General enquiries on this form should be made to:
General enquiries on this form should be made to:
Defra, Procurements and Contracts Division (Science R&D
Team)
Telephone No.0207 238
5734E-mail:[email protected]
SID 5Research Project Final Report
Note
In line with the Freedom of Information Act 2000, Defra aims to
place the results of its completed research projects in the public
domain wherever possible. The SID 5 (Research Project Final Report)
is designed to capture the information on the results and outputs
of Defra-funded research in a format that is easily publishable
through the Defra website. A SID 5 must be completed for all
projects.
· This form is in Word format and the boxes may be expanded or
reduced, as appropriate.
ACCESS TO INFORMATION
The information collected on this form will be stored
electronically and may be sent to any part of Defra, or to
individual researchers or organisations outside Defra for the
purposes of reviewing the project. Defra may also disclose the
information to any outside organisation acting as an agent
authorised by Defra to process final research reports on its
behalf. Defra intends to publish this form on its website, unless
there are strong reasons not to, which fully comply with exemptions
under the Environmental Information Regulations or the Freedom of
Information Act 2000.
Defra may be required to release information, including personal
data and commercial information, on request under the Environmental
Information Regulations or the Freedom of Information Act 2000.
However, Defra will not permit any unwarranted breach of
confidentiality or act in contravention of its obligations under
the Data Protection Act 1998. Defra or its appointed agents may use
the name, address or other details on your form to contact you in
connection with occasional customer research aimed at improving the
processes through which Defra works with its contractors.
Project identification
1.Defra Project code
BD1454
2.Project title
Modified management of agricultural grassland to promote
in-field structural heterogeneity, invertebrates and bird
populations in pastoral landscapes
3.Contractororganisation(s)
Royal Society for the Protection of Birds
The Lodge
Sandy
Beds
SG19 2DL
54.Total Defra project costs
£799,882
(agreed fixed price)
5.Project:start date
01 December 2005
end date
30 November 2009
6.It is Defra’s intention to publish this form.
Please confirm your agreement to do so.YES FORMCHECKBOX NO
FORMCHECKBOX
(a)When preparing SID 5s contractors should bear in mind that
Defra intends that they be made public. They should be written in a
clear and concise manner and represent a full account of the
research project which someone not closely associated with the
project can follow.
Defra recognises that in a small minority of cases there may be
information, such as intellectual property or commercially
confidential data, used in or generated by the research project,
which should not be disclosed. In these cases, such information
should be detailed in a separate annex (not to be published) so
that the SID 5 can be placed in the public domain. Where it is
impossible to complete the Final Report without including
references to any sensitive or confidential data, the information
should be included and section (b) completed. NB: only in
exceptional circumstances will Defra expect contractors to give a
"No" answer.
In all cases, reasons for withholding information must be fully
in line with exemptions under the Environmental Information
Regulations or the Freedom of Information Act 2000.
(b)If you have answered NO, please explain why the Final report
should not be released into public domain
Executive Summary
7.The executive summary must not exceed 2 sides in total of A4
and should be understandable to the intelligent non-scientist. It
should cover the main objectives, methods and findings of the
research, together with any other significant events and options
for new work.
1. The UK Government is committed to reversing large scale
declines in farmland bird populations by 2020. A major obstacle is
the provision of invertebrate-rich habitats to allow farmland birds
to successfully raise chicks. This is particularly true of
livestock-rearing areas dominated by intensively managed grassland,
which lacks invertebrate resources especially larger-bodied groups
like grasshoppers and Lepidoptera which are important dietary
components for farmland birds.
2. Complete and frequent defoliations of fast-growing, uniform,
fertilized swards characterise the modern management of
agricultural grassland. These conditions are generally hostile to a
broad range of grassland invertebrates, which require a
heterogeneous sward structure and composition (providing a variety
of niches) and vegetation left in situ (allowing life cycles to be
completed). Frequent, low height mowing of grassland may disrupt
the breeding of species like skylark which nest at high densities
in grass silage fields.
3. The general aim of this study was to use simple management
techniques to enhance invertebrate communities and invertebrate
prey availability for farmland birds on grazed grassland. A further
aim was to enhance skylark nesting success in grass silage fields.
Specific objectives were: (1) to investigate the effects of leaving
grass in situ over winter and of reducing summer grazing intensity
on sward structure, invertebrate abundance and availability to
birds; (2) to assess whether the magnitude of any invertebrate
response was large enough to influence reproductive success of
priority birds; (3) to measure the agronomic and economic costs of
the grazing treatments; (4) to assess the benefits to nesting
skylarks and agronomic costs of raised cutting heights on grass
silage fields, and (5) to recommend combinations of grassland
management options suitable for inclusion in Defra’s Environmental
Stewardship Scheme.
4. Different combinations of grazing treatments were established
for 2-4 years on a split-plot design across twenty-one grass fields
in South Devon (ranging from semi-improved to improved grassland),
and across four improved fields in Herefordshire and North
Yorkshire. Sites in South Devon were in localities known to support
populations of priority farmland birds (buntings, finches and
skylarks). Different combinations of two extensification measures
were tested: reduced intensity (lenient) early season grazing (to a
surface sward height of 12-15cm compared to moderate grazing sward
height of 7-9cm) and the early cessation (early closure) of grazing
whereby livestock were excluded from trial paddocks from mid-July
until the following spring. Lenient grazing was predicted to
enhance structural heterogeneity and invertebrate prey abundance
while providing access to foraging birds. Early closure was
predicted to enhance densities of two food resources that are
scarce in agricultural grasslands: seeds (by allowing plants to
reproduce) and large-bodied invertebrates in spring (by promoting
their overwinter survival). Loss of beef cattle live weight output
was used as a measure of agronomic cost.
5. The impact of raised cutting heights and delayed mowing on
skylark reproductive success was tested on thirty multiple-cut
silage fields in Dorset over three years, while the impact of safe
nesting plots (areas of grass left undisturbed after a first cut)
was tested on seven fields during one year. The impacts on
reproductive success of different combinations of high and low
cuts, different intervals between cuts and of using different
mowing and collection machinery were investigated using stochastic
simulation modelling.
6. Lenient early season grazing was associated with large and
immediate increases in invertebrate abundance, including those
invertebrate groups that are important in bird diets (abundance on
semi-improved lenient plots during July was 34%, 48%, 78% and 65%
higher than on moderately grazed plots during 2006-2009
respectively). This beneficial impact of lenient grazing was
evident for all invertebrate size-classes, and for groups that are
important in farmland bird diets such as grasshoppers, ground
beetles, rove beetles and Dipteran flies, as well as leafhoppers,
true bugs and springtails. Similar impacts of lenient grazing were
evident on improved sites.
7. Early closure of semi-improved grazed grassland had smaller
and delayed impacts on the abundance of invertebrate groups
important in bird diets (abundance on moderately-grazed plots
during July was 12%, 17%, 31% and 35% higher than on control plots
during 2006-2009). Impacts of early closure on invertebrate
abundance on improved fields were larger and also increased over
the course of the study (equivalent differences in the abundance of
bird prey taxa were 21%, 38%, 61% 40% during 2006-2009). These
impacts of early closure were associated mainly with small and
medium-sized invertebrates. Early closure had significant positive
impacts on the abundance of rove beetles, leafhoppers and true
bugs. With the exception of all beetles and rove beetles, early
closure did not enhance invertebrate abundance during April. Early
closure was associated with a reduction in plant species richness
on semi-improved fields, but not on improved fields.
8. Foraging priority birds mainly used extensification plots
during the breeding season. Skylarks responded strongly and
positively to both lenient early season grazing and to early
closure. Buntings (yellowhammer and cirl bunting) and the wider
group of mixed diet granivores (buntings, sparrows and finches)
strongly selected early closure plots, but did not respond to
lenient grazing. Obligate granivores (goldfinch & linnet)
showed no strong selection of either extensification treatment.
Measurements of fine-scale sward structure at bunting foraging
sites indicated that lenient grazing reduced structural
heterogeneity on early-closure swards below the levels required by
buntings, thereby restricting their access to the enhanced prey
resources. Models of bunting responses to sward condition implied
that an intermediate grazing pressure (mean sward height ca.
9-12cm) could optimise the trade-off between prey density and
accessibility. The primary requirement of the obligate granivores
appeared to be the seed of particular forbs (notably Taraxacum and
Leontodon). These plants occurred on too few fields to explore the
effects of the extensification treatments. During winter, meadow
pipits and, to a lesser extent, yellowhammers preferred plots
subjected to early closure.
9. The agronomic costs of lenient early season grazing (relative
to moderate early season grazing) on semi-improved grassland were
2%, 25% and 32% during 2007-2009 respectively. The costs of early
closure were 31%, 51% and 43% during 2007-2009 respectively, while
the combined costs of lenient grazing plus early closure were 32%,
63% and 62%. Proportional costs on improved fields were similar in
Yorkshire but lower in Herefordshire. The high costs of early
closure were attributable both to the loss of late season grazing
and to deleterious changes in sward composition (increased cover of
Ranunculus repens, Holcus lanatus, Agrostis species and litter;
reduced cover of Lolium perenne and Trifolium repens). These
changes in sward composition were most pronounced in leniently
grazed plots in 2009, when there was a significant decline in
forage quality (metabolisable energy).
10. On semi-improved sites, there was no evidence that cover of
injurious weeds was promoted by lenient grazing or by early
closure. There were indications that the early closure treatments
had some value in controlling Cirsium vulgare infestations,
reducing thistle cover over the last two years of the study. On
improved sites, there was evidence of increased cover of nettle and
creeping thistle after four years of lenient grazing, but the cover
of each weed species remained below 5% on all sites.
11. Reproductive success of skylarks under normal low-cut
(6-8cm) silage management is very poor (only 4-8 fledglings raised
per 100 skylark pairs), mainly due to high rates of egg and chick
mortality soon after mowing caused mainly by abandonment (after
having been covered in cut grass), predation and trafficking
(crushed under vehicle wheels). Raised mowing heights (8-10cm)
increased daily survival rates of nests and chicks, increased
productivity of successful nests and promoted earlier initiation of
replacement nests.
12. For a two-cut silage system, reproductive success was
substantially enhanced (ca. 21 fledglings / 100 pairs) by imposing
a raised cutting height at the first cut, and a normal low cutting
height at the second cut. Delaying the second cut by 7 and 14 days
further enhanced reproductive success (to 36 and 47 fledglings /
100 pairs respectively). Raising cutting height at the first cut
(only) increased (by 9%) total dry matter yield from two successive
cuts, while delaying the second cuts by 7 days had no significant
impact on silage yield or quality in the one year it was measured.
Substantial (9-fold) increases in skylark reproductive success are
therefore possible at no or little agronomic cost. A single late
haylage cut had the potential to raise reproductive success further
(e.g. 94 fledglings / 100 pairs for a late July cut), although this
is still probably insufficient to maintain stable breeding
populations.
13. Lenient grazing of agricultural grassland (to a sward height
of 12-15cm) has been shown to have large and immediate beneficial
impacts on invertebrate abundance and utility to skylarks at
agronomic costs that may be affordable within agri-environment
schemes. Early exclusion of cattle had fewer biodiversity benefits
but substantial agronomic costs and should probably not be
considered as a future agri-environment option. Continuous moderate
grazing (to a sward height 9-12cm) would have a smaller positive
impact on invertebrate abundance but produce swards that are more
accessible to a wider range of priority farmland birds particularly
buntings. However, it remains to be tested whether these benefits
would be realised in the absence of early closure.
14. Raising mowing heights (to 8-10cm) at the first silage cut
and delaying the second cut by one week has the potential to
substantially raise skylark reproductive success in grass silage
fields at negligible agronomic cost. This combination of silaging
options could be considered as the basis for a future option within
the Entry Level Scheme.
15. We highlight four key gaps in current agri-environment
provision for farmland birds on English livestock systems: (1)
lenient grazing options tailored to the needs of different bird
groups, (2) optimised combinations of rest periods and lenient
grazing for field margins/corners, (3) a means of providing
abundant seed during winter from grassland and (4) a means of
promoting the breeding success of in-field nesting species on
intensive silage fields. Solutions to the third and fourth
requirements have been developed (here and in BD1455) although
agronomic costs need clarification. Further work is needed to
optimise lenient grazing measures for different bird groups and to
develop rest-graze combinations for field margins.
Project Report to Defra
8.As a guide this report should be no longer than 20 sides of
A4. This report is to provide Defra with details of the outputs of
the research project for internal purposes; to meet the terms of
the contract; and to allow Defra to publish details of the outputs
to meet Environmental Information Regulation or Freedom of
Information obligations. This short report to Defra does not
preclude contractors from also seeking to publish a full, formal
scientific report/paper in an appropriate scientific or other
journal/publication. Indeed, Defra actively encourages such
publications as part of the contract terms. The report to Defra
should include:
the scientific objectives as set out in the contract;
the extent to which the objectives set out in the contract have
been met;
details of methods used and the results obtained, including
statistical analysis (if appropriate);
a discussion of the results and their reliability;
the main implications of the findings;
possible future work; and
any action resulting from the research (e.g. IP, Knowledge
Transfer).
1. OBJECTIVES.
The general aim of this project was to trial modified grassland
management techniques that aimed to enhance farmland bird
populations by increasing the abundance (and availability) of their
main invertebrate prey and by providing safe nesting sites on
lowland agricultural grassland. Specific objectives were:
(1) To quantify the extent to which in-field heterogeneity and
grassland invertebrate prey for farmland birds can be improved on
intensive and semi-improved grassland by modifying grazing
management at key periods of the year, and specifically: (a) To
investigate the effects of leaving grass in situ over winter and
limiting summer grazing and fertiliser inputs.(b) To evaluate
whether manipulating grazing intensity can produce the preferred
foraging sward structures for priority farmland birds. (c) To
assess whether the magnitude of any invertebrate responses to
modified grazing management is sufficient to influence the
reproductive performance of priority farmland birds.(d) To evaluate
the agronomic and economic impacts of various grazing options to
increase the heterogeneity and biodiversity value of improved
grassland.
(2) To evaluate whether raised cutting heights on grass silage
fields can increase the nesting success of skylarks to the level
needed for sustainable populations, and quantify the agronomic
costs.(3) To recommend combinations of practical and costed
measures suitable for inclusion in Defra’s Environmental
Stewardship (ES) Scheme that will lead to increased densities of
invertebrates & birds on agricultural grassland.
All objectives were met except objective 1c (see section
4.2.4).
2. METHODS.
2.1. Experiment 1: Extensive grazing on semi-improved grassland
with over-winter refugia for invertebrates
Treatments were imposed on 13 permanent grass fields with a
history of grazing by beef cattle and inorganic fertilizer inputs
of less than 50 kg N/ha/year. The fields were widely spread across
fourteen livestock farms (mainly beef and mixed beef) in South and
East Devon (from Seaton in the east to Holbeton in the west) in
localities known (from bird surveys conducted in 2003) to hold
target farmland birds including buntings, larks, finches, sparrows
and thrushes. Areas of trial fields averaged 3.4ha (range 2.76-4.79
ha) and were selected to allow the imposition of three split-plot
paddocks of equal area (approximately 1ha each) and having similar
aspect, slope and boundary characteristics. Two conservation
measures – lenient grazing pressure and early closure - were tested
in a partially crossed, split-plot experimental design. Lenient
grazing pressure was expected to simultaneously promote densities
of foliar invertebrate and seeds and to allow cattle to graze more
selectively, enhancing the structural heterogeneity of the swards
and providing access for foraging birds. Early closure was expected
to enhance densities of invertebrate prey (by promoting the
survival of taxa that overwinter in sward or litter layer) and
seeds (by allowing plants to reproduce). On each field, the
following three treatment combinations were imposed at random
during the summer of 2006 and maintained until September 2009:
Treatment 1: Moderate grazing with early closure. Beef cattle
grazed plots to a target sward height (TSH) of 6-9cm between April
and mid-July, after which all livestock were excluded until spring
April of the following year. This treatment was designed to
increase invertebrate densities and sward heterogeneity
(overwinter, through early closure) and to then provide short
swards in spring, suitable for soil- and surface-feeding birds like
starlings, thrushes and wagtails.
Treatment 2: Lenient grazing with early closure. Beef cattle
grazed plots to a TSH of 12-16cm (10-14cm in 2006 only) between
April and mid-July, after which all livestock were excluded until
April of the following year. This treatment was designed to
increase foliar invertebrate and seed densities and sward
heterogeneity (to promote accessibility) thereby providing suitable
spring and summer foraging conditions for sward-feeding birds like
buntings, larks, sparrows and finches.
Control: Moderate grazing without early closure. Beef cattle
grazed plots to a target sward height of 6-9cm between April and
October (or when grass growth ceased). Under this treatment,
continuous defoliation was expected to prevent grasses and forbs
from flowering and setting-seed and to prevent useful invertebrate
prey resources from developing.
Usage of inorganic fertilizer was prohibited under all three
treatments. A put-and-take grazing system was employed to achieve
TSHs, with plots being rested when swards fell to the lower TSH
limit. During grazing periods, sward height was monitored
approximately weekly by project staff, who then asked farmers to
move animals where necessary. Sward height was measured using a
HFRO sward stick which is the height at which a descending Perspex
window first comes into contact with live foliage excluding
inflorescences and their supporting stems. Grazing control was
based on the mean sward height from 30-40 HFRO measurements
collected from a W-shaped transect across each treatment plot.
The timing of plot closure was staggered across the 14 study
sites to ensure that sward and invertebrate sampling took place
within a few days of sward closure. As this sampling took
approximately two days at each site, the first site was typically
closed in early July and the last in early August (Appendix A).
During March, two passes of a chain harrow were used to dislodge
and partially remove accumulated dead vegetation and litter from
the moderate and lenient treatment plots with the aim of opening
swards up and promoting new grass growth. Two sites (one Experiment
1 site and one Devon extension site; Appendix A) were abandoned
during the autumn of 2008 due to persistent serious management
difficulties.
2.1.1. Agronomic measures.
Grazing days: Throughout the season a record was kept by farmers
at all sites of the number of cattle grazing each paddock. These
data were accumulated for the early and late season periods to
provide grazing day totals for each treatment at each site in each
year. Sites that were unable to provide reliable data or did not
achieve TSH were excluded from analyses (Appendix A).
Cattle weights and animal performance: On six core sites, animal
live weight was monitored at three key dates; turnout (April),
early closure (mid-July) and end of the grazing season
(September/October) during 2007-2009. Individual cattle growth
rates for early and late season periods were calculated by weighing
core cattle on three sites, and estimating (by expert eye) cattle
live weights on the other three sites. Estimated weights were
highly correlated with actual weights measured on the same animals
on the same day (R2 = 91%, 95% and 95% for 2007-2009 respectively;
Appendix B), which suggests that any measurement error was small.
Rates of animal weight gain (kg/day) and season-long livestock
output (kg/ha) were calculated for each treatment paddock at each
core site in each year.
Herbage quality: On the six core sites, herbage pluck samples
were collected three times (May, mid-July and September (control
paddocks only)) during each season (2007-2009) from three locations
within each paddock that were being actively grazed by cattle.
Three bulked pluck samples were gathered from each paddock on each
occasion and analysed for crude protein (nitrogen), modified acid
detergent fibre (MAD fibre) and soluble carbohydrate (WSC). From
the MAD fibre result, D-value and metabolisable energy (ME) values
were derived for the purpose of estimating feed quality.
Soil sampling: Soil nutrient status was assessed for whole
fields on core sites during April of 2006 or 2007, and for all
treatment paddocks on all sites during April 2009. Soil cores were
collected to a depth of 7.5cm along W-shaped transects and bulked
into a single sample prior to laboratory analysis. Bulked samples
were analysed for pH, available P, K, Mg, ammonium N, nitrate N,
total N and soil mineral nitrogen (SMN).
Statistical analysis: Herbage quality data were analysed with
four-way analysis of variance (ANOVA) with site, year, month and
treatment as main factor, fitting all possible interactions. Animal
weight gain was analysed using three-way ANOVA with main factors
site, year and treatment and fitting all possible interactions.
Turnout weight was included as a covariate. Livestock output was
analysed using a repeated measures generalised linear model (GLM)
in which plot was declared as the subject which was measured
repeatedly across years. Cattle breed (5 levels), year and
treatment were main factors and a year* treatment interaction was
incorporated. For both sets of animal performance data, separate
analyses were conducted for early season output (all three
treatments), late season output (controls only) and a comparison of
early and late season output (controls only). Where necessary, raw
data were logarithm or square root transformed prior to analysis,
and normal errors were assumed.
2.1.2. Litter measurements.
In order to monitor any accumulation of dead litter and to
assess the effectiveness of the chain harrowing in dislodging and
removing accumulated litter, we measured litter on all treatment 1
and 2 plots during March 2007-2009 and on control plots in 2008 and
2009. The method aimed to measure the vertical distribution of
litter within the sward and in 2007 and 2008 was conducted within a
period two days before harrowing and repeated approximately 7 days
later in order to assess the impact of this restoration technique.
In 2009, litter measurements were only collected before harrowing
was conducted.
On each sampling occasion, the vertical distribution of litter
within the sward was measured at 40 random locations spread across
each trial paddock. Litter (attached or detached dead plant
material, excluding any brown tips of green leaves) was measured
using a modified sward stick, by recording the presence/absence of
dead litter (or stems) within sample cells of 5cm horizontal radius
in each 5cm height stratum forming a vertical cylindrical column
centred on the sward stick. Live growth was also measured at each
point using the standard HFRO sward stick method. Statistical
analyses tested for evidence of litter accumulation over the course
of the study and for effects of harrowing on the vertical
distribution of litter.
2.1.3. Swards and invertebrates.
Sample positions: On each treatment plot, four 10m-spaced,
parallel sampling transects were established to assess within-field
variation for both invertebrate and sward assessments. Sampling
points were located at five regularly-spaced points along each
transect, yielding 20 samples per plot spaced at 3, 20, 40, 60 and
80m from the field boundary edge.
Sward structure measurements: Within each treatment paddock, 10
HFRO sward stick measurements were taken within a 3m-radius circle
at each of the 20 sample points. Height was recorded from all plant
material, excluding stems and inflorescences above the top leaf.
The height of each sward stick measurement was taken from
undisturbed areas within each circle. Sampling occurred in July of
all 4 years and in April / May of 2007 and 2009. Sward
heterogeneity at each of the sampling points was expressed as the
coefficient of variation (CV) of the HFRO measures. A higher CV is
indicative of structurally more heterogeneous vegetation than a low
CV.
Drop disc measurements: Three drop disc measurements (drop disc
20 cm in diameter) were taken from undisturbed areas within each
3m-radius circle at each of the 20 sample points within each
treatment plot in July 2006 and 2009. Compressed sward height and
percent cover of functional groups under the disc were
recorded.
Vegetation quadrats: Percent cover was recorded in four fixed
position 1x1m quadrats in each treatment plot, located along the
central two parallel transects, in July of all four years.
Vegetation composition was recorded as percent cover of functional
groups (e.g. grass, forbs, agricultural clovers, bare ground, dung
and leaf litter, defined as detached vegetation) and the individual
percent cover of all higher plant species.
Injurious weeds assessment: This was additional to the original
proposal in order to assess the extent of injurious weeds within
the experimental paddocks and the effects of the treatments
thereon. Most of the weeds recorded are prescribed under the 1959
Weeds Act (e.g. spear thistle (Cirsium vulgare), creeping thistle
(Cirsium arvense), broadleaved dock (Rumex obtusifolius) and
additionally nettle (Urtica dioica)). Within a 10m radius of each
botanical quadrat position (approx. 314 m2), the percent cover was
estimated. Recordings took place during July 2007 to 2009.
Flower head counts: During spring and summer 2009, flower head
assessments were made in order to assess the suitability of the
treatments for granivorous birds. Within a 3m radius around each of
the 20 sampling points the number of flowering heads of all species
was counted.
Seed head counts: Seed head counts of grasses and forbs were
undertaken during September/October of 2007 and 2009 on the six
core sites. The number of seed heads, identified to genus, were
recorded at 20 points in each paddock along a W-shaped transect
using a 0.25m x 0.25m quadrat.
Sweep net sampling: Within each treatment plot, 20 double sweeps
were taken at all 20 sample positions. These were taken from the
edge of the 3m-radius sampling area to avoid any disturbance of the
vegetation. Sampling occurred in July of all 4 years and in April
of 2007 & 2009 on the same day as other measurements. The
abundance of all Orthoptera, larvae of Symphyta and Lepidoptera
were recorded and the Orthoptera identified to species.
Vortis sampling: Invertebrate densities were measured at 20
sample points in each treatment using a Vortis suction sampler
(Burkard Manufacturing Co. Ltd, UK). A single sample consisted of
10 sub-sample positions: each position was vacuum-sampled for 15
seconds and the sub samples combined (total area sampled 0.174m2).
These were taken from within the 3m circles around the centre of
each sampling position. The sampling provided measures of the total
number of invertebrates (>2mm), and those groups identified by
RSPB as potential bird food (e.g. beetles, beetle larvae, spiders,
true bugs, flies, Orthoptera and Lepidoptera larvae, Wilson et al.
1996). These invertebrates were mainly identified to orders and
sub-orders, but the following were identified to species: Carabidae
(ground beetles), Chrysomelidae (leaf beetles), Curculionidea
(weevils), Orthoptera (grasshoppers), and Auchenorrhyncha
(leafhoppers). Invertebrates were assigned to three body size
classes (2-5mm, 5-10mm and >10mm). Sampling occurred in July of
all 4 years and in April of 2007 & 2009.
Pitfall traps: Pitfall traps were used to sample the
invertebrates for two weeks in the summers of 2007 and 2009. Twenty
pitfall traps were set per treatment plot in clusters of four,
located along the central transect line and positioned to avoid
overlap with botanical quadrats. Polyethene pots with a diameter of
6.8cm and a saturated salt solution as a preservative were used.
Carabidae were identified to species and the abundance of all other
invertebrates (>2mm) was recorded.
Statistical analyses were performed using the Community-Ecology
package “vegan” in R (R Core Development Team 2009) and Genstat 11
(Payne, Murray et al. 2007). The effects of the treatments over
time on sward height, sward structure (Coefficient of Variation of
the HFRO measurements) and invertebrate abundances were described
using a linear mixed-effects model with treatment as fixed factor
and farm as a random factor. The experiment was set-up as a nested
randomized-block design, e.g. distance nested in treatment and
farm. Target sward height duration (the number of days between
achieving target sward height and plot closure to grazing) was used
as a covariate in all multivariate analyses. The effects of the
grazing treatments on botanical composition were analysed using
Redundancy Analysis (RDA). In all cases, the data were
log–transformed prior to analyses. Changes in treatment effects on
botanical composition over the four years of the study were
assessed using principal response curves (PRC). These show temporal
differences in community composition relative to the control
treatment (the latter indicated by a horizontal line through y=0)
and the analysis effectively tests the treatment*time interaction.
The abundance of injurious weeds was analysed using Kruskal-Wallis
one-way analysis of variance as the data were highly skewed.
The effects of the grazing treatments on invertebrate community
composition were assessed using the same multivariate techniques as
those used for the vegetation data. However, changes in vegetation
composition are likely to be the cause of differences in
invertebrate assemblages among the treatments. We therefore tested
for relationships between individual components of the vegetation
and the abundance of invertebrate groups or species using RDA,
where invertebrate abundance was the response variable and
vegetation composition the explanatory variable. Injurious weeds
were analysed using Kruskal-Wallis one-way analysis of variance as
the data were highly skewed.
2.1.4 Plot- and patch-scale usage of trial swards by birds Bird
usage of trial plots was measured as the number of foraging visits
to each plot by each species during timed summer watches (conducted
during May-September) each lasting 45 minutes. Each survey
consisted of three 45-minute watches, one in each plot. The surveys
took place in good weather, during the morning and mid-late
afternoon, when foraging activity peaked. The number of watches
undertaken at each site varied and in most years, and watches did
not start until target sward heights had been attained. These
variations in survey effort were controlled for by ‘year’ and
‘closure period’ (pre or post-closure) explanatory variables in the
analyses. Bird usage of Experiment 1 plots was also assessed during
the winter of 2007-08.
We first consider the empirical responses of birds during the
breeding season to factors under the direct control of the land
manager: namely the target sward height and time of closure. The
aim of the analysis is to predict how management can be optimised
for key species and groups of conservation concern particularly
skylarks, buntings and the obligate seedeater guild. In addition to
plot-scale treatment analyses, we consider influences of potential
covariates also measured at the plot scale. These include
invertebrate abundance and seed / flower densities measured soon
after plot closure in July.
For the treatment models, GLMs were fitted to the count data
with a log link and either a Poisson error structure (corrected for
over-dispersion) or a negative binomial structure (for the more
numerous species/guilds, where dispersion factors exceeded c.2).
Where possible, a site identifier factor was forced into the models
as a fixed effect (most species/guilds occurred on too few sites to
model the site effect as a random factor). Covariance structures
were omitted from the models, as correlations between the sparse
usage data were too weak to improve model performance. In any one
year, a site can only contribute information on treatment
preferences if birds occurred on it during that year. Site-years
where the species (or guild) of interest did not occur were
excluded from the treatments models. Also, site-years where the
full set of treatments was not implemented correctly were excluded
from the analysis. The effects of year, closure period (pre- or
post-closure) and the duration of TSH delivery were also tested in
the treatment models. In each year, TSH duration was quantified as
the number of days between achieving TSH and the early closure
date. TSH duration was calculated at the plot level (number of days
at TSH on each plot) and the site level (number of days when all
three treatment plots were at TSH). Data from Experiment 1 and
Devon extension sites were analysed separately. The Reddaway site
held the best numbers of several key species in the Devon Extension
experiment (skylark, buntings), but cattle rejected a large area of
grass on the control plot, which resembled a lenient treatment
plot. Models were fitted with or without the Reddaway data to check
whether the rejected area had any effect on bird usage.
In addition to the treatment response models, the bird usage
data were related to plot-level covariates describing sward
conditions (sward structure, composition, seed and invertebrate
resources). Bird usage at the plot scale was related to plot level
assessments of sward structure derived from the weekly monitoring
of sward height for grazing control purposes (above). Plot sward
height means and CVs were calculated for each sward monitoring
visit and the values on intervening dates were interpolated. The
resulting sward structure dataset was used to model bird usage on
each plot during every bird-counting visit. The covariate models
incorporated data from all plot-visits, regardless of whether or
not the planned treatments were correctly implemented. The
modelling approach was the same as that used for the treatment
models.
Foraging patch selection by buntings was also examined at a
finer, within-plots scale, as part of an attempt to clarify why
bunting usage of lenient plots was lower than expected. This took
place during the 2008 and 2009 breeding seasons and generated 54
foraging observations, 39 for yellowhammer and 15 for cirl bunting.
All foraging sites were paired with unused randomly located control
sites in the same plots and at the same distance from the field
boundary. Ten sward height measurements were collected at each
sampling patch defined as a 3m diameter circles centred on each
foraging or control location (i.e. the same HFRO measurements used
for characterising sward structure at the plot level). GLMs with a
logit link and binomial errors were used to model the probability
that a patch was used for foraging by buntings. A binary response
variable distinguished foraging sites (1) from unused controls (0)
and a dummy variable identifying each foraging site-control site
pair was forced into the models. We tested for any influence of
sward height means and CV (linear and quadratic effects) on patch
usage by birds. Initially, all the foraging site data were analysed
in a single model, then the effects of year, bunting species and
plot treatment were tested as interactions with the sward structure
variables.
A sward-opening experiment was undertaken during the summer of
2009, to test whether access to invertebrate-rich lenient swards
was limited by sward accessibility. Patches of grass were kept
short by regular mowing and clippings were removed. The experiment
took place on five sites, in the margins of the lenient treatment
plots. Two configurations of mown patches were tested: two
2m-diameter circles and two groups of four 1m-diameter circles
(i.e. the same area, but with twice the perimeter length).
Observations of bird usage were collected as part of the timed
foraging watches on these plots and none of these watches were
undertaken within three days of mowing. In total, 47 timed watches
were undertaken on lenient plots that included mown patches (range
8 to 11 visits per site).
2.2. Experiment 2: Extensive grazing on intensive grassland with
over-winter refugia for invertebrates
Four grazing treatments were imposed on four permanent grass
fields (two fields at ADAS Rosemaund, Herefordshire, and two at
ADAS High Mowthorpe, Yorkshire) each with a history of grazing by
beef cattle, and a history of fertilizer inputs of approximately
150kg N/ha/year. Fertiliser restrictions and the two
extensification measures described for Experiment 1 were tested in
a partially crossed design. The following combinations were imposed
as treatments on each of the four fields (split-plot design) during
summer 2006 and maintained until September 2009:
Treatment 1: Moderate grazing with early closure, and reduced
fertilizer. Beef cattle grazed 0.6 ha plots to a TSH of 6-9cm
between April and mid-July, after which all livestock were excluded
until April of the following year. Annual fertilizer input was
restricted to a single 50 kgN/ha application during March/April.
This treatment was designed to promote invertebrate densities and
sward heterogeneity, and to provide suitable spring and summer
foraging conditions for soil and surface-feeding birds (e.g.
starlings, thrushes, wagtails).
Treatment 2: Lenient grazing with early closure, and reduced
fertilizer. Beef cattle grazed 0.6 ha plots to a TSH of 12-16cm
(10-14cm in 2006) between April and mid-July, after which all
livestock were excluded until April of the following year. Annual
fertilizer input was restricted to a single 50 kgN/ha application
during March/April. This treatment is designed to promote
invertebrate densities and sward heterogeneity, and to provide
suitable spring and summer foraging conditions for sward-feeding
birds (e.g. buntings, sparrows, finches and larks).
Control: Moderate grazing without early closure, and reduced
fertilizer. Beef cattle grazed 0.6 ha plots to a TSH of 6-9cm
between April and the cessation of grass growth (typically
October). Annual fertilizer input was restricted to a single 50
kgN/ha application during March/April. This treatment is designed
to provide some sward heterogeneity and suitable foraging habitat
for soil and surface-feeding birds.
Fertilized Control: Moderate grazing without early closure, and
normal fertilizer. Beef cattle grazed 0.6 ha plots to a TSH of
6-9cm between April and the cessation of grass growth (typically
late October). Annual fertilizer input was 150 kgN/ha/year (three
50kg/ha applications spread over the growing season).
A put-and-take grazing system was employed to achieve TSHs, with
plots being rested when swards fell below the lower TSH limit.
Grazing control was based upon regular (approximately weekly)
measures of sward height along W-shaped transects covering each
grazing plot. During March of each year, two passes of a chain
harrow were used to dislodge and partially remove accumulated dead
vegetation and litter from the moderate and lenient treatment plots
with the aim of opening swards up and promoting new grass
growth.
Devon extension study: During 2008 and 2009, the Experiment 2
treatment that was associated with the largest enhancement of
invertebrate abundance (Treatment 2, lenient grazing followed by
early closure), was established with a control treatment (as above)
on eight fields in South Devon with a presumed history of intensive
management (i.e. grazed pastures with fertilizer inputs of at least
50kg N / ha /year). Soil nutrient assessments conducted in April
2009 suggested two of these sites had probably not been subject to
such intensive management, but a relatively high nutrient status
was confirmed at the remaining sites (Appendix E, Table E1). One of
the eight sites was discontinued after 2008 following persistent
breaches of treatment protocols. The main aim of the Devon
extension was to assess the biodiversity benefits (particularly
bird usage) of the most promising Experiment 2 treatment (which
lacked sufficient replication to assess bird usage). Assessments of
sward structure/composition, invertebrates and bird usage were
therefore conducted on all extension sites during 2008 and
2009.
2.2.1 Agronomic measures
Using a similar technique to that in Experiment 1 (section
2.1.1), cattle grazing days were recorded and accumulated over the
early and late grazing season periods. On all Experiment 2 sites,
individual cattle growth rates were calculated by weighing core
animals at turnout, early closure and at the end of the grazing
season. Grazing output was expressed as kg of live weight gain per
hectare. During 2007-09, herbage pluck samples were collected from
all treatment paddocks at all sites. Three samples were collected
from each paddock during May, mid-July and September. Laboratory
analysis was as described for Experiment 1 (section 2.1.1). Soil
nutrient status was assessed for whole fields during spring 2006,
and for all individual treatment paddocks in spring 2009. Soil
sampling and analysis techniques were similar to those used in
Experiment 1 (section 2.1.1). Seed head counts of grasses and forbs
were undertaken on each treatment paddock on each site during
September of 2007 and 2009 using the method outlined above for
Experiment 1. Statistical analyses of these data were similar to
those conducted on Experiment 1 data (section 2.1.1). No agronomic
measures were collected for the eight Devon extension sites.
2.2.2. Swards and invertebrates
The design and sampling regime of Experiment 2 were largely
identical to Experiment 1 (section 2.1.3). Swards, vegetation
composition and invertebrates were recorded, assessed and analysed
using the same methods as described above for Experiment 1.
Experiment 2 differed as described below. Transects were spaced at
10, 30, 50, 70 and 90 metres from the field boundary for both the
ADAS and Devon extension sites. Detailed recordings of the
botanical composition were taken at the ADAS sites in 2006 and 2009
only. Neither drop discs nor pitfall traps were used on Devon
extension sites. Litter accumulation was assessed on the four main
Experiment 2 sites during March 2007-09 using the same methods used
in Experiment 1 (section 2.1.2).
2.2.3 Bird usage
Following the field methods adopted for Experiment 1 (section
2.1.4), regular summer bird usage surveys were conducted on all
Devon extension fields during 2008-09. Analyses of these data are
described in section 2.1.4.
2.3Experiment 3: Effects of modified grass silage management on
skylark breeding productivity
2.3.1. Study sites and silage management practices
Between 2006 and 2008 silage fields holding breeding skylarks
were identified in two areas of Dorset, England, centred on
Dorchester and Sherborne. Silage fields subjected to at least two
cuts were preferred as study sites, though some of these reverted
to single cut regimes after problems with weather or poor grass
growth. Some silage fields were allocated to experimental mowing
regimes (below) while others were subject to normal management and
served as controls. The majority of the farms were dairy farms. The
silage fields ranged from three-cut systems on high-yielding,
short-term Italian ryegrass leys, to two-cut systems on
semi-improved, permanent grassland. Organic silage fields were only
included if there was high residual soil fertility.
All farms used disc mowers with one, two or three cutting bars
and two farms used single-bar “swather” mowers. Each bar cut swaths
ca. 3m wide. On most mowers, adjustable baffles could optionally be
set to spread the cut grass to wilt or into rows for collection.
Swather mowers incorporate a conveyor belt to deflect the cut grass
onto a previously cut row, thereby avoiding the need to rake cut
grass together for collection. Swather mowers were used in
preference where there was a risk of raking up stones. Tedders were
used to spread wet cut grass or to produce drier silage for baling.
Windrowing rakes or swather mowers were used to gather up the cut
grass for collection. The effective operating widths of windrowing
rakes varied from 4m (single rotor) to 12.5m (four rotors). Silage
was picked up using self-propelled forage harvesters (followed by
tractors towing silage trailers) or big balers. Bales were picked
up and removed on trailers and were sometimes wrapped in plastic in
situ. The process of grass collection resulted in a high proportion
of the field being trafficked by machinery, particularly when
baling. Early cuts generally produce higher yields of wetter grass,
requiring at least a day of wilting. Most early crops are put into
silage clamps, rather than baled. Late cut yields tend to be lower
and require less drying. Consequently, late crops can be cut and
cleared in a single day and the grass is better suited to the use
of swather mowers and for baling.
2.3.2. Experimental mowing
Split-field experiments were used in 2006 and 2007 to compare
the effects of a raised cutting height and a low (control) cutting
height on skylark reproductive success. These treatments were
applied to the first and second cuts, after which management
reverted to the farmer. In 2006, farmers were asked to implement a
raised cutting height of 10-12cm and a control cut as low as
possible (resulting mean cutting surface heights: raised cut
12.3cm, SD 1.8cm and control 5.9cm, SD 1.0cm). In addition to this,
the raised cutting height plot was picked up one row at a time,
without using a windrowing rake. In 2007 farmers were asked to cut
the raised cutting height plot 8cm taller than the cutting height
on the control, but the same grass collection machinery could be
used across the whole field (raised cut 12.4cm, SD 2.7cm; control
6.8cm, SD 0.7cm). In 2008, whole fields were mown at a raised
cutting height of 10cm at the first cut (11.4cm, SD 0.6cm on first
cut), followed by a normal low cutting height at the second cut
(8.0cm, SD 0.6cm). Extra payments were offered to farmers willing
to accept a minimum 7-week cutting interval (6 weeks on three-cut
systems), but uptake of this option was low. The numbers of trial
and control fields contributing to this cutting height study were,
respectively, 12 and 11 in 2006, 8 and 4 in 2007 and 10 and 15 in
2008.
A second experiment was carried out on another set of fields in
2007. Raised cutting heights were used in an attempt to attract
skylarks onto areas that could be protected from subsequent silage
cuts. At the first cut, the field was cut at the normal low cutting
height, except for half-hectare Safe Nesting Plots (SNPs), each of
which was cut 10cm higher than the rest of the field (resulting
mean cutting surface heights: SNP 27.3cm, SD 12.2cm and control
9.0cm, SD 2.3cm). Thirteen SNPs were sited in the centres of 7
fields in areas known to hold skylark territories. At the second
cut, the SNPs were not mown and the retained grass was not cut or
grazed until after skylarks finished breeding in August.
Cutting dates and intervals were generally left to the
discretion of the farmers, so they could cut when the grass reached
target condition and/or weather permitted.
2.3.3. Yield measurements
Silage yields were measured on the split-field experiments only.
Yields for both treatments were measured at the first two cuts. Cut
herbage was weighed from 2m lengths of cut grass swaths in at least
eight random positions (depending on variability) on each plot. The
swaths were sampled using the “bread knife technique” and analysed
in the laboratory for dry matter content and quality (crude
protein, water soluble carbohydrates, MAD fibre, D value and ME).
The calculation of yield was based on swath weight and mower
cutting width measured at each sampling point. Mower cutting height
was also recorded at six positions per sampling point.
In 2008 a nine-swath wide (c.25m), 100m long strip was used to
measure yields from taking a low cut at both the first and second
cuts, whilst the remainder of the field was subjected to a high
first cut and low second cut. Yields at first cut were measured as
during 2006-07 (above). Following first cut, adjacent areas of high
and low cut grass were marked to allow second cuts to be sampled at
experimentally controlled cutting intervals of 7 and 8 weeks. These
samples were cut by hand to a height of 6cm, within 1m x 1m sample
plots.
2.3.4. Invertebrate sampling
During late June 2007, just before the second silage cut was due
to be taken, sward height (HFRO) and invertebrates were sampled at
20 locations in each treatment at each of eight study sites.
Invertebrates were sampled using sweep nets, Vortis suction
samplers and pitfall traps.
2.3.5. Measuring skylark reproductive success
Skylark nests were found and monitored to measure initiation
dates (first egg dates), daily survival rates, event-specific
survival rates for silaging operations and productivity. Particular
attention was paid to identifying pairs of consecutive nesting
attempts by the same skylark pair to establish replacement
intervals. Radiotags were used to measure post-fledging survival
and to establish the age at which fledglings were no longer
vulnerable to silaging machinery. All breeding parameters were
carefully tested for any effects of cutting height, date and chick
age or differences between alternative silaging machinery.
The consequences for skylarks’ annual productivity of the
different silage harvesting manipulations were estimated using
re-nesting models (Appendix C). Each model simulated the annual
productivity of 4999 populations of 100 pairs of skylarks nesting
on silage fields. The population productivity summaries included
the number of nesting attempts, number of chicks fledged, number of
chicks reaching independence and the numbers of nests and
fledglings failing due to silaging operations or other causes.
Various models were constructed to mimic different silage
harvesting scenarios. The model specified cutting dates, cutting
heights and the combinations of machinery used (each requiring
different sets of skylark nest/chick survival probabilities). Most
models focussed on two-cut and three-cut systems. In the basic
series of models, cutting dates were set to the observed means
(three-cut systems being earlier, with marginally shorter cutting
intervals). Subsequent models tested the effects of varying the
dates of mowing (maintaining the mean cutting intervals, to
represent inter-season variation in cutting dates relative to the
nesting season) and varying the lengths of the cutting intervals.
Cutting dates and intervals were increased or decreased by periods
of one or two weeks. Cutting height effects were compared by
testing varying combinations of high and low cuts at first and
subsequent cuts: low-low, high-low, high-high, low-low-low,
high-low-low, etc. Cutting date and height variations were explored
for the most typical silaging machinery combination: disc mower
plus windrower plus forage harvester. A separate set of models set
the low cut re-nesting delay to zero, to test the sensitivity of
annual productivity estimates to this factor.
A simplified set of models compared different machinery
combinations representing the best (swather mower followed by
forage harvester) and worst (standard mower, grass spread once,
windrower, then baled) extremes for nest survival. Two other common
scenarios were modelled: forage harvester collection for clamped
silage at the first cut, followed at the second cut by either
baling or the use of a swather mower. Further models examined
single, late cuts by testing single low or high cuts at two-week
intervals throughout June and July. These scenarios are
representative of late, single hay or haylage cuts, and delayed two
cut systems where the second cut would occur after the end of the
skylark breeding season. Late cuts such as these would typically be
baled.
3. RESULTS
3.1. Grazing treatment delivery
3.1.1. Experiment 1 and the Devon extension
The timing and duration of delivery of target sward heights
varied markedly between sites and years. The main factors
influencing treatment delivery were the speed with which farmers
responded to requests to move animals and prevailing weather
conditions. Wet weather during spring 2008 delayed stock being
turned out and encouraged lush grass growth, which combined to
cause delays in the attainment of TSHs at many sites. The removal
of cattle for vaccinations (TB and blue tongue) also hindered
grazing control. Once TSHs were achieved, they were usually
maintained, so the duration of TSH maintenance prior to plot
closure in July provides an indirect measure of the effectiveness
of grazing control at each site in each year (Appendix A).
Duration of TSH delivery was shortest in 2006 (mean duration of
TSH delivery prior to closure = 1.6 days; only 5 sites achieved TSH
by closure) due to delays associated with the recruitment and
establishment of trial sites (i.e. the provision of fencing and
water troughs). Treatment delivery on Experiment 1 sites was much
improved in 2007 and 2009 (mean TSH duration = 46.4 and 50.3 days
respectively) but poorer in 2008 (mean = 17.8 days) due mainly to
weather-related delays to the start of grazing and lush grass
growth. Treatment delivery on Devon extension sites was similar to
that on Experiment 1 sites during 2008 and 2009 (Appendix A).
We allowed for variation in the timing of treatment delivery by
testing for any influence of the duration of TSH attainment in
analyses of agronomic and biodiversity data. Complete failure to
implement treatments at a small number of sites in particular years
(usually caused by a severe lack of grazing due to cattle illness
or farmer problems) resulted in the exclusion from all analyses of
data from those sites and years (see Appendix A).
3.1.2. Experiment 2
Target sward heights were generally achieved quickly and
maintained successfully on both Rosemaund fields during all years
of the study (Appendix D). However, grazing control at the less
productive High Mowthorpe site was more problematic partly as a
consequence of the vegetation on those fields ‘burning up’ during
dry summer weather. This often resulted in sward heights (which
only measure live green foliage) continuing to fall after the
removal of all cattle from trial plots, particularly on lenient
treatment plots. TSHs were not adequately achieved at either of the
two High Mowthorpe fields in 2006 or at Elbow South in 2009 (the
two relatively dry summers of the study), but TSHs (or at least
sustained sward height separation) were achieved on the Elbow South
Field in 2007 and 2008, and on the Warren Dale field in 2008 and
2009 (Appendix D). Although post-closure grass growth on the early
closure treatments was more vigorous on the more fertile Rosemaund
fields (sward heights typically reaching ca. 20cm by early
September), grass growth was much slower on the High Mowthorpe
paddocks (sward heights rarely exceeded 15cm). Fertilizer
application extended the grazing season by approximately 14 days at
Rosemaund but had little effect at High Mowthorpe (Appendix D).
3.2. Effects of grazing treatments on soil nutrient status
At Experiment 1 and Devon extension sites, soil mineral nitrogen
(SMN) and nitrate (NO3-) were significantly higher in both the
lenient and moderate treatments compared to the control (Appendix
E). On Experiment 1 sites, potassium (K) and magnesium (Mg)
concentrations were also significantly higher on lenient and
moderate plots by 2009 (Table E2). Ammonium (NH4+) concentration
was also higher on the moderate and lenient plots, but the
differences were not significant. Although there were no
significant treatment effects on the soil parameters measured at
Experiment 2 (ADAS) sites, there was a tendency for higher
concentrations of nitrate and potassium in the early closure
treatments (Table E2).
3.3. Effects of grazing treatments on agronomic measures
3.3.1. Experiment 1
Tables of mean grazing days (for core and non-core sites) and
animal growth rates are presented in Appendix F (Tables F1 to F3).
Season- and year-specific breakdowns of mean livestock output are
presented in Table 3.3.1. Although there was no impact of grazing
treatments on early season output during 2007, quite large impacts
were evident in 2008 (14% loss on moderate and 35% loss on lenient
plots) and 2009 (losses of 23% and 48% respectively) (Fig 3.3.1).
Early season livestock output differed significantly between sites
(P<0.01) but not between years (P<0.09) or treatments
(treatment: P=0.15; treatment*year: P=0.14). However, paired
comparisons suggest that livestock output differed significantly
between lenient and control plots during 2008 (P<0.08) and 2009
(P<0.005), but the differences in output between moderate and
control plots were not significant. Separate analyses for each
year, indicated a significant treatment effect in 2009 only
(P<0.0001).
Figure 3.3.1: Annual mean livestock output (kg/ha) during early
season (until mid-July) based on core cattle at six sites.
Confidence intervals are +/- 1 SE.
0
100
200
300
400
200720082009
liveweight output (kg/ha)
Con
Mod
Len
Average season-long livestock output on control plots remained
stable across years (493, 534 and 504 kg/ha in 2007-2009
respectively), although the relative proportion of output during
early and late season varied between years (Table 3.3.1). Early
season output (before mid-July) was significantly higher than late
season output (after mid-July) on control sites (P<0.002) and
accounted for 65% of total annual output across all three years
(Table 3.3.1). Thus, the early closure of lenient and moderate
treatments entailed a loss of livestock output of approximately
35%. Combining the early (grazing) and late season (plot closure)
losses, the total season-long losses of livestock output were
31.3%, 51.1% and 42.6% during 2007-2009 respectively for the
moderate treatment, and 32.3%, 63.0% and 61.6% respectively for the
lenient treatment.
The economic equivalent of the livestock output data were
estimated assuming beef cattle to be worth £1.52 per live kg (as in
September 2009; Table 3.3.1). These figures highlight the high cost
associated with early closure in 2008 and 2009. The cost (per ha)
of early closure (control vs. moderate) was £236 in 2007, £415 in
2008 and £327 in 2009. The cost (per ha) of lenient grazing during
the early season (lenient vs. moderate) was £7 in 2007, £97 in 2008
and £144 in 2009 reflecting a large decline in economic output from
the lenient sward over the study period. The combined cost (per ha)
of early closure plus lenient grazing (lenient vs. control) was
£243 in 2007, £512 in 2008 and £471 in 2009.
Table 3.3.1: Mean cattle live weight output (kg/ha) and economic
output (£/ha) for Devon core sites. Data are observed mean weights
for early and late season (and SE) plus the economic value of these
outputs.
Year
Early- late
Control
Moderate
Lenient
(kg/ha)
£/ha
(kg/ha)
£/ha
(kg/ha)
£/ha
2007
Early
314 (55.8)
477
338 (50.9)
513
333 (47.4)*
506
(6 sites)
Late
179 (42.7)
272
-
-
-
-
Total
493 (109.2)
749
338 (50.9)
513
333 (47.4)
506
2008
Early
302 (73.2)
459
261 (73.4)
396
197 (57.4)
299
(5 sites)
Late
232 (94.2)
352
-
-
-
-
Total
534 (137.5)
811
261 (73.4)
396
197 (57.4)
299
2009
Early
376 (23.0)
571
289 (54.7)
439
194 (29.8)
295
(6 sites)
Late
128 (32.6)
195
-
-
-
Total
504 (73.7)
766
289 (54.7)
439
194 (29.8)
295
Herbage quality data showed complex patterns of variation with
significant interactions for most measures involving treatment,
month and/or year (Table F4). ME (an overall measure of feed
quality) did not differ significantly between control and moderate
treatments in any year, but was significantly lower in lenient
plots during 2007 and especially 2009 (Fig. 3.3.2). ME did not
differ significantly between treatments in 2008.
Figure 3.3.2: Mean ME (+/- SE) from pluck samples for different
months, years and treatments
Pluck samples were collected from areas where cattle were
actively grazing and does not therefore include rejected
vegetation. A visual sward assessment in April 2009, suggested that
swards in the moderate and, especially, the lenient paddocks had
deteriorated in agronomic quality relative to control plots, with a
marked reduction in the extent of high quality grass forage and an
increase in dead litter at the base of the sward, and in the extent
of dead grass. This reduction in sward quality may account for the
reduction in live weight gain in 2009 (Fig. 3.3.1). The heavier
early season grazing on moderate plots may have encouraged the
growth of higher quality vegetative grass that cattle were
apparently able to select during May and July (Fig. 3.3.2).
8.0
8.5
9.0
9.5
10.0
10.5
11.0
MayJulMayJulMayJul
200720082009
ME (MJ/kg DM)
Con
Mod
Len
3.3.2 Experiment 2
Analyses of livestock output indicated significant county and
county*treatment effects, and we therefore analysed data from the
two counties separately. County-, year- and treatment-specific
tables of mean grazing days, cattle growth rates and livestock
output are presented in Appendix G.
At Rosemaund, early season output did not differ significantly
between treatments (P=0.13), but was significantly lower in 2008
than in the other three years (year effect: P=0.0014) (Fig.
3.3.3a). At High Mowthorpe, output also differed between years
(P<0.002; output in 2009 was significantly lower than in all
other years), and between treatments (P<0.02) with output on the
lenient treatment being significantly lower than on any of the
other three treatments (Fig. 3.3.3b). In neither county was there a
significant year*treatment interaction (P=0.59 for Rosmaund, P=0.26
for High Mowthorpe). The moderate treatment was associated with
reduced early season output in 2008 and 2009 at Rosemaund (by 28%
and 21% respectively, relative to the control) but not at High
Mowthorpe (Fig. 3.3.3). The lenient treatment was associated with a
modest (non-significant) early season loss at Rosemaund in 2009
only (19% loss relative to the control), but large significant
losses were recorded at High Mowthorpe during 2008 and 2009 (61%
and 51% respectively) suggesting possible sward deterioration.
Early season output was significantly greater than late season
output in both counties (P<0.0001), and accounted for 66% and
70% (averaged across 2007-2009) of the total output of the
fertilized and unfertilized control treatments (Appendix G). Late
season output was significantly higher on the fertilized than on
the unfertilized controls at Rosemaund (P<0.03) but not at High
Mowthorpe. Average season-long live weight output is summarised for
each treatment and county in Figure 3.3.4. At Rosemaund, total
losses were higher on moderate plots than on lenient plots during
2006-08 (20%, 36% and 50% for moderate vs. 10%, 19% and 4% for
lenient) and similarly large (ca. 40%) in 2009. At High Mowthorpe,
total losses were much higher on lenient plots (41%, 72% and 65%
during 2007-09) than on moderate plots (13%, 27%, 27%).
Figure 3.3.3: Annual mean early season livestock output (kg/ha)
at (a) Rosemaund, Herefordshire and (b) High Mowthorpe, Yorkshire.
Confidence intervals are +/- 1 SE.
(a)
(b)
0
200
400
600
800
1000
2006200720082009
liveweight output (kg/ha)
Con + N
Con
Mod
Len
0
100
200
300
400
500
600
2006200720082009
liveweight output (kg/ha)
Con + N
Con
Mod
Len
Figure 3.3.4: Annual mean season-long livestock output (kg/ha)
at (a) Rosemaund, Herefordshire and (b) High Mowthorpe, Yorkshire.
Late season output data for High Mowthorpe in 2006 were considered
unreliable. Confidence intervals are +/- 1 SE.
(a)
(b)
0
200
400
600
800
1000
2006200720082009
liveweight output (kg/ha)
Con + N
Con
Mod
Len
0
200
400
600
200720082009
liveweight output (kg/ha)
Con + N
Con
Mod
Len
The economic equivalents of the output data are summarised in
Table 3.3.2 and highlight the much greater livestock output at
Rosemaund. The costs (per ha) of early closure (control vs.
moderate treatments) were £202, £419, £364 and £456 during
2006-2009 respectively (mean £360) at Rosemaund, compared to £99,
£203, and £116 during 2007-2009 (mean £139) at High Mowthorpe.
There were no costs associated with early season lenient grazing
(moderate vs. lenient) at Rosemaund (livestock output was always
higher in the lenient treatment than in the moderate treatment),
while at High Mowthorpe the costs of early season lenient grazing
were £219, £333 and £164 during 2007-09 (mean £239) (Table 3.3.2).
The combined (per ha) costs of early closure plus early season
lenient grazing were £103, £217, £33 and £440 at Rosemaund during
2006-2009 (mean £198), and £318, £536 and £280 at High Mowthorpe
during 2007-2009 (mean £378). The additional fertiliser applied to
the fertilised control plots resulted in an average increase in
livestock output (relative to control plots) of 15% (£146/ha) at
Rosemaund and 21% (£116/ha) at High Mowthorpe, although the effect
of fertilizer addition was variable across sites and years (e.g.
from £5/ha at High Mowthorpe, to £197/ha at Rosemaund both in
2009).
Table 3.3.2: Summary of total economic output (£/ha) for
Experiment 2 sites. Estimates are based on mean live weight output
(see Tables G7 and G8 in Appendix G for details)
Year
Fertilised Control
Control
Moderate
Lenient
Rosemaund
2006
1169
1030
828
927
2007
1242
1171
752
954
2008
912
732
368
699
2009
1341
1144
688
704
High Mowthorpe
2006 *
588
400
411
552
2007
900
784
685
466
2008
895
741
538
205
2009
434
429
313
149
* Late season output for control treatments was estimated from
incomplete data
Figure 3.3.5: Mean ME from May & July pluck samples.
8
8.5
9
9.5
10
10.5
11
11.5
200720082009
ME (MJ/kg DM)
Con + N
Con
Mod
Len
The herbage quality data were heterogeneous with significant
interactions involving site, treatment, month and year (Table G9
& G10). Early season ME varied significantly between sites
(being consistently higher at the Rosemaund fields), years (lower
in 2009), months and treatments (lowest in the lenient, highest in
the controls) with significant year*month and year*treatment
interactions (Table G9). ME was significantly lower on lenient
plots during all three years, but in 2009 there was a marked
reduction in herbage quality on lenient plots relative to other
treatments (Fig. 3.3.5). Similar patterns of variation in other
herbage quality measures are summarised in Appendix G (Tables G11–
G15).
3.4 Effects of grazing treatments on swards
3.4.1 Experiment 1
Sward Height and structure: During all four summers (July),
sward height (HFRO and drop disc) was significantly higher in the
lenient treatment than in the control or moderate treatments (P
< 0.05; Fig. 3.4.1.1). During Spring 2007 and 2009, mean HFRO
sward height was significantly higher in lenient plots than in
moderate or control plots (Table H5). Compressed vegetation height
(measured using drop discs) was also significantly higher on
lenient plots than on control or moderate plots, indicating greater
standing biomass on lenient plots (Table H7). In 2006, there were
no significant differences in sward height CV between treatments,
but during subsequent years the vegetation in the lenient plots
(and by 2009 the moderate plots) became less heterogeneous (lower
CV) than the vegetation in the control plots (treatment*year: P
< 0.001; Fig. 3.4.1.2). Differences in vegetation structure were
less marked during spring, but there was a significant treatment
effect in 2007 (taller swards in lenient plots with a lower CV;
Table H5). Distance from field edge had no significant effect on
sward height or CV.
Plant species richness and diversity: Although there was no
significant overall treatment effect on the number of plant species
in quadrats, by 2009 there had been a significant decline in the
number of species (and in Shannon’s Evenness index, Fig. H7b) on
moderate and lenient plots but not on control plots (treatment*year
interaction: P < 0.01; Fig 3.4.1.3). A significant change in
plant community composition occurred in the lenient and moderate
treatments (Fig 3.4.1.4; Appendix H). The change in species
composition was mainly due to increasing cover of Ranunculus
repens, Holcus lanatus and Agrostis spp. and reductions in the
cover of Trifolium repens and Lolium perenne. Litter cover was also
higher in lenient and moderate plots than in control plots. The
positive and negative changes in species grass cover cancelled each
other out and there was no overall change in the abundance of
grasses as a functional group. Consequently, variation in the
relative abundance of functional groups between treatments was
caused mainly by a reduction in the cover of agricultural clovers
and small increases in the cover of non-leguminous forbs and litter
(Appendix H).
Flower head counts: In spring 2009 there were significantly more
flower heads in the control plots than in the moderate and lenient
treatment plots (P < 0.05). However, by July, flower heads were
significantly more abundant in lenient plots than in moderate or
control plots (P < 0.05; Fig 3.4.1.5). The difference in pattern
between the spring and summer samples was due to the higher
abundance of the early-flowering species, such as Anthoxanthum
odoratum, in the control treatment and the higher abundance of the
summer-flowering grasses Holcus lanatus and Agrostis spp. and the
lower grazing pressure in the lenient treatment.
Figure 3.4.1.1: Mean HFRO Sward Height. Summers of 2006 - 2009.
White, grey and black bars indicate control, moderate and lenient,
respectively. The error bars indicate one SE.
Figure 3.4.1.2: Mean HFRO Sward CV - summers of 2006 - 2009.
Shading as in Fig. 3.4.1.1.
Figure 3.4.1.3: Number of plant species per m2. Shading as in
Fig. 3.4.1.1.
Figure 3.4.1.4: Principal Response Curve of plant species. PRC
graphs show the temporal evolution of differences in community
composition relative to the control treatment (the latter
standardised to remain constant and indicated by a horizontal line
through y=0). The solid black line indicates the lenient and the
dashed red line the moderate treatment.
Figure 3.4.1.5: Total number of flower heads per 3m-radius
circle during spring and summer 2009. Shading as in Fig.
3.4.1.1.
Figure 3.4.1.6: Mean September grass seed head counts. Shading
as in Fig. 3.4.1.1.
Figure 3.4.1.7: Mean September forb seed head counts. Shading as
in Fig. 3.4.1.1.
0
10
20
30
40
50
60
70
20072009
Year
Grass seed heads (+/- SE)
0
10
20
30
20072009
Year
Forb seed heads (+/- SE)
September seed head counts: Grass seed heads were significantly
more abundant on moderate and lenient plots than on control plots
(treatment P<0.001), although grass seed heads were far less
abundant on moderate plots in 2009 than in 2007 (Fig 3.4.1.6). Forb
seed heads were significantly more abundant on moderate plots than
on lenient or control plots (Fig. 3.4.1.7).
Injurious weeds: Within each year, there were no significant
treatment effects on the cover of the four weed species Cirsium
vulgare, Cirsium arvense, Rumex obtusifolius and Urtica dioica.
Cover of Cirsium vulgare declined in all treatments, and
significantly so in lenient and moderate plots (P < 0.05)
(Appendix H, Table H2). By 2009, the cover of Cirsium arvense
exceeded 5% cover on two study fields (both organically managed),
but cover of this species was not influenced by the experimental
treatments (Table H2).
3.4.2 Experiment 2 (ADAS sites)
HFRO and drop disk sward measurements: Although sward heights on
lenient plots were significantly higher than on all other
treatments in all four years, lenient sward were on average shorter
than the target of 12-16cm (Fig. 3.4.2.1, Appendix I1). Compressed
mean sward height, measured using drop discs, was greater in the
lenient compared with the other treatments and the control in 2006
and 2009 but only significantly so in 2009. There were no
significant effects of the treatments on sward structural
heterogeneity (i.e. CV derived from HFRO or drop disk measures)
(Appendix I1).
Figure 3.4.2.1: Mean sward height, as measured using HFRO sward
sticks, in Experiment 2. White, grey, black and hatched bars
indicate control, moderate, lenient and fertilized control
treatments. Error bars indicate SE.
Figure 3.4.2.2: Principal Response Curve (PRC) graph showing
changes in relative abundance of functional groups in Experiment 2.
The dashed, solid and dotted lines show the moderate, lenient and
fertilized control treatments respectively.
Figure 3.4.2.3: Mean September grass seed head counts. Shading
and error bars as in Fig. 3.4.2.1.
Figure 3.4.2.4: Mean September forb seed head counts. Shading
and error bars as in Fig. 3.4.2.1.
0
50
100
150
200
250
2007
2009
Year
Grass seed heads (+/- SE)
0
20
40
60
80
100
2007
2009
Year
Forb seed heads (+/- SE)
Vegetation composition: Assessments of the vegetation
composition in permanent quadrats revealed that forbs were less
abundant, and grasses more abundant, in the lenient treatment
especially in 2008 and 2009 (Fig. 3.4.2.2; Appendix I3).
Differences in vegetation composition between the other three
treatments were relatively minor. There were no significant shifts
in the abundance of individual species, which were only assessed in
2006 and 2009 (Appendix I3).
September seed head counts: Grass seed heads were significantly
more abundant on early closure plots during both years, and the
higher counts on control plots in 2009 were associated mainly with
High Mowthorpe plots (Fig. 3.4.2.3). Forb seed heads (mainly white
clover) were significantly more abundant on early closure plots in
2007 and on moderate plots in 2009 (Fig. 3.4.2.4).
Injurious weeds assessment: Within each year, there were few
significant differences between treatments in the cover of the four
injurious weed species. Only during 2009 were treatment effects
detected for Cirsium arvense and Urtica dioica (both species were
more abundant in lenient plots than in other treatment plots (Table
I6). However, the mean percentage cover of each species remained
well below 5% on all sites. Cirsium vulgare cover increased
significantly over the course of the study, but this occurred
across all treatments (Table I6).
3.4.3. Experiment 2 (Devon extension sites)
Lenient grazing resulted in significantly taller swards than in
the control plots with target sward heights being delivered in July
2009 but not in July 2008 (Fig. 3.4.3.1; Appendix J). Sward
structural heterogeneity (CV of HFRO sward height) was higher in
control plots than in lenient plots, but the difference was only
significant in 2009 (Fig. 3.4.3.2; Table J2). There were no
treatment effects on the abundance of functional groups in the
vegetation or on individual plant species.
The abundance of flower heads was higher in the control
treatment during spring 2009, but higher in the lenient treatment
by July 2009 (Appendix J1). There were no significant changes in
the cover of injurious weeds between the two years and no
differences in weed cover between control and lenient treatments
(Appendix J1).
Figure 3.4.3.1: Mean sward height in the Devon Extension sites,
measured using HFRO sward sticks. White and black bars represent
control and lenient grazing treatments, respectively and error bars
indicate SE.
Figure 3.4.3.2: Mean Structural heterogeneity of the vegetation
in the Devon Extension sites. White and black bars represent
control and lenient grazing treatments, respectively and error bars
indicate SE.
3.4.4. Effects of treatments and harrowing on litter
Harrowing was seen to remove substantial amounts of litter, but
this did not result in any significant change in litter
distribution through the sward column (Appendix K). Harrowing
evidently spread the remaining litter throughout the sward, and the
litter stick measurements could not quantify the magnitude of the
litter removal.
Litter measurements prior to harrowing in each year were used to
examine whether the treatments affected rates of litter
accumulation over time. Litter densities increased substantially
over time on the two early closure treatments, but remained
constant on the control plots (Appendix K, Figure K2.1). Litter
accumulation rates did not differ between the moderate and lenient
treatments, though litter cover scores suggested that litter
accumulation rates were greater on the lenient treatment (Appendix
H). The rates of litter accumulation on early closure plots did not
differ between plots that were harrowed with spring-tine harrows,
chain harrows or fields that were not harrowed (P> 0.7). This
suggested that harrowing had little effect on the rates of litter
accumulation on early closure plots.
3.5 Factors affecting invertebrate abundance and communities on
experimental grazing plots
3.5.1 Experiment 1
Invertebrate abundance: The total number of invertebrates
sampled in July by both Vortis and pitfall trapping was
significantly higher in the lenient than in the control or moderate
treatments (P < 0.001 for vortis samples; P = 0.002 for pitfall
samples). Total invertebrate abundance (recorded with vortis
sampling) did not differ between the moderate and control
treatments in any of the four years (Appendix H1) although mean
abundance was slightly higher in the moderate treatment (Fig.
3.5.1.1). Total invertebrate abundance increased over the course of
the study in all treatments, as did the difference between the
lenient and other treatments (Fig. 3.5.1.1). Changes in total
invertebrate abundance were mainly caused by changes in the
abundance of invertebrates belonging to the smallest size class
(body length 2-5cm), although significant and proportionally larger
changes were evident in the medium and large body size classes
(Table H6).
The abundance of important bird food taxa showed a similar
pattern of treatment effects and temporal change (Fig. 3.5.1.2).
Large and significant treatment effects were evident for a diverse
range of invertebrate taxa including large and small-bodied groups
(Fig. 3.5.1.3; Table H6). In all cases, abundance was significantly
higher in the lenient treatment (suggesting a beneficial impact of
lenient grazing), and in several cases abundance in the moderate
treatment was significantly higher than that in control plots (e.g.
Auchenorrhyncha, Heteroptera, Staphylinidae, Fig. 3.5.1.3)
suggesting beneficial impact of early sward closure. Orthoptera
populations on the Devon study sites were almost exclusively
represented by a single species, Chorthippus parallelus. The number
of Orthoptera sampled with sweep nets showed significant treatment
(P<0.001) and treatment *year interaction (P<0.001) effects.
Although Orthopteran abundance was higher in the lenient than in
the control treatment in 2007, abundance subsequently declined and
by 2009 there were no significant differences between treatments
(Fig. 3.5.1.3 g). A similar pattern was evident for sawfly larvae
for which an initial treatment effect disappeared following a
cross-treatments reduction in abundance (Fig. 3.5.1.3 h; Appendix
H1).
Figure 3.5.1.1: Total invertebrate numbers from Vortis samples
in summers 2006 – 2009. Letters indicate significant within-year
treatment effects. Shading as in Fig. 3.4.1.1.
Figure 3.5.1.2: Total "birdfood" invertebrate numbers from
Vortis samples in summers 2006 – 2009. Letters indicate significant
within- year treatment effects. Shading as in Fig. 3.4.1.1.
Figure 3.5.1.3: Mean total abundance per paddock of selected
invertebrate taxa, summers 2006 – 2009. Taxa shown are Colembola,
Diptera, Auchenorrhyncha, Heteroptera, Araneae and Staphylinidae
from Vortis samples. Taxa from Sweep Net samples are also shown;
Orthoptera and Sawfly larvae. Where within year differences were
found) letters denote which treatments differ from which tested
using LSD (least significant differences, P=<0.05). Shading as
in Fig. 3.4.1.1.
a: Collembola
b: Diptera
C: Auchenorrhyncha
05001000150020002500
2006200720082009mean total abundance per paddock
(+/-1SE)ControlModerateLenient
baabaababa
0100200300400500600700
2006200720082009mean total abundance per paddock
(+/-1SE)ControlModerateLenient
baabaa
01002003004005006007008009001000
2006200720082009mean total abundance per paddock
(+/-1SE)ControlModerateLenient
aaaabbbbaacc
d: Heteroptera
e: Araneae
F: Staphylinidae
0102030405060708090
2006200720082009mean total abundance per paddock
(+/-1SE)ControlModerateLenient
cbabaabaa
050100150200250300
2006200720082009mean total abundance per paddock
(+/-1SE)ControlModerateLenient
050100150200250300
2006200720082009mean total abundance per paddock
(+/-1SE)ControlModerateLenient
aabbbaaabacbc
g: Orthoptera (Sweep Net)
h: Sawfly (Sweep Net)
020406080100120
2006200720082009
Mean total abundance per paddock (+/-1 SE)
ControlModerateLenient
abbaabababa
024681012
2006200720082009
Mean total abundance per paddock (+/-1SE)
ControlModerateLenient
baa
Figure 3.5.1.4: PRC of invertebrate community. Line patterns as
in Fig. 3.3.1.4.
As well as increasing invertebrate abundance, the treatments
were also associated with a shift in invertebrate community
composition (Fig. 3.5.1.4). During most of the experiment, the
invertebrate communities in the lenient plots differed from those
in the control plots. The moderate plots were more similar to the
control plots than the lenient plots. These differences in
community structure were least pronounced in 2007 (Fig. 3.5.1.4).
The lenient and moderate treatments had higher numbers of
Heteroptera, Auchenorrhyncha, Collembola, Araneae, Staphylinidae
and Diptera, but fewer Curculionidae, than the control plots. For
example, in 2009 average Auchenorrhyncha abundance was three times
higher in the lenient plots than in the control plots (Fig. 3.5.1.3
c). Species-level analyses of the Curculionidae, Auchenorrhyncha,
Chrysomelidae and Carabidae indicated that variation in the
abundance of a few (up to five) species within each group caused
most of the treatments effects.
The relative abundances of Auchenorrhyncha species in the
lenient plots diverged from those on the control plots; abundance
on moderate plots was intermediate (Fig. 3.5.1.5). This effect was
driven by few abundant species, e.g. Muellerianella faimairei, a
monophagous species feeding on Holcus spp., and Javesella dubia, a
species normally associated with Agrostis spp. Other species that
were responsive to our treatments were Macrosteles viridigriseus
and Streptanus sordidus which both feed on Agrostis species.
Descriptions of the ecology of the invertebrate species that were
found to be most responsive to our treatments are provided in
Appendix H5. Responsive species associated with tall swards include
Arthaldeus pascuellus. In parallel to the shifts in the abundance
of Auchenorrhyncha species, the number of species found in the
lenient plots was higher than that in the control, with species
richness being intermediate in moderate plots (P &
