatments produced a signifi-
cantly higher mean R-score (5.72) than three-yearly FYM
(5.41) (P = 0.025), with nil FYM, intermediate (5.52).
3.2. Relationship between species-richness in 2005 and
the mean amounts of fertilizer applied 1999–2005
There was a significant relationship at each site between spe-
cies-richness in 2005 and the amount of FYM applied per year,
averaged over 1999–2005, among treatments receiving either
annual or three-yearly or nil FYM without lime, but no corre-
sponding significant relationship was found among inorganic
equivalent treatments (Fig. 1). At Raisbeck, a polynomial func-
tion gave a significantly (P = 0.01) improved fit (adjusted
R2 = 0.37; F2,21 = 7.74; P = 0.003) compared with a linear func-
tion, suggesting a threshold of FYM application above which
there was an increasingly strong negative effect on species-
richness. The equation predicts maximum species-richness
when FYM application averages about 6.5 tonne ha1 year1,
whilst species-richness falls below levels equivalent to nil in-
put controls at about 12-13 tonne ha1 year1. The latter
amount approximates to the amount applied annually to the
remainder of the field at that site (i.e. to the ‘continuation’
treatment), although a slightly lower total amount was applied
during the study period due to the missed application in 2001.
The relationship with FYM application was linear at Gaisgill
(adjusted R2 = 0.20; F1,19 = 5.87; P = 0.026) and at Pentwyn and
Bush (adjusted R2 = 0.26; F1,22 = 8.95; P = 0.007 and adjusted
R2 = 0.15; F1,22 = 5.00; P = 0.036, respectively), with no significant
improvement provided by a polynomial function, although
there was little difference in mean species-richness between
treatments within the 0–5 tonne ha1 year1 range of average
application at any of these three sites (Fig. 1 b–d). The relation-
ship shown for inorganic treatments at Bush was only margin-
ally non-significant (adjusted R2 = 0.13; P = 0.075) and the
regression equation (y = 17.4–0.14x) differed little either from
that shown by FYM treatments (see Fig. 1 d) or from an analysis
of FYM and inorganic treatments combined (i.e. y = 17.5–0.16x,
adjusted R2 = 0.17; F1,34 = 8.22; P = 0.007).
3.3. Relationships between composite variables in 2005
As might be expected, total species-richness and the richness
of MG3 and of MG5 positive indicator species were highly
1418 B I O L O G I CA L C O N S E RVAT I O N 1 4 1 ( 2 0 0 8 ) 1 4 1 1 –1 4 2 7

32 20
Raisbeck FYM Inorganic Gaisgill
30
18
28
Species m-2

Species m-2
26 16
24
22 14
20
y = 25.7+0.41x - 0.03x 2 12 y = 16.5 - 0.14x
18
16 10
0 5 10 15 20 25 0 5 10 15 20 25
-1 -1 -1 -1
Tonnes FYM ha year and inorganic equivalent Tonnes FYM ha year and inorganic equivalent

28 22
Pentwyn Bush
26 20

24 18
Species m-2

Species m-2
22 16

20 14

18 y = 22.40 - 0.19x 12 y = 17.3 -0.15x

16 10
0 5 10 15 20 25 0 5 10 15 20 25
-1 -1 -1 -1
Tonnes FYM ha year and inorganic equivalent Tonnes FYM ha year and inorganic equivalent

Fig. 1 – Relationships between the mean amount of FYM or inorganic equivalent fertilizer applied per year averaged over
1999–2005 and species-richness in 2005 at four hay meadow sites. Each point represents the mean of three replicate plots,
except for the zero input treatment at Pentwyn and Bush (c and d) and 12 tonne FYM ha1 annually at Raisbeck (a) (n = 6). Error
bars are the standard error of the mean in each case. The line or curve shown on each graph was fitted using linear or
polynomial regression analysis of data from FYM-treated plots (without lime) and nil input controls (no significant
relationship was shown with inorganic equivalent treatments).

correlated over the whole 2005 data set (r = 0.90 and r = 0.85,
respectively). Species-rich plots tended to have a high herb:-
grass ratio in vegetation cover (r = 0.65), higher than average
R-scores (r = 0.48) and weighted stress-tolerator radius (Sr-)
score (r = 0.79) and lower than average N-scores (r = 0.32)
and weighted competitor radius (Cr-) score (r = 0.70) (all
P < 0.001). Sr-scores were inversely and highly significantly
(P < 0.001) related to both Cr- and ruderal (Rr-) scores
(r = 0.90 and r = 0.84, respectively), whilst the latter two
variables were positively correlated with each other (r = 0.72,
P < 0.001). High N-scores were closely associated with high
Cr-scores (r = 0.70) and low Sr-scores (r = 0.68). Plots with a
high number and cover of PI species (both MG3 and MG5)
tended to have high Sr-scores and lower than average Cr-
and Rr- scores, with these relationships accounting for 57–
90% of the variation in each case.
Species-richness was positively correlated with higher
than average cover of bryophytes at both semi-improved sites
(P < 0.01 at Gaisgill, P < 0.05 at Bush) with a marked tendency
at Pentwyn (P = 0.053), but not at Raisbeck (r = 0.258,
P = 0.087). N-scores and bryophyte cover were negatively cor-
related at each site, except Raisbeck (P < 0.001 at Pentwyn
and Bush, P < 0.05 at Gaisgill). Legume abundance was associ-
ated with no other variable at Raisbeck, but was associated
positively with bryophyte cover (P < 0.05) and negatively with
grass cover (P < 0.001) at Pentwyn and negatively with both
grass and herb cover at Gaisgill (P < 0.001 and P < 0.001,
respectively) where it showed a positive, though not quite sig-
nificant, association with the herb:grass ratio in vegetation
cover (P = 0.054). At Bush, legume cover was negatively associ-
ated with grass cover (P < 0.001), N-score and weighted Cr-
and Rr-scores (all P < 0.01) and positively with Sr-scores
(P < 0.05).
3.4. Temporal changes 1999–2005 in response to FYM,
inorganic equivalents and lime
Figs. 2–5 show the temporal changes in the three key variables
during the study period that led to the treatment effects in
2005 described in Sections 3.1 and 3.2 above. Figs. 2 and 3
show the changes in these variables at unimproved sites, as
influenced by FYM and equivalent inorganic treatments
(Fig. 2) or liming with or without FYM application (Fig. 3). Cor-
responding data for semi-improved sites are shown in Figs. 4
and 5.
There was a highly significant (P < 0.001) effect of Year on
all variables at all sites, with the exception of MG3 positive
indicator (PI) species at Gaisgill, where, as already noted,
these species were too sparse to provide meaningful statisti-
cal analysis. At Raisbeck clear temporal trends occurred over
the seven years of the study, whereas at Pentwyn significant
inter-year effects on individual treatments tended to obscure
longer term trends. Year · treatment interactions occurred in
the N-score at Raisbeck (F40,96 = 1.55, P = 0.043), the MG5 PI
 B I O L O G I C A L C O N S E RVAT I O N 1 4 1 ( 2 0 0 8 ) 1 4 1 1 –1 4 2 7 1419

Nil input 6 t FYM /ha annual 12 t FYM / ha annual 24 t FYM / ha annual

6 t FYM /ha interm. 12 tFYM / ha interm. 24 t FYM / ha interm. Inor. equiv. 12 t annual
Inor. equiv. 24 t annual Inor. equiv. 12 t interm. Inor. equiv. 24 t interm.

Raisbeck Pentwyn
30
30 a b
28 28

26 26
-2
Species m

24 24

22 22

20 20

18 18
1999 2000 2001 2002 2003 2004 2005 1999 2000 2001 2002 2003 2004 2005
Year Year

9 7
c d
8
-2
Pos. indicator species m

6
7

6 5

5
4
4

3 3
1999 2000 2001 2002 2003 2004 2005 1999 2000 2001 2002 2003 2004 2005
Year Year

5.0 e 5.0 f
4.8 4.8

4.6 4.6
N -score

4.4 4.4

4.2 4.2

4.0 4.0

3.8 3.8

3.6 3.6
1999 2000 2001 2002 2003 2004 2005 1999 2000 2001 2002 2003 2004 2005
Year Year

Fig. 2 – Changes 1999–2005 in total species-richness (a and b), the number of MG3 and MG5 positive indicator species per m2
(c and d, respectively) and weighted British Ellenberg N indices (N-score) (e and f) at Raisbeck (a, c and e) and Pentwyn (b, d
and f) as influenced FYM or inorganic fertilizer application. FYM treatments are represented by squares, inorganic equivalent
treatments by circles, annual treatments by solid lines and three-yearly treatments (applied in 1999, 2002 and 2005) by dotted
lines.

species m2 at Pentwyn (F52,124 = 1.74, P = 0.007) and tended to treatments were consistently less species-rich than their
occur for total species-richness at Bush (F30,109 = 1.32, P = inorganic equivalents from 2000 onwards (Fig. 2a). MG3 PI
0.095). species tended to increase on all treatments between 2000
and 2003 (Figs. 2c and 3c). The N-score tended to show a slight
3.4.1. Changes at unimproved sites overall decline in all except Treatment 5, which showed a
Species-richness tended to increase generally after 2002 at marked increase after 2004 (Figs. 2e and 3e).
Raisbeck on all except Treatment 5, which received At Pentwyn, both Treatment 5 and lime application in
24 tonne FYM ha1 per annum (Figs. 2a and 3a). Annual FYM combination with annual FYM application (Treatment 14)
1420 B I O L O G I CA L C O N S E RVAT I O N 1 4 1 ( 2 0 0 8 ) 1 4 1 1 –1 4 2 7

Nil input Annual FYM - lime Intermittent FYM - lime

Lime only Lime + annual FYM Lime + intermittent FYM

Raisbeck 30
Pentwyn
30
a b
28 28

26 26
-2
Species m

24 24

22 22

20 20

18 18

16 16
1999 2000 2001 2002 2003 2004 2005 1999 2000 2001 2002 2003 2004 2005
Year Year

9 7
c d
-2
Pos. indicator species m

6
7

5
4

3 3
1999 2000 2001 2002 2003 2004 2005 1999 2000 2001 2002 2003 2004 2005
Year Year

5.0
5.0 e f
4.8 4.8

4.6 4.6
N -score

4.4 4.4

4.2 4.2

4.0 4.0

3.8 3.8

3.6 3.6
1999 2000 2001 2002 2003 2004 2005 1999 2000 2001 2002 2003 2004 2005
Year Year

Fig. 3 – Changes 1999–2005 in total species-richness (a and b), the number of MG3 and MG5 positive indicator species per m2
(c and d, respectively) and weighted British Ellenberg N indices (N-score) (e and f) at Raisbeck (a, c and e) and Pentwyn (b, d
and f), as influenced by lime applied in 1999, either alone or combined with annual or three-yearly (1999, 2002 and 2005) FYM
application.

reduced both total species-richness and the number per m2 Gaisgill and Bush (Figs. 4 and 5, respectively). Plots at Gaisgill
of MG5 PI species from 2000 onwards (Fig. 2b, d and Fig. 3b, became more species-rich, increasing from 12.9 to 15.4 spe-
d). N-scores increased in general at this site from 1999 cies per m2 overall between 1999 and 2005 averaged over all
until 2003 and then declined, although considerable changes treatments (P < 0.001). At Bush, the overall trend was down-
in the order of treatments had occurred by 2005 (Figs. 2f ward, though slight, from 17.5 species per m2 to 16.8
and 3f). (P < 0.05). There was also a downward trend on the nil input
control plots at Bush, although, except for a dip in species-
3.4.2. Changes at semi-improved sites richness in 2004, the overall trend was small (from 18.1 down
Trends in each of the three key variables differed markedly to 16.9 species per m2). PI species increased at both sites,
between the Cumbrian and Welsh semi-improved sites, although these were very sparse at Gaisgill. The PI numbers
 B I O L O G I C A L C O N S E RVAT I O N 1 4 1 ( 2 0 0 8 ) 1 4 1 1 –1 4 2 7 1421

Nil input 6 t FYM /ha annual 12 tFYM / ha annual 24 t FYM / ha annual

6 t FYM /ha interm. 12 t FYM / ha interm. 24 tFYM / ha interm. Inor. equiv. 12 t annual
Inor. equiv. 24 t annual Inor. equiv. 12 t interm. Inor. equiv. 24 t interm. Continued past inputs
Gaisgill Bush
22 22
a b
20 20

18 18
-2
Species m

16 16

14 14

12 12

10 10
1999 2000 2001 2002 2003 2004 2005 1999 2000 2001 2002 2003 2004 2005
Year Year

1.0 2.5
c d
-2

0.8 2.0
Pos. indicator species m

0.6 1.5

0.4 1.0

0.2 0.5

0.0 0.0
1999 2000 2001 2002 2003 2004 2005 1999 2000 2001 2002 2003 2004 2005
Year Year

6.0 6.0
e f
5.5 5.5
N -score

5.0 5.0

4.5 4.5

4.0 4.0
1999 2000 2001 2002 2003 2004 2005 1999 2000 2001 2002 2003 2004 2005
Year Year

Fig. 4 – Changes 1999–2005 in total species-richness, the number of MG3 and MG5 positive indicator species per m2 and
weighted British Ellenberg N indices (N-score) at Gaisgill (a, c and e) and Bush (b, d and f) as influenced by FYM or inorganic
fertilizer application. Other details as in caption of Fig. 2.

were relatively low at Bush compared with the unimproved
sites, but the overall increase – from an average of 0.6 per
m2 in 1999 to 1.5 per m2 in 2005 – was highly significant
(P < 0.001) and was largely related to ingress of Rhinanthus min-
or. N-scores declined in general at Gaisgill and, despite a sub-
stantial increase between 1999 and 2002 on Treatment 2 plots
(continued past inputs), the decline averaged over all treat-
ments – from a mean of 5.04 in 1999 to 4.75 in 2005 – was
highly significant (P < 0.001). Nevertheless, this represented
a substantially higher range of values than shown at Bush,
where the overall mean rose from 4.26 in 1999 to a peak of
4.43 in 2003 and then declined to 4.29 in 2005, with both dif-
ferences highly significant (P < 0.001).
Plots of Treatment 2 at Gaisgill were consistently less spe-
cies-rich than those of other treatments from 2000 onwards
(Figs. 4a and 5a). There was no significant effect of liming at
this site, either on species-richness or on N-score. N-scores
were consistently and markedly higher for Treatment 2 than
for other treatments from 2000 onwards, tending to peak in
2002 followed by a slight decline (Figs. 4e and 5e).
1422 B I O L O G I CA L C O N S E RVAT I O N 1 4 1 ( 2 0 0 8 ) 1 4 1 1 –1 4 2 7

Nil input Annual FYM - lime Interm. FYM - lime Lime only
Lime + annual FYM Lime + interm. FYM Continued past inputs

Gaisgill Bush
22 a 22 b
20 20

18 18
-2
Species m

16 16

14 14

12 12

10 10
1999 2000 2001 2002 2003 2004 2005 1999 2000 2001 2002 2003 2004 2005
Year Year

1.0 c 2.5 d
-2

0.8 2.0
Pos. indicator species m

0.6 1.5

0.4 1.0

0.2 0.5

0.0 0.0
1999 2000 2001 2002 2003 2004 2005 1999 2000 2001 2002 2003 2004 2005
Year Year

6.0 6.0
e f
5.8 5.8
5.6 5.6
5.4 5.4
5.2 5.2
N -score

5.0 5.0
4.8 4.8
4.6 4.6
4.4 4.4
4.2 4.2
4.0 4.0
1999 2000 2001 2002 2003 2004 2005 1999 2000 2001 2002 2003 2004 2005
Year Year

Fig. 5 – Changes 1999–2005 in total species-richness, the number of MG3 or MG5 positive indicator species per m2 and
weighted British Ellenberg N indices (N-score) at Gaisgill (a, c and e) and Bush (b, d and f), as influenced by liming with or
without FYM. Other details as in caption of Fig. 3.

4. Discussion 24 tonne FYM ha1 had been applied annually, retained close
affinity to the MG3b (B. media) NVC sub-community (Kirkham
4.1. Plant community characteristics at the experiment et al., 2002; R.A. Sanderson, unpublished data) and supported
sites 25–29 species per m2, which appears to be typical for the MG3
community (C.N.R. Critchley, M.J. Burke and J.A. Fowbert,
The unimproved species-rich meadow communities exam- unpublished report to Defra). Where 24 tonne FYM ha1 had
ined in this study show close affinity to lowland oceanic been applied annually, species-richness declined to less than
and upland meadows in Europe (Rodwell et al., 2007) and 22 species per m2 and the vegetation had developed a close
therefore these results should have a broader applicability affinity to the less species-rich MG3a (Bromus hordeaceus spp.
than to just the UK. During this study at the unimproved up- hordeaceus) sub-community (Rodwell, 1992) by 2005 (R.A. San-
land meadow site at Raisbeck, all treatments, except where derson, unpublished data). Unfertilized plots in the lowland
 B I O L O G I C A L C O N S E RVAT I O N
unimproved meadow at Pentwyn contained 23.9 species per
m2 in 2005, which was in line with an overall average of
23.6 species per m2 in surveys of MG5 grassland (Gibson,
1996, 1997). Annual application of 24 tonne FYM ha1 and lim-
ing in combination with annual application of
12 tonne FYM ha1 had by 2005, reduced species-richness to
18.9 and 17.3 species per m2, respectively, values typical of
poorer examples of MG5 grassland.
Most of the treatments at the lowland semi-improved mea-
dow at Bush, including unfertilized plots, supported levels of
species-richness in 2005 that were fairly typical of MG6 plant
communities at 15–19 species per m2 (Critchley et al.,
2002a,b). At the upland semi-improved meadow at Gaisgill
there were initially 13 species per m2 in 1999, and species-rich-
ness declined under the treatment representing a continua-
tion of past inputs to about 11 species per m2 by 2005. These
values are typical of agriculturally improved grasslands repre-
sented by MG7 NVC communities (Critchley et al., 2002a,b).
The vegetation at Gaisgill was initially close to the MG7e
community (Lolium perenne–Plantago lanceolata grassland) sub-
community (Kirkham et al., 2002; R.A. Sanderson, unpub-
lished data). However, whilst most treatments at Gaisgill
increased in species-richness and became more similar to
MG6 communities, plots receiving continued past inputs
became more similar to the MG7d (Lolium perenne–Alopecurus
pratensis grassland) sub-community (R.A. Sanderson, unpub-
lished data).
Smith et al. (2003) identified target weighted Ellenberg
N-scores for MG3b NVC communities of between 4.33 and
4.67, based upon the floristic tables for this sub-community
(Rodwell, 1992). All treatment means at Raisbeck lay within
these limits in 1999. By 2005 some treatments that repre-
sented a cessation of, or reduction in, nutrient inputs com-
pared with past management had declined below 4.33. Only
the 24 tonne FYM per annum treatment exceeded the 4.33–
4.67 range, at 4.83, in 2005, and by even more in 2003 and
2004. By contrast, the overall mean N-score for Gaisgill in
1999 was 5.04. Although scores declined for all treatments ex-
cept that representing continued past inputs, the majority of
treatments at Gaisgill were still above the target range in 2005,
and even the most species-rich plots were much less species-
rich than would be typical of even the poorest examples of
MG3 grassland (C.N.R. Critchley, M.J. Burke and J.A. Fowbert,
unpublished report to Defra).
Target N-scores are not available for MG5 communities.
The range of values at Pentwyn in 1999 was considerably low-
er than that at Raisbeck, but by 2005 there was considerable
overlap in mean scores between the two sites.
4.2. Frequency and rate of application
Species-richness at each site was significantly inversely re-
lated to the mean amount of FYM applied per year, averaged
over all years of the study, though not to equivalent amounts
of N, P and K applied in inorganic form. The effect of fre-
quency (three-yearly vs. annual) of application on species-
richness was more often significant than the mean amounts
of fertilizer applied per year when averaged over the whole
1999–2005 period. There was, however, little difference in
the mean amount of each nutrient applied with annual treat-
1 4 1 ( 2 0 0 8 ) 1 4 1 1 –1 4 2 7 1423
ments at the low and medium rates compared with three-
yearly treatments at the medium and high rates, respectively
(see Table 2). Despite this, when pair-wise comparisons were
made, among the three key variables and over all sites,
between those three-yearly and annual treatments that pro-
vided similar mean fertilizer amounts per year (e.g. three-
yearly FYM at 24 tonne ha1 compared with annual FYM at
12 tonne ha1), such comparisons favoured three-yearly
treatments much more often than annual treatments. This
indicates that there may be advantages for vegetation quality
in applying treatments intermittently, rather than applying
correspondingly smaller amounts each year. However, an
alternative explanation may lie in the fact that the last instal-
ment of the three-yearly treatments was applied only a
month or so before the 2005 botanical assessments. Whilst
this instalment was included in the calculation of the mean
amounts of fertilizer applied per year, it may not have con-
tributed greatly to the effects of the treatment when assessed
only a month later. Monitoring over a longer period is re-
quired to clarify these effects.
4.3. The significance of the amounts of individual
nutrients supplied
The amounts of N applied (in FYM) were relatively small com-
pared with amounts used in other fertilizer experiments (e.g.
Mountford et al., 1993; Kirkham et al., 1996; Silvertown et al.,
2006), but the proportions of P and K relative to N in FYM were
much higher than in most commercial compound fertilizers
and compared with treatments applied in previous experi-
ments with inorganic fertilizers. Other studies at these sites
(F. Kirkham, J. Tallowin, and R. Sanderson, unpublished data)
suggest that treatment effects may be more closely related to
amounts of P applied than N, except at the Cumbrian semi-
improved site where initial soil P levels were substantially
higher than at the other three sites (see below). This is consis-
tent with the results of an earlier controlled experiment in
species-rich unimproved hay meadows on a Somerset peat
moor, where P was the most influential of the three elements
applied at different rates as inorganic fertilizers (Kirkham
et al., 1996).
Differences in overall site means for N-score in 1999 in
relation to the availability of soil nutrients (see Table 1) pro-
vide further evidence of a relationship between soil P avail-
ability and botanical composition. Mean N-scores for
Pentwyn and Raisbeck in 1999 (4.00 and 4.47, respectively)
and Bush and Gaisgill (4.29 and 5.04, respectively) correspond
more closely with mean amounts of extractable soil P re-
corded in that year than with exchangeable K or with total
soil N. This apparent relationship between soil P and N-score
is, however, accentuated by the high soil P levels at Gaisgill,
albeit that the order of values is the same for both N-score
and soil P among the remaining three sites. The values for soil
P at all the sites except Gaisgill correspond with the lowest
category of P availability recognized by UK ministry guide-
lines, i.e. index 0 (Anon, 2000). Mean values for K availability
at both Cumbrian sites correspond to the target values for
productive grassland, i.e. index 2 – (Anon, 2000), whilst those
at both Welsh sites correspond to K index 1. Levels of soil P at
Gaisgill are equivalent to an index of 3, typical of those found
1424 B I O L O G I CA L C O N S E RVAT I O N
in productive grassland. Whatever the ongoing role that the
relatively high soil P levels might be playing at Gaisgill, these
levels also serve as an indicator of past intensive manage-
ment there, which included the application of substantial
amounts of N, P and K. High soil P levels per se do not there-
fore necessarily provide reliable evidence of a direct causal
relationship between soil P and botanical composition when
making a comparison across sites.
Extensive surveys of permanent grasslands show that
grassland communities and sub-communities of high
botanical value are associated with low soil-extractable P
(Critchley et al., 2002a,b) and that maximum species-rich-
ness is confined to a very narrow band of low soil P avail-
ability (Janssens et al., 1998; Critchley et al., 2002a). Other
surveys have shown that many more endangered species
persist under P-limited than under N-limited conditions
(Wassen et al., 2005). As has been acknowledged (Critchley
et al., 2002a,b; Walker et al., 2004), these surveys cannot
demonstrate a causal link, for the reasons given in the par-
agraph above. Nevertheless, species that are particularly
adapted to low P availability are, characteristically, deep-
rooted herbs and other species with low relative growth
rates (Grime and Hunt, 1975), typically including species
that form associations with micorrhizal fungi which en-
hance their access to soil P resources (van der Heijden
et al., 1998). These properties allow them to co-exist in rel-
atively large numbers in P-limited situations, albeit in a
mainly subordinate position, along with much smaller num-
bers of potentially more competitive species whose domi-
nance is restrained by the nutrient-poor conditions.
Enhancing P supply in such situations would nullify these
advantages, ultimately leading to competitive exclusion of
the subordinate species as a result of increased dominance
by the competitors.
4.4. Differences between FYM and inorganic fertilizers
The tendency for inorganic treatments to be less damaging to
vegetation composition than their FYM counterparts runs
contrary to expectation, if only because of the traditional
use of FYM on unimproved meadows (Simpson and Jefferson,
1996; Crofts and Jefferson, 1999) and the results of other stud-
ies showing damaging effects on the botanical composition of
meadows by low inputs of inorganic N (e.g. Mountford et al.,
1993) The obvious physical differences between the two
forms of fertilizer could have resulted in differential effects
on low-growing species with a prostrate growth habit, or on
bryophytes, for example. However, the nature of the differ-
ences, in particular the differences in N-score, suggests differ-
ential nutrient supply as a more likely explanation. There
may be differences in rate of release of nutrients from the
two forms of fertilizer in the period following application.
Nevertheless, one would normally expect the more ready
availability of nutrients provided by inorganic fertilizers to fa-
vour species with a higher potential relative growth rate, such
as grasses, that would allow them to capitalise on an en-
hanced nutrient supply by making more rapid spring growth
than less competitive species (Grime, 1979). There was, in
fact, no statistically significant overall difference between fer-
tilizer forms in the contribution of grasses to vegetation cover
1 4 1 ( 2 0 0 8 ) 1 4 1 1 –1 4 2 7
at any site, but where such differences were apparent, grass
cover was invariably higher with FYM.
Another possibility is that FYM may have increased the or-
ganic matter content of soils, thereby increasing moisture
retention which could favour more competitive species, par-
ticularly on lighter soils. This might explain why differences
between the two forms of fertilizer were more clear-cut at
the Welsh sites, where soil carbon and organic matter con-
tents were relatively low compared with the Cumbrian sites.
Nevertheless, there was no significant treatment effect on
these soil variables at any site, nor on total soil N with which
both variables were significantly correlated, although there
may have been effects that were too subtle to be identified
within the timescale involved (IGER/ADAS/QUB, 2005).
A further, more likely explanation, is that the amounts of
N and/or P supplied by FYM over the seven year period may
have been underestimated. The MANNER model used to esti-
mate N availability from FYM in this study predicts only the
amount available in the season immediately following appli-
cation (Chambers et al., 1999) and these were the amounts
upon which estimated inorganic equivalent treatments were
based. FYM may enhance the supply of N over a period of
more than one year and with annual FYM applications, for
example, this could lead to a slight underestimation of the
amounts of mineralized organic N available over an extended
period of repeated applications. However, the amounts in-
volved are relatively small, probably leading to underestima-
tion of about 2% of the organic N applied (Williams et al.,
2001). Greater underestimation of P availability may have oc-
curred. It was assumed that 60% of the total P in FYM
was readily available (Anon, 2000), but there is evidence to
suggest that, in the medium-longer term, as much as 100%
of the total P in FYM may become available to plants (Smith
et al., 1998).
If, as suggested by this and other studies, enhanced P sup-
ply is influential in determining vegetation composition in
situations where base levels of soil P are low, differentials of
this order in the amount of P applied in inorganic form com-
pared with FYM could go some way towards explaining the
apparent differences in response between the two forms.
Re-inspection of the graphs shown in Fig. 1 suggests that if
the equivalence between the two forms had been underesti-
mated by as much as 67% (i.e. corrected FYM amount = origi-
nal FYM · 100/60), there would have been little or no
discernible difference in effect on species-richness between
the two forms, except possibly for the highest rate of annual
application at Bush (see Fig. 1d). It is therefore probably safe
to conclude that, at least when applied at rates equivalent
to those shown to be sustainable in this study and over the
timescale encompassed by it, any difference between the
two forms of fertilizer in their effect on botanical composition
is negligible. It should be emphasized, however, that the
amounts of inorganic nutrients applied in these treatments,
particularly the amounts of N, are very much lower than
would normally be used in commercial practice.
4.5. Response to lime
Soil pH was relatively low at all sites in 1999 at 4.9–5.2. The
target pH of 6.0 for liming treatments was exceeded at
 B I O L O G I C A L C O N S E RVAT I O N
Pentwyn and Bush (6.5 and 6.3, respectively in 2002, averaged
over all lime treatments), and slightly under-achieved at Rais-
beck and Gaisgill at 5.7 and 5.9, respectively in 2002 (IGER/
ADAS/QUB, 2005). This may explain the smaller and non-sig-
nificant response to liming at the Cumbrian sites compared
with the sites in Wales. The overall effects of lime alone were
not dramatic at either Welsh site, although the detrimental
effects of lime in conjunction with annual FYM application
were substantial at Pentwyn. The magnitude of this effect
may be attributable to the greater increase in soil pH that oc-
curred in response to liming at Pentwyn compared with other
sites, coupled with a plant community that reflected a long
history of no lime input. It is probably significant that Pent-
wyn was the only site at which there was a significant overall
response to liming in terms of the weighted Ellenberg reac-
tion (R-) score. Liming also appears to have resulted in an in-
creased capacity for nutrient uptake and dry matter response
to annual FYM application at this site, since this combination
was consistently among the most productive treatments both
in terms of hay yield and in terms of the amounts of N and P
present in the hay crop (IGER/ADAS/QUB, 2005). The resultant
changes in botanical composition were characterized not
only by a reduction in species-richness, but also by a signifi-
cant increase in the weighted N-score compared with other
treatments. An increased N-score was also evident with the
same treatment at Bush, but here the effect was much less
pronounced and liming had a positive overall influence on
the species-richness of the vegetation.
4.6. Factors influencing vegetation enhancement at
semi-improved sites
Significant changes in fertilizer regime to the semi-improved
meadows resulted in comparatively little botanical change.
At Gaisgill, the differences that developed between treatments
were attributable both to continued eutrophication where past
fertilizer management was maintained and to small opposing
shifts where inputs ceased or were reduced. Species-richness
increased between 1999 and 2005 with cessation of fertilizer
inputs at this site (from 12 to 15 species per m2), but changed
very little on unfertilized plots at Bush and even declined
slightly there. These differences probably reflect the higher
initial species-richness at Bush, whilst the relatively slow
change even at Gaisgill probably reflected seed and dispersal
limitations within the locality (Bullock et al., 2002). Other stud-
ies have shown that pro-active measures such as the addition
of seeds of desired species, and/or the addition of the
hemi-parasitic species Rhinanthus minor to weaken competi-
tive species, are required to increase the plant diversity of spe-
cies-poor grassland (Smith et al., 2002; Pywell et al., 2004,
2007). The Gaisgill site lay some distance from the nearest spe-
cies-rich meadow and was surrounded by grazed improved
pasture. Ingress of species characteristic of MG3 communities
(e.g. Alchemilla glabra, Conopodium majus and L. pratensis) ac-
counted for some of the modest increases in species-richness
that occurred there, but gains in generalist species such as Vi-
cia cracca, Plantago lanceolata, Cynosurus cristatus and bryo-
phytes were more common. Only a few plants of Rhinanthus
minor were recorded at this site and none before 2005. By con-
trast, the two Welsh meadows were adjacent to each other,
1 4 1 ( 2 0 0 8 ) 1 4 1 1 –1 4 2 7 1425
with other species-rich meadows nearby. R. minor abundance
increased at Bush during the course of the study, from negligi-
ble amounts in 1999 to over 10% mean cover by 2005 on unfer-
tilized plots, but this ingress was apparently inhibited by
annual FYM applied at moderate or high rates and R. minor
cover was reduced by these treatments at Pentwyn where ini-
tial levels were higher (2–8% cover). Careful tailoring of fertil-
izer application may therefore be necessary to allow the
potential benefits of R. minor presence to be realized. More-
over, despite the ingress of R. minor, negligible overall change
occurred in the vegetation at Bush, in agreement with other
work that suggests that relatively high cover of this species
is needed to facilitate the ingress of other species (Smith
et al., 2003; Pywell et al., 2004). This supports a case for artifi-
cial introduction of R. minor seed, preferably as a precursor to
sowing seeds of other desirable species (Pywell et al., 2004).
4.7. Implications for the sustainable management
of semi-natural hay meadows
4.7.1. Maintenance of species-rich meadows
Annual amounts of FYM application higher than those already
in use at Raisbeck (i.e. higher than 12 tonne FYM ha1) may
not be ecologically sustainable within similar MG3 meadows,
and amounts as high as 24 tonne ha1 year1 almost certainly
are not sustainable within meadows comparable to any of
those in this study. Moreover, whilst current inputs to the
MG3 meadow appear to be sustainable at Raisbeck, there
was some indication that a reduction in the frequency or rate
of application might be beneficial, particularly where lime had
recently been applied.
At Pentwyn, FYM inputs no greater than the equivalent of
about 5–6 tonne FYM ha1 per annum with or without lime
appeared to be sustainable for the maintenance of a high nat-
ure value meadow. Liming alone or in conjunction with an-
nual FYM at a higher rate appeared to be unsustainable for
high nature value maintenance at this site. Low levels of
nutrient input at Pentwyn produced hay yields averaging less
than 3 tonne of dry matter ha1(IGER/ADAS/QUB, 2005) and
these levels of production are unlikely to be attractive from
the agronomic point of view. The negative effects of the lime
application and modest inputs of FYM at Pentwyn meadow
may reflect limited genotypic adaptation of a site with no re-
cent history of these inputs. Therefore, caution is needed in
extrapolating the 1999–2005 responses at Pentwyn to MG5
meadows with a more recent history of either FYM or lime
use. Continuation of these treatments and further monitoring
of the changes occurring over time are needed to ascertain
whether adaptation and maintenance of high nature value
will occur under moderate FYM inputs. Intermittent applica-
tions of inorganic fertilizers at the rates used in this study
were apparently more sustainable than applying FYM, both
at Pentwyn and at Raisbeck, although the inorganic amounts
of N and P may have represented lower levels of nutrient in-
put than the FYM counterparts. Furthermore, a final judge-
ment on the acceptability or otherwise of inorganic
fertilizers at any level should await the results of further
monitoring, as well as efforts to identify any effect on soil
microbial populations in the longer term (Smith et al., 2003;
IGER/ADAS/QUB, 2005).
1426 B I O L O G I CA L C O N S E RVAT I O N
4.7.2. Restoration of diversity to agriculturally improved
meadows
Changes occurring at the Cumbrian semi-improved site sug-
gest that, as shown at other species-poor sites dominated by
Lolium perenne (e.g. Hayes and Sackville Hamilton, 2001), some
initial improvement in species-richness can be achieved by
simple cessation or significant reduction in nutrient inputs.
Nevertheless, optimum longer term benefits will probably not
be achieved without some form of pro-active measure aimed
at accelerating gains in species-richness, such as the artificial
introduction of R. minor seeds followed by seeds of other desir-
able species (Smith et al., 2002; Pywell et al., 2004). The high
residual levels of soil P at Gaisgill suggest that initial measures
aimed specifically at reducing soil P may be necessary,
although the extent to which high soil P levels appear to limit
restoration success is very variable (Walker et al., 2004). More-
over, whilst approaches involving the application of substan-
tial amounts of N and K without P, for example, might be
appropriate in the initial stages of restoration (Pywell et al.,
2007), these measures applied at a later stage (as represented
by Gaisgill in 2005) could reverse the gains in species-richness
already made there. Measures aimed at reducing soil fertility
would almost certainly not be worthwhile in more species-rich
semi-improved grassland sites typified by Bush, but seed intro-
duction may well be justified. Applying FYM or other fertilizers
should probably be avoided at such sites, but occasional liming
to maintain pH at about 6.0 would appear to be beneficial.
Acknowledgements
We would like to thank Helen Adamson, Roger Smith, John
Fowbert, Jo Goodyear, Anna Gundrey, Jan Winder, Ken Milner,
Barry Wright, James Towers, Julia Tallowin and Anne Moon
for carrying out botanical assessments, and Helen Adamson,
Jo Goodyear, Gail Bennett and other staff from IGER and ADAS
for applying treatments. Mike Burke and Alison Mole were
responsible for data management. The project was conceived,
supervised and funded by the Department for Environment,
Food and Rural Affairs, English Nature (now Natural England)
and the Countryside Council for Wales, and we would like to
acknowledge the contributions made by Richard Brand-
Hardy, Richard Jefferson, Andrew Cooke, Steve Peel, and Tim
Blackstock. We are grateful to CCW for providing additional
funding specifically for the preparation of this paper. We also
thank Richard Jefferson, Carrie Rimes, Stuart Smith and Tim
Blackstock for comments on the initial draft of the paper,
and Mr. and Mrs. Joe Winder, Tim Winder and the Gwent
Wildlife Trust (Julian Branscombe) for provision of sites and
for cooperation and assistance with field operations.
R E F E R E N C E S
Anon, 2000. Fertiliser Recommendations for Agricultural and
Horticultural Crops (RB 209). Ministry of Agriculture, Fisheries and
Food Reference Book 209, Her Majesty’s Stationery Office, London.
Bakker, J.P., 1989. Nature Management by Grazing and Cutting
Geobotany Series No. 14. Kluwer Academic Publishers,
Dordrecht.
1 4 1 ( 2 0 0 8 ) 1 4 1 1 –1 4 2 7
Berendse, F., Oomes, M.J.M., Altena, H.J., Elberse, W. Th., 1992.
Experiments on the restoration of species-rich meadows in
the Netherlands. Biological Conservation 62, 59–65.
Blackstock, T.H., Rimes, C.A., Stevens, D.P., Jefferson, R.G.,
Robertson, H.J., Mackintosh, J., Hopkins, J.J., 1999. The extent of
semi-natural grassland communities in lowland England and
Wales: a review of conservation surveys 1978–96. Grass and
Forage Science 54, 1–18.
Bullock, J.M., Moy, I.L., Pywell, R., Coulson, S.J., Nolan, A.M.,
Caswell, H., 2002. Plant dispersal and colonisation processes at
local and landscape scales. In: Bullock, J.M., Kenward, R.E.,
Hails, R. (Eds.), Dispersal Ecology. Blackwell Science, Oxford,
pp. 279–302.
Chambers, B.J., Lord, E.I., Nicholson, F.A., Smith, K.A., 1999.
Predicting nitrogen availability and losses following
application of manures to arable land. MANNER. Soil Use and
Management 15, 137–143.
Critchley, C.N.R., Chambers, B.J., Fowbert, J.A., Bhogal, A., Rose,
S.C., Sanderson, R.A., 2002a. Plant species-richness, functional
type and soil properties in grasslands and allied vegetation in
English environmentally sensitive areas. Grass Forage Science
57, 82–92.
Critchley, C.N.R., Chambers, B.J., Fowbert, J.A., Sanderson, R.A.,
Bhogal, A., Rose, S.C., 2002b. Association between lowland
grassland plant communities and soil properties. Biological
Conservation 105, 199–215.
Crofts, A., Jefferson, R.G., 1999. The Lowland Grassland
Management Handbook, second ed. English Nature/The
Wildlife Trusts, Peterborough.
Ellenberg, H., 1988. Vegetation Ecology of Central Europe.
Cambridge University Press, Cambridge.
Fuller, R.M., 1987. The changing extent and conservation interest of
lowland grasslands in England and Wales: a review of grassland
surveys 1930–1984. Biological Conservation 40, 281–300.
Genstat 5 Committee, 1997. Genstat for Windows Command
Language Manual, third ed. Numerical Algorithms Group,
Oxford.
Gibson, C.W.D., 1996. The effects of horse grazing on species-rich
grasslands. English Nature Research Report No. 164. English
Nature, Peterborough.
Gibson, C.W.D., 1997. The effects of horse and cattle grazing on
English species-rich grasslands. English Nature Research
Report No. 210. English Nature, Peterborough.
Goulson, D., Hanley, M.E., Darvill, B., Ellis, J.S., Knight, M.E., 2005.
Causes of rarity in bumblebees. Biological Conservation 122, 1–
8.
Grime, J.P., 1977. Evidence for the existence of three primary
strategies in plants and it’s relevance to ecological and
evolutionary theory. American Naturalist 111, 1169–1194.
Grime, J.P., 1979. Plant Strategies and Vegetation Processes. John
Wiley and Sons, London.
Grime, J.P., Hunt, R., 1975. Relative growth rate: its range and
adaptive significance in a local flora. Journal of Ecology 63,
393–422.
Hayes, M.J., Sackville Hamilton, N.R., 2001. The effect of sward
management on the restoration of species-rich grassland: a
reassessment of IGER’s grassland restoration experiment,
Trawsgoed. Countryside Council for Wales Science Report No.
438, Bangor.
Hill, M.O., Mountford, J.O., Roy, D.B., Bunce, R.G.H, 1999.
Ellenberg’s indicator values for British plants. ECOFACT vol. 2,
Technical Annexe. HMSO, Norwich.
Hodgson, J.G., Grime, J.P., Hunt, R., Thompson, K., 1995. The Electronic
Comparative Plant Ecology. Chapman and Hall, London.
IGER/ADAS/QUB, 2005. The impact of organic fertilizers on semi-
natural grasslands. Final report by ADAS UK Ltd, the Institute
of Grassland and Environmental Research and Queen’s
University, Belfast, on Defra project BD1415. http://www2.
 B I O L O G I C A L C O N S E RVAT I O N
defra.gov.uk/research/project_data/More.asp?I=BD1415&M=
KWS&V=BD1415 & SUBMIT1= Search&SCOPE=0.
Janssens, F., Peeters, A., Tallowin, J.R.B., Bakker, J.P., Bekker, R.M.,
Verweij, G.L., Fillat, F., Oomes, M.J.M., 1998. Relationship
between soil chemical factors and grassland diversity. Plant
and Soil 202, 68–78.
Jefferson, R.G., 2005. The conservation management of upland
hay meadows in Britain: a review. Grass and Forage Science 60,
322–331.
Kirkham, F.W., Mountford, J.O., Wilkins, R.J., 1996. The effects of
nitrogen, potassium and phosphorus addition on the
vegetation of a Somerset peat moor under cutting
management. Journal of Applied Ecology 33, 1013–1029.
Kirkham, F.W., Tallowin, J.R.B., Sanderson, R.A., Hoppé, G.M.,
Goodyear, J., 2002. Role of organic fertilizers in the sustainable
management of semi-natural grasslands, II – Botanical
studies. In: Frame, J. (Ed.), Conservation Pays? BGS Occasional
Symposium No. 36 British Grassland Society, Reading, pp. 153–
157.
Mountford, J.O., Lakhani, K., Kirkham, F.W., 1993. Experimental
assessment of the effects of nitrogen addition under hay
cutting and aftermath grazing on the vegetation of hay
meadows on a Somerset peat moor. Journal of Applied Ecology
30, 321–332.
Myklestad, Å., Sætersdal, M., 2004. The importance of traditional
meadow management techniques for conservation of vascular
plant species richness in Norway. Biological Conservation 118,
133–139.
Ormerod, S.J., 2003. Restoration in applied ecology. Journal of
Ecology 40, 44–50.
Pywell, R.F., Bullock, J.M., Walker, K.J., Coulson, S.J., Gregory, S.J.,
Stevenson, M.J., 2004. Facilitating grassland diversification
using the hemiparasitic plant Rhinanthus minor. Journal of
Applied Ecology 41, 880–887.
Pywell, R.F., Bullock, J.M., Tallowin, J.B., Walker, K.J., Warman, E.A.,
Masters, G., 2007. Enhancing diversity of species-poor
grasslands: an experimental assessment of multiple
constraints. Journal of Applied Ecology 44, 81–95.
Ratcliffe, D.A. (Ed.), 1977. A Nature Conservation Review (Two
volumes). Cambridge University Press, Cambridge.
Ratcliffe, D.A., 1984. Post-mediaeval and recent changes in British
vegetation: the culmination of human influence. New
Phytologist 98, 73–100.
Rich, T.C.G., Woodruff, E.R., 1996. Changes in the vascular plant
floras of England and Scotland between 1930–60 and 1987–88:
the BSBI monitoring scheme. Biological Conservation 75, 217–
229.
Robertson, H.J., Jefferson, R.G., 2000. Monitoring the condition of
lowland grassland SSSIs. I. English Nature’s rapid assessment
method. English Nature Research Report No.315. English
Nature, Peterborough.
Rodwell, J.S. (Ed.), 1992. British Plant Communities. Grasslands
and montane Communities, vol. 3. Cambridge University
Press, Cambridge.
Rodwell, J.S., Morgan, V., Jefferson, R.G., Moss, D., 2007. The
European context of British lowland grasslands. Joint Nature
Conservation Committee Report No. 394. JNCC, Peterborough.
Silvertown, J., Poulton, P., Johnston, E., Edwards, G., Heard, M.,
Biss, P.M., 2006. The Park Grass Experiment 1856–2006: its
contribution to ecology. Journal of Ecology 94, 801–814.
Simpson, N.A., Jefferson, R.G., 1996. Use of farmyard manure on
semi-natural (Meadow) grassland. English Nature Research
Report No. 150. English Nature, Peterborough.
1 4 1 ( 2 0 0 8 ) 1 4 1 1 –1 4 2 7 1427
Smith, R.S., 1988. Farming and conservation of traditional
meadowland in the Pennine Dales environmentally sensitive
area. In: Usher, M.B. Thompson, D.B.A. (Eds.), Ecological
Change in the Uplands. British Ecological Society Special
Publication No. 7. Blackwell Scientific Publications, Oxford pp.
183–199.
Smith, R.S., Jones, L., 1991. The phenology of mesotrophic
grassland in the Pennine Dales, Northern England: historic
hay cutting dates, vegetation variation and plant species
phenologies. Journal of Applied Ecology 28, 42–59.
Smith, R.S., Rushton, S.P., 1994. The effects of grazing
management on the vegetation of mesotrophic (meadow)
grassland in Northern England. Journal of Applied Ecology 31,
13–24.
Smith, R.S., Buckingham, H., Bullard, M.J., Shiel, R.S., Younger, A.,
1996. The conservation management of mesotrophic
(meadow) grassland in northern England. I. Effects of grazing,
cutting date and fertilizer on the vegetation of a traditionally
managed sward. Grass and Forage Science 51, 278–291.
Smith, K.A., Chalmers, A.G., Chambers, B.J., Christie, P., 1998.
Organic manure phosphorus accumulation, mobility and
management. Soil Use and Management 14, 154–159.
Smith, R.S., Shiel, R.S., Millward, D., Corkhill, P., 2000. The
interactive effects of management on the productivity and
plant community structure of an upland meadow: an 8-year
field trial. Journal of Applied Ecology 37, 1029–1043.
Smith, R.S., Shiel, R.S., Millward, D., Corkhill, P., Sanderson, R.A.,
2002. Soil seed banks and the effects of meadow management
on vegetation change in a 10-year meadow field trial. Journal
of Applied Ecology 39, 279–293.
Smith, R.S., Shiel, R.S., Bardgett, R.D., Millward, D., Corkhill, P.,
Rolph, G., Hobbs, P.J., Peacock, S., 2003. Soil microbial
community, fertility, vegetation and diversity as targets in the
restoration management of a meadow grassland. Journal of
Applied Ecology 40, 51–64.
Stace, C.A., 1997. New Flora of the British Isles, second ed.
Cambridge University Press, Cambridge.
Tallowin, J.R.B., 1998. Use and effects of lime application on semi-
natural grasslands in Britain. Countryside Council for Wales
Contract Science Report No. 261. Countryside Council for
Wales, Bangor.
Tallowin, J.R.B., Kirkham, F.W., Goodyear, J., Hoppé, G.M., 2002.
Role of organic fertilizers in the sustainable management of
semi-natural grasslands, I – Agronomic studies. In: Frame, J.
(Ed.), Conservation Pays? BGS Occasional Symposium No. 36
British Grassland Society, Reading, pp. 149–152.
van der Heijden, M.G.A., Klironomos, J.N., Ursic, M., Moutoglis, P.,
Streitwolf-Engel, R., Boller, T., Wiemken, A., Saunders, I.R.,
1998. Mycorrhizal fungal diversity determines plant
biodiversity, ecosystem variability and productivity. Nature
396, 69–72.
Walker, K.J., Stevens, P.A., Stevens, D.P., Mountford, J.O.,
Manchester, S.J., Pywell, R.F., 2004. The restoration and re-
creation of species-rich lowland grassland on land formerly
managed for intensive agriculture in the UK. Biological
Conservation 119, 1–18.
Wassen, M.J., Venterink, H.O., Lapshina, E.D., Tannenberger, F.,
2005. Endangered plants persist under phosphorus limitation.
Nature 437, 547–550.
Williams, J.R., Chambers, B.J., Chadwick, D.R., Bhogal, A., King,
J.A., 2001. Organic nitrogen release from farm manures. In:
Eleventh Nitrogen Workshop, Institut National de la
Recherche Agronomique, Reims, pp. 381–382.