SPATIAL VARIATION IN GRAZING INTENSITY OVER HEATHER MOORLAND IN ENGLAND A. E. Riding , D. S. Allen and M.J.W. Burke ADAS Woodthorne, Wergs Road, Wolverhampton, WV6 8TQ, U.K. Heather moorland is an internationally valuable habitat that has suffered a decline in extent and condition in recent years largely because of increased stocking rates. As a consequence of this decline the British government has used various policy instruments to encourage farmers to reduce sheep stocking rates on vulnerable moorland. However, reductions in stocking do not take into account spatial variation in grazing intensity, which can result in localised suppression of heather growth. A geostatistical approach was used to investigate grazing intensity on heather by sheep over a five-year period on five grazing units under reduced stock densities within the North Peak Environmentally Sensitive Areas Scheme. Spatial analysis showed that in most of cases studied, grazing intensity was randomly distributed. However, for four grazing units, evidence of large-scale patchiness in grazing intensity was found. The implication of this spatial variation in grazing intensity for the conservation of heather moorland is discussed. Keywords: moorland, Calluna, sheep grazing, geostatistics Running title: Spatial variation in sheep grazing Author to whom correspondence should be addressed: 1
163
Embed
Evaluating management techniques for controling Molinia ...randd.defra.gov.uk/Document.aspx?Document=BD1217_8… · Web viewWord count (incl. Tables and Figure legends) = 5,559.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
SPATIAL VARIATION IN GRAZING INTENSITY OVER HEATHER
grassland, U5; and Deschampsia flexuosa grassland, U2); in some areas these
grassland communities, as well as areas dominated by Molinia caerulea, are more
extensive.
The main agricultural use of the open moorland is for grazing of sheep. In addition,
many moors have a shooting interest and Calluna is managed for red grouse
(Lagopus lagopus scoticus). Increases in sheep numbers (Anderson and Yalden,
1981) and the threat that overgrazing posed to the conservation value of the area led
to the designation of the ESA in 1988. By 1996, 74% of moorland was managed
under an ESA agreement (ADAS, 1997a). The significance of the area in landscape
and conservation terms is recognised in its status as a National Park and a Site of
Special Scientific Interest.
At the time of the study, the North Peak ESA offered two basic tiers of management
under which moorland could be managed. Tier 1 covers all moorland and farmland
and contains two sub-divisions that apply to moorland; Tier 1A is the basic entry-
level tier and restricts cultivation and agricultural inputs and Tier 1C dictates limits
on stocking density. Tier 2 requires a greater degree of change to management
practices and contains two sub-divisions; Tier 2A provides for still lower stocking
rates and off-wintering of stock, and Tier 2B is concerned with exclosure of
5
moorland to encourage regeneration. The grazing units included in this study were
managed under tiers 1C and 2A.
SAMPLING STRATEGY
The annual programme of monitoring in the North Peak ESA formed part of an
extensive programme established to assess the effectiveness of the ESA scheme in
maintaining and enhancing the wildlife conservation, landscape and historical value
of the area (Hooper, 1992). Whilst the sampling strategy was not designed
specifically to provide data for spatial analysis, the approach gave sufficient
coverage over each grazing unit to enable an investigation of the nature of spatial
pattern in grazing intensity.
A grazing unit is an area of moorland that is managed as a single entity, with a
specified number of livestock. This was used as the primary sampling unit. Only
those grazing units with > 25 ha of Calluna, as shown on a land-cover map derived
from aerial photographs (ADAS, 1997b), were included within the sampling frame.
A random sample of five grazing units was then selected from a population of 34
units.
On each grazing unit estimates were made of the grazing intensity on Calluna.
Sampling within grazing units followed a stratified random sampling procedure
using 100 quadrats. Forty quadrats were allocated initially to each of two strata -
dominant Calluna (>75% cover) and sub-dominant Calluna (<75% cover) - with the
remaining 20 quadrats distributed proportionately in relation to the areas of the two
Calluna cover classes. Initial cover of Calluna was determined from land cover
6
maps and aerial photography. This resulted in the sampling plan outlined in Table 1.
Random co-ordinates within the stratified areas were plotted for each grazing unit on
1:10,000 Ordnance Survey maps, until the required number of points were obtained.
These points were located in the field from the maps using compass bearings and by
pacing. Quadrats were not fixed, an independent sample being obtained in each year
of monitoring. The grazing units were sampled first in the period late March-late
May in 1993 and annually within the same time frame until 1997.
ESTIMATION OF GRAZING INTENSITY
Grazing intensity was estimated by assessing the proportion of the current year’s
Calluna shoots that had been grazed by sheep. The technique has two stages: a field
stage involving the collection of Calluna and a laboratory stage to calculate a
grazing index. By conducting the assessment of grazing index in the laboratory,
observer bias was minimised.
The field technique involved the collection of a sample of Calluna stems. At each
sample point, four Calluna stems were collected from the corners of a 1 m 0.5 m
quadrat. If there was no Calluna under the corner of the quadrat, the stem nearest to
the corner was chosen. If Calluna was entirely absent then the quadrat was moved
to the nearest area of Calluna within a radius of 10 m and the above procedure was
repeated. The sample bags were opened in the laboratory and each stem cut a
distance of 4 cm from the crown of the stem to yield a sample of shoots. All shoots
less than 1 cm in length were discarded. Each of the remaining shoots was then
assessed for sheep grazing and any shoot with a terminal zone intact was classified as
7
being “ungrazed”. The number of shoots defined as “grazed” or “ungrazed” was
used to calculate the following index:
Grazing Index (GI) = Grazed Shoots / Total shoots
ANALYSIS OF VARIANCE
The effects of the main survey factors (grazing unit, ESA agreement status, Calluna
dominance and year) on GI were determined by Analysis of variance (ANOVA).
The ANOVA was a mixed model treating Grazing Unit as a random factor nested
within Agreement Status, with Agreement Status, Year and Calluna Dominance as
fixed factors. Grazing units were allocated Agreement Status codes 1 or 2,
corresponding to ESA tiers 1C and 2A respectively. GI was arcsine-square root
transformed to improve the distribution and satisfy assumptions of normality and
homogeneity of variance.
GEOSTATISTICAL ANALYSIS
Geostatistical analysis (e.g. Burrough, 1987; Rossi et al., 1992) was used to quantify
the nature of spatial variation present within each data set. Geostatistics provide
quantitative tools for the description and unbiased prediction of spatially distributed
variables. The spatial correlation structure is described by the semivariogram, which
quantifies spatial dependence by measuring the variation among samples
(semivariance) separated by the same distance (lag classes). Models fitted to the
semivariograms provide a quantitative representation of the spatial variation
exhibited in the field.
8
Where distinct spatial variation occurs, the values of semivariance rise gradually to a
point known as ‘the range’ at which it levels off. The range parameter (usually
denoted a) indicates the scale of the spatial dependence. Values of a variable
separated in space by distances less than the range are spatially autocorrelated, or
predictable. The value of semivariance at the point where a plateau occurs is known
as the sill, and is theoretically equal to the variance of the data. Theoretically, the
semivariogram should pass through the origin, because differences in values of a
variable at a point in space separated by zero distance should be zero. However,
semivariograms often appear to intersect the y-axis at a positive value of
semivariance. This value is known as the nugget variance (denoted CO) and has two
components: first, sampling error, and, secondly, unmeasured variation below the
smallest sampling distance (Isaaks and Srivastava, 1989).
Patterns of spatial variation in grazing intensity for each year surveyed were
examined using geostatistical analyses with GEO-EAS (Englund and Sparks, 1991)
and S-Plus (Mathsoft, 1995). Within the geostatistical analysis, each lag class was
represented by at least 40 pairs of points, and the maximum lag class was half of the
total distance measured in the field. Semivariograms were calculated at 50 metre
lags. Where distinct spatial structure was identified, spherical semivariogram
models (Isaaks and Srivastava, 1989) were fitted by weighted least squares (Cressie,
1985). The proportion of the nugget variance (CO) in relation to the overall
structural variance, or sill (CO+C1) was used as a measure of unexplained variation at
distances smaller than the smallest distance class, and any other sampling error for
each data set.
9
COMPARISONS BETWEEN YEARS
Since the location of surveyed points was not consistent over time, comparison
between years was carried out using interpolated values of GI at fixed locations over
each grazing unit. Interpolation and comparisons between years were only carried
out for those sites and years where spatial dependence was identified. The
semivariograms constructed were used with a kriging approach to interpolate values
of GI onto regular 100m×100m grids superimposed upon the grazing units. Sets of
interpolated values of GI for each year were then plotted against each other to assess
the extent to which patterns were repeated between years.
10
Results
LEVELS OF GRAZING INDEX
A total of 2,461 quadrats were sampled over 5 years on the 5 grazing units. 47% of
these quadrats had GI values below 0.2, with 93% having values below 0.6 The
overall mean GI across years was 0.26. Grazing Units 1, 2 and 3 all had overall
mean GI's of 0.28. Grazing Units 4 and 5 had mean GI's of 0.25 and 0.20
respectively.
Year (F(4,16)=24.8; p<0.001), Grazing Unit (F(4,2411)=18.9; p<0.001) and their
interaction (F(16,2411)=2.9; p<0.001) were all significant factors affecting changes in
GI (Fig 1). All grazing units showed broadly similar trends in GI between 1993 and
1997. This trend is one of a large decrease in GI between 1993 and 1994, followed
by a slight increase in 1995 and then a decrease in 1996, with similar intensities in
1997. The significant grazing-unit-by-year interaction is represented by
considerable variation between grazing units in changes in the relative increase in GI
in 1995. Grazing Units 1 and 2 showed greater increases than 3-5.
GEOSTATISTICAL ANALYSIS
The results of the geostatistical analysis are summarised in Table 2. Spatial
dependence in GI was seen in at least one of the survey years for four out of the five
grazing units studied, the exception being Grazing Unit 5. Semivariograms for
Grazing Unit 1 are shown in Figure 2 and typify the results found in the analysis. In
many cases, no evidence of spatial structure in GI was observed, and the
semivariogram form was flat (e.g. years 1993 - 1995 in Figure 2). This ‘pure
nugget’ semivariogram form is indicative of random spatial behaviour, as there is no
11
similarity between values of GI either at locations near to each other or those further
apart.
Where distinct spatial variation occurs in GI over the grazing unit, semivariance is
lower at small lag distances and the value of semivariance then increases with
increasing lag distance until it reaches a plateau at the range distance (e.g. years 1996
and 1997 in Figure 2). Figure 3 shows the spatial variation in GI in 1996 and 1997
for Grazing Unit 1. These kriged maps give a visual representation of the spatial
behaviour described by the semivariogram models. A more homogeneous
distribution is seen for the year where a longer range of spatial dependence was seen
(Figure 3b) compared with the spatial variation observed where a shorter range of
spatial dependence exists (Figure 3a).
The range of spatial dependence observed varied between grazing units and years
(Table 2). These values should be interpreted with reference to the overall area of
the grazing unit surveyed (Table 1) to gain an idea of the degree to which they
represent heterogeneous conditions. Spatial dependence over a small distance
relative to the size of the grazing unit indicates heterogeneity in grazing levels. This
was the case for Grazing Units 2, 3 and 4, and Grazing Unit 1 in 1996. In contrast,
spatial dependence over a longer distance relative to the overall size of the grazing
unit was seen for Grazing Unit 1 in 1997 which indicates less overall heterogeneity.
The nugget variance expressed as a proportion of the overall semivariogram
structural variance (Table 2) varied from 9.5 to 61.1%. These values indicate that
where it was possible to characterise spatial variation with the semivariogram, there
12
was considerable spatial variation at smaller distances not adequately detected by the
sampling approach. Since for most sites information was not collected on spatial
variations below 150 m, it is likely that much of this unexplained variation relates to
spatial variation over this distance.
Temporal change
Evidence of distinct spatial distribution in levels of grazing intensity was observed in
more than one year for three grazing units. These were 1996 and 1997 for Grazing
Unit 1; 1993 and 1996 for Grazing Unit 3; and 1994 and 1996 for Grazing Unit 4.
For these sites, the similarity between the years was seen by comparison of data
interpolated onto a common grid using scatter plots (Figure 4). Varying degrees of
similarity between years were seen for the three grazing units
For Grazing Unit 1 (Figure 4a), the scatterplot indicates broadly similar values in GI
at points between 1996 and 1997, and there was no real change in mean levels of GI
(0.254% in 1996 and 0.264% in 1997). For some points on the scatterplot, low
values in 1996 showed higher corresponding values of GI in 1997. The locations of
these points over the grazing unit were found in spatially distinct areas and can be
identified in Figure 3a as areas of low grazing intensity. There are also points on the
scatterplot where high values of GI were observed in 1996 which were then lower in
1997 which has much the same effect in making the distribution of GI more
homogeneous over space. The increased range of spatial dependence in levels of GI
from 495m in 1996 to 879m in 1997 (Table 2) quantifies this shift towards a more
homogeneous distribution of GI over the grazing unit.
13
The scatterplot for Grazing Unit 3 (Figure 4b) indicates considerable change in
values of GI at the same points from 1993 to 1996. A decrease in the mean value of
GI was seen, from 0.446 in 1993 to 0.224 in 1996. The reduction in values of GI did
not occur in a spatially consistent manner over the grazing unit. At sites where low
values of GI were recorded in 1993, these sites were also low in 1996. In contrast,
where high values of GI were recorded in 1993 these sites were, for the main part,
much lower when monitored in 1996.
For Grazing Unit 4 (Figure 4c) the scatterplot shows a slight increase in the levels of
GI from 1994 to 1996, with mean levels of GI changing from 0.169 in 1994 to 0.215
in 1996. A map of differences between the two years was constructed which
indicated that the temporal change in levels of GI was represented by a ‘blanket’
increase in GI over a large area of the moor and one region of size 25 ha that showed
a major increase in GI.
14
Discussion
In this study we have identified the existence of large-scale spatial variation in the
distribution of grazing of calluna by sheep. For four of the five grazing units
examined, spatial heterogeneity in grazing intensity was observed in at least one
year. This heterogeneity was evident in the form of spatially distinct areas of
uniformly high or low grazing intensity. The geostatistical approach used made it
possible 1) to examine temporal change in the spatial pattern of grazing intensity, 2)
to quantify changes to the scale of heterogeneity and 3) to identify change from a
heterogeneous distribution to a random distribution or vice versa.
Spatial variation in grazing pressure may result from factors such as social behaviour
of the grazing animal, the distribution of preferred vegetation types, availability of
shelter and provision of supplementary feed. These factors may be interrelated and
it may be difficult to identify causes of observed distributions of grazing pressure.
Furthermore, temporal variation in grazing patterns may make understanding
underlying causes more difficult. Considerable temporal change in grazing pattern
was identified in this study. Changes to the spatial distribution of grazing intensity
may be the result of a broad range of influences including management and climatic
effects. There were no major changes in stocking densities between the surveys on
the grazing units in this study as the ESA agreements had already been in place for
some time. However, changes in stock management and/or burning practices could
have been responsible for changes in grazing patterns. From field visits to Grazing
Unit 1, it became clear that the main localised area of high grazing intensity
corresponded with an area where sheep were regularly turned out, resulting in this
area receiving a higher density of sheep on a regular basis.
15
The existence of variation in the distribution of grazing sheep means that in some
areas effective stocking densities will exceed the overall stocking density for a
grazing unit; in other areas effective densities will be lower than the grazing unit
average. This has implications for the effective conservation of heather moorland.
A reduction of stocking levels alone would appear not to be a wholly effective
control against overgrazing. The occurrence of patches of localised high grazing
intensity may affect the competitive balance of moorland plant species by reducing
the vigour of Calluna, allowing invasion by other species (Welch, 1984; Welch and
Scott, 1995).
The specific locations where overgrazing occurs in relation to the distribution of
vegetation communities may have a considerable effect on the nature of any
vegetation change. Increased grazing intensity on Calluna at Calluna/grass
interfaces (Clarke et al., 1995a) will have a more marked effect on vegetation
change than in the middle of extensive monospecific areas of Calluna. Continued
incidence of high grazing intensity in particular areas (e.g. as seen in Grazing Unit 1)
may result in erosion of Calluna cover at the edges of Calluna stands in the same
way as that predicted by Clarke et al., (1995b) at smaller scales.
Uneven grazing pressure across moorland is not a problem per se. Localised high
grazing pressure is only undesirable if vegetation cannot sustain this level of grazing
and is likely to lead to a reduction in cover or condition of a valued resource, such as
calluna. Indeed, in some areas it may be beneficial to encourage heavier grazing in
order to reduce dominance by competitive grasses such as nardus stricta and molinia
caerulea or to prevent succession to woodland. Decisions on the manipulation or 16
control of spatial variation in grazing will depend on the overall management aims
for a moor. Spatial variation in grazing pressure may provide benefits if it provides
a diverse range of habitats for invertebrates (e.g. Gardner et al., 1997) and other
fauna (e.g. Fuller, 1996). Conversely, deleterious effects of uneven grazing are
likely to be most evident where patches of heavily grazed moorland vegetation
occur, especially where this results in the decline in condition and cover of plant
species which are of conservation importance (e.g. Calluna). Therefore where the
prime objective of a moorland management plan is the conservation of calluna, the
spatial pattern of grazing intensity on this species needs to be monitored and
managed.
In the absence of intervention by land managers, it seems that even under reduced
stocking levels sheep continue to congregate in favoured areas, rendering any
calluna located there vulnerable to grazing. Attracting sheep away from such areas
should be a key objective of any management plan that aims to conserve heather
moorland. There are several methods that can be used to redistribute grazing
intensity, either indirectly through changing the spatial pattern of vegetation through
burning and cutting (phillips and watson, 1995), or directly through amendment of
shepherding practices, such as the selective positioning of feed blocks (davies and
griffiths, 2000). Routine evaluation of practices such as the location of
supplementary feeding should be made by land managers to prevent overgrazing
hotspots in the same location year on year.
Whilst the monitoring and analysis techniques described in this paper are too labour
intensive to be used for moorland management in practice, alternative simple
assessment measures could be employed. Localised overgrazing of calluna and 17
other signs of heavy grazing pressure (e.g. Dunging) can be relatively easily
identified (e.g. Through recognition of growth forms typical of overgrazing
(macdonald, 1990)). Regular assessment of moors in respect of the occurrence and
distribution of such features would provide adequate warning of potentially
deleterious effects. Managers could then amend shepherding or feeding practices
appropriately to maximise the potential for the maintenance or enhancement of
moorland under their stewardship.
18
Acknowledgements
The financial support of the Ministry of Agriculture Fisheries and Food for the
monitoring of the ESA programme is gratefully acknowledged. We thank the
landowners for access and all that have assisted in the field and laboratory work.
The authors are grateful to Sarah Gardner and Francis Kirkham for helpful
comments on the work presented. This paper was written with financial assistance
from MAFF under project BD1217.
19
REFERENCES
ADAS (1997a). Environmental Monitoring in the North Peak ESA 1988-1996. ADAS report to the Ministry of Agriculture, Fisheries and Food.
ADAS (1997b). Landscape Monitoring in the North Peak ESA 1988-1996. ADAS report to the Ministry of Agriculture, Fisheries and Food.
ADAS (1997c). Biological Monitoring in the North Peak ESA 1988-1996. ADAS report to the Ministry of Agriculture, Fisheries and Food.
Anderson, P. and Yalden, D.W. (1981). Increased sheep numbers and the loss of heather moorland in the Peak District, England. Biological Conservation 20, 195-213.
Anon. (1989). Environmentally Sensitive Areas (First Report). MAFF. HMSO, London.
Armstrong, H.M. and Macdonald, A.J. (1992). Tests of different methods for measuring and estimating utilization rate of heather (Calluna vulgaris) by vertebrate herbivores. Journal of Applied Ecology 29, 285-294.
Burrough, P.A. (1987). Spatial aspects of ecological data. In Data analysis in community and landscape ecology (R.H. Jongman, C.J.F. ter Braak and O.F.R. van Tongeren, eds), pp. 213-251. PUDOC, Wageningen.
Clarke, J.L., Welch, D. and Gordon, I.J. (1995a). The influence of vegetation pattern on the grazing of heather moorland by red deer and sheep. I. The location of animals on grass/heather mosaics. Journal of Applied Ecology 32, 166-176.
Clarke, J.L., Welch, D. and Gordon, I.J. (1995b). The influence of vegetation pattern on the grazing of heather moorland by red deer and sheep. II. The impact on heather. Journal of Applied Ecology 32, 177-186.
Cressie, N., (1985). Fitting variogram models by weighted least squares. Mathematical Geology 17, 563-586.
Davies, o.d. and griffiths, j.b. (2000). The strategic use of feed blocks to manage
sheep grazing in the uplands. In grazing management (a.j. rook and p.d. penning,
eds), occasional symposium no. 34, british grassland society.
Englund, E. and Sparks, A. (1991). GEO-EAS User's Guide. Environmental Protection Agency, Las Vegas.
Felton, M. and Marsden, J.H. (1990). Heather Regeneration in England and Wales. Nature Conservancy Council, Peterborough.
Fuller, R.J. (1996). Relationships between grazing and birds with particular reference to sheep in the British uplands. BTO Research Report No. 164. British Trust for Ornithology, Thetford.
20
Gardner, S.M., Hartley, S.E., Davies, A. and Palmer, S.C.F. (1997). Carabid communities on heather moorland in northeast Scotland: The consequences of grazing intensity for community diversity. Biological Conservation 81, 275-286.
Hooper, A. (1992). Field monitoring of environmental change in the Environmentally Sensitive Areas. In Land use change: the causes and consequences (M.C. Whitby, ed) pp. 53-59. ITE Symp. No. 27. HMSO, London.
Isaaks, E H and Srivastava, R.M. (1989). An Introduction to Applied Geostatistics. Oxford University Press, New York.
Lloyd, M. (1967). Mean Crowding. Journal of Animal Ecology 36, 1-30.
MacDonald, A. (1990). Heather damage: a guide to types of damage and their causes. Nature Conservancy Council, Peterborough.
Mathsoft (1995). S-Plus. Mathsoft Inc, Seattle.
Phillips, J. and Watson, A. (1995). Key requirements for management of heather moorland: now and for the future. In Heaths and Moorland: Cultural Landscapes (D.B.A. Thompson, A.J. Hester and M.B. Usher, eds) pp. 344-361. HMSO, Edinburgh.
Robertson, G.P. (1987). Geostatistics in Ecology, interpolating with known variance. Ecology 68, 744-748.
Rodwell, J. ed., (1991). British Plant Communities. Volume 2. Mires and Heaths. Cambridge University Press, Cambridge.
Rodwell, J. ed., (1992). British Plant Communities. Volume 3. Grasslands and Montane Communities. Cambridge University Press, Cambridge.
Rossi, R.E., Mulla, D.J., Journel, A.G. and Franz, E.H. (1992). Geostatistical tools for modeling and interpreting ecological spatial dependence. Ecological Monographs 62, 277-314.
Sydes, C. and Miller, G.R. (1988). Range management and nature conservation in the British Uplands. In Ecological change in the uplands, (M.B. Usher and D.B.A. Thompson, eds) pp. 323-337. Blackwell Scientific Publications, Oxford.
Taylor, L.R. (1984). Aggregation, variance and the mean. Nature 189, 732-735.
Thompson, D.B.A., MacDonald, A.J., Marsden, J.H. and Galbraith, C.A. (1995). Upland moorland in Great Britain: a review of international importance, vegetation change and some objectives for nature conservation. Biological Conservation 71, 163-178.
Tudor, G.J. and Mackey, E.C. (1995). Upland land cover change in post-war Scotland. In Heaths and Moorland: Cultural Landscapes, (D.B.A. Thompson, A.J. Hester and M.B. Usher, eds) pp. 28-42. HMSO, Edinburgh.
21
Welch, D. (1984). Studies in the grazing of heather moorland in north-east Scotland. III. Floristics. Journal of Applied Ecology 21, 209-225.
Welch, D. and Scott, D., (1995). Studies in the grazing of heather moorland in north-east Scotland. VI. 20-year trends in botanical composition. Journal of Applied Ecology 32, 596-611.
22
Table 1. Areas of total moorland, dominant and sub-dominant heather and the consequential number of quadrats in each stratum for each grazing unit.
Grazing Unit Total area (ha) Moorland area surveyed (ha) Number of quadrats
Quadrats were distinguished according to those that had a large membership value for
a specific group (value >0.66, maximum value 1), and those that were intermediate
31
between several groups. The latter are marked with an open symbol on Figure 1 and
were assigned to the group for which they had the largest membership value.
A fuzzy analysi of the vegetation on Areas 3 and 4 at ADAS Pwllpeiran identified
two distinct plant groups, one dominated by Vaccinium and Nardus (Table 1: group
6) and a second dominated by Calluna, Vaccinium and Nardus (Table 1: group 7).
Response of different plant communities to reduced grazing
Significant temporal differences in ordination axis 1 quadrat scores were
recorded for both areas 1 (1990 - 1997) and 2 (1990-1995) at adas redesdale
(f5,290= 5.22, p<0.001 and f5,245= 5.085, p<0.001 for areas 1and 2 respectively),
suggesting an overall change in vegetation composition on both study areas. In
both cases, there was also a significant interaction between plant community
group and year (f20,290=2.91, p<0.001 and f20,245= 1.713, p<0.001 for areas 1 and 2
respectively), suggesting that plant groups differed in their response to the
stocking treatments imposed on each area.
Analysis of between-year differences for the five plant community groups on Area 1
(stocking level 1.5 sheep/ha) revealed no significant changes in community
composition for the Calluna-dominated (group 1), Molinia-dominated (group 2) or
Calluna/Eriophorum dominated (group 4) vegetation. There was, however, a
significant decrease in the Axis 1 scores for Nardus-dominated vegetation (group 3),
suggesting a shift in the composition of this group towards Molinia-dominated
vegetation (Tukey HSD, p<0.01). A change was also detected in the mean Axis 1
scores for mixed grassland (group 5). These increased significantly for the period
1990-1994 (Tukey HSD, p<0.04) indicating an increased dominance in rushes,
particularly Juncus effusus, within the mixed grassland vegetation (group 5). for No
32
significant shifts in community compostion were detected for any of the plant
community groups on Area 2 (stocking level 2.1 sheep/ha), despite the detection of an
overall change in Axis 1 ordination scores.
At ADAS Pwllpeiran, a significant change in community composition was observed
for Calluna/Vaccinium/Nardus (group 7) vegetation on Area 3 (1.4-1.5 sheep/ha),
suggesting a shift towards Calluna-dominated vegetation. No significant change was
observed in either Vaccinium/Nardus (group 6) or Calluna/Vaccinium/Nardus (group
7) vegetation on Area 4 (1.0-1.1 sheep/ha), although the analysis confirmed the
occurrence of two distinct communities on this site.
Response of different species to reduced grazing
No significant between-year differences in Calluna cover were observed for any of
the plant community groups on Area 1 (1.5 sheep/ha) at ADAS Redesdale, although
the ANOVA analyses confirmed that the five plant community groups did differ from
each other in terms of Calluna cover (F4,58= 25.929, p<0.0001). There was a
significant increase in Calluna cover on Area 2 (2.1 sheep/ha, F5,245=2.831, p<0.02),
but this was confined to Calluna/Eriophorum vegetation (group 4) where a significant
increase was observed four years after the start of the experiment (p<0.018). Calluna
cover in Calluna/Vaccinium/Nardus (group 7) vegetation remained unchanged on
Area 3 (1.4-1.5 sheep/ha) at ADAS Pwllpeiran and showed a marked decrease on
Area 4 (1.0-1.1 sheep/ha) after seven years (Tukey HSD test, p<0.01).
Molinia increased rapidly on Area 1 (1.5 sheep/ha), over the first two years of the
study, in Calluna- (group 1, Figure 2c), Molinia- (group 2, Figure 2a) and Nardus-
dominated (group 3, Figure 2b) vegetation. Although no further change was
observed after this initial period, the increased cover of Molinia persisted in both the
33
Nardus- and Molinia- dominated vegetation (Figure 2b&a) but did not persist in
Calluna-dominated vegetation. Molinia cover in Calluna-dominated vegetation in
1997 did not differ significantly from that in 1990. A significant increase in Molinia
cover also occurred across Area 2 (F5, 245=16.688, p<0.0012) but there was no
consistent directional change within any of the five plant groups. There was
insufficient Molinia present on either of the two study areas at ADAS Pwllpeiran to
enable formal analysis.
Vaccinium cover in Calluna/Vaccinium/Nardus (group 7) vegetation decreased after
four years on Area 3 (1.4-1.5 sheep/ha) at ADAS Pwllpeiran (F4,68= 2.8302, p<0.03),
but no consistent change in Vaccinium cover was observed on Area 4 (1.0-1.1
sheep/ha) in either this plant community group or in Vaccinium/Nardus (group 6)
vegetation. No significant change in the cover of Nardus was observed in either plant
groups 6 or 7 (Table 1) on either area.
34
Discussion
The response of seven different vegetation types, three dominated by Calluna and
four dominated by grasses, to reduced stocking levels is reported. Stocking levels on
three of the sites were consistent with those used in ESAs, whilst on the remaining
site, levels approximated those of normal farming practice.
Calluna cover was maintained on both sites with stocking levels of 1.4-1.5 sheep/ha
(equivalent to normal ESA Tier 1 stocking prescriptions – ADAS 1997b), but
declined by 14% on the lowest stocking level (1.0-1.1 sheep/ha - normally equivalent
to higher Tier 2 ESA prescriptions). A small but significant increase in Calluna was
recorded for Calluna /Eriophorum vegetation at the higher stocking rate (2.1
sheep/ha) but no other consistent changes in vegetation composition were recorded at
this site. Community analyses suggested enhancement of the dwarf shrub community
in only one plant group, Calluna/Vaccinium/Nardus, but this occurred as a result of a
decrease in the cover of Vaccinium rather than from an enhancement of Calluna.
These data indicate, therefore that whilst a reduction in stocking levels was effective
in maintaining the existing cover of Calluna, it was insufficient for stimulating a
significant enhancement in its abundance.
Greater differences were seen in the grass dominated communities. The rapid
increase in Molinia observed within both Molinia and Nardus-dominated
communities is commonly seen in overgrazed upland heath in response to a reduction
or removal of grazing (Hill et al. 1992). The different response reflects differences in
the physiognomy of grasses and dwarf shrubs with respect to grazing (Grime et al.
1988) The overall impact of reducing grazing on moorland vegetation change will,
therefore, depend on the respective competitive vigour of Calluna and the grass
35
species present. Where Calluna is vigorous, then it will out-compete grasses by over-
shading (Alonso et al., 1998). If, as is common on formerly over-grazed sites in
ESAs, Calluna growth is woody and old, then grasses such as Molinia and Nardus
can easily encroach into areas formerly dominated by dwarf-shrub (Watt, 1955;
Gimingham, 1972).
For all seven communities, it was evident that change in community composition was
very slow. This is unsurprising for established moorland vegetation, where expansion
of each species occurs as a result of vegetative growth rather than from the
germination and establishment of seed . In this circumstance, Calluna can only
increase in extent at the interface with other vegetation types. Significant seed
germination and establishment will only occur on burns or similar areas of bare
ground (Gimingham 1972) and occurs more slowly than vegetative regrowth
(Gardner et al., 1993).
Change was most readily detected at the level of the individual species rather
than at the community level. Change in the latter may be detectable only after
shifts have occurred among several species and at a larger scale. Focusing on
species that are characteristic of a community type or environmental variable
(critchley, 2000) may provide more reliable data on the impacts of management
rather than reliance on shifts in community composition. However, the initial
characterisation of the variation in vegetation composition present on a moor is
important, since as seen in this study, communities differ in their response to
changes in management.
Results from this study suggest that for formerly over-grazed moorland, the reduction
in stocking levels is beneficial in maintaining the existing cover of Calluna but does
36
little to enhance its extent. The latter is likely to require more pro-active management
measures such as burning, seeding and cutting. In each case, careful consideration of
the composition and condition of the moorland vegetation, particularly Calluna, will
be needed to ensure that additional management will be effective in enhancing the
cover of dwarf shrubs rather than that of aggressive less-preferred grass species that
tend to coexist with them.
Acknowledgements. The authors are grateful for comments and assistance from
ADAS staff, particularly Dr Sarah Hetherington, in the preparation of this
manuscript. This research was funded by the UK Ministry of Agriculture,
Fisheries and Food under contracts BD0101, 0106 and 1217.
37
ReferencesADAS. 1997a. Biological Monitoring of Moorland in the North Peak ESA 1988-
1996. Unpublished report to MAFF, April 1997.
ADAS. 1997b. Environmental Monitoring IN the Exmoor ESA 1993-1996. Unpublished report to MAFF, April 1997.
Alonso, I. & Hartley, S.E. 1998. Effects of nutrient supply, light availability and herbivory on the growth of heather and three competing grass species. Plant Ecology 137: 203-212.
Anderson, P. & Yalden, D.W. 1991. Increased sheep numbers and the loss of heather moorland in the Peak District, England. Biological Conserv. 20: 195-213.
Bardgett, R.D., Marsden, J.H. & Howard, D.C. The extent and condition of heather moorland in the uplands of England and Wales. Biological Conserv. 71: 155-161.
Barr, C.J. (Ed.) 1997. Current status and prospects of key habitats in England: Upland landscapes. Department of the Environment, Transport and Regions, London.
Clarke, J.L., Welch, D. & Gordon, I.J. 1995. The influence of vegetation patterrn on the grzing of heather moorland by red deer and sheep. II. The impact on heather. J. Appl. Ecol. 32: 177-186.
Critchley, C.N.R. 2000. Ecological assessment of plant communities by reference to species traits and habitat preferences. Biodiversity and Conservation 9: 87-105.
Equihua, M. 1990. Fuzzy clustering of ecological data. J. Ecol. 78: 519-534.
Gardner, S.M., Leipert, C. & Rees, S. 1993. Managing heather moorland: Impacts of heather burning on Calluna regeneration. Journal Environmental Planning and Management 36: 283-293.
Gimingham, C.H. 1972 Ecology of Heathlands. Chapman and Hall, London.
Grime, J.P., Hodgson, J.G. & Hunt, R 1988. Comparative Plant Ecology: a Functional approach to Common British Species. Unwin Hyman, London.
Hartley, S.E. & Amos, L. 1999. Competitive interactions between Nardus stricta L. and Calluna vulgaris (L.) Hull. The effects of fertilizer and defoliation on above- and below-ground performance. J. Ecol. 87: 330-340.
Hill, M.O., Evans, D.F. & Bell, S.A., 1992. Long term effects of exluding sheep from hill pastures in North Wales. J. Ecol. 80: 1-13.
Rodwell J.S. 1991. British Plant Communities: Volume 2. Mires and Heaths. Cambridge University Press, Cambridge.
Stace, C. 1991. New Flora of the British Isles. Cambridge University Press, Cambridge.
ter Braak, C.J.F. 1987. Ordination. In (eds) Jongman, R.H.G. ter Braak, C.J.F. & van Tongeren O.F.R. Data analysis in community and landscape ecology. PUDOC Wageningen, The Netherlands.
Thompson, D.B.A., Macdonald, A.J. Marsden, J.H. & Galbraith, C. A. (1995) Upland heather moorland in Great Britain, A review of international importance,
38
vegetation change and some objective for nature conservation. Biol. Conserv. 71: 163-178.
Tudor, G.J. & Mackey, E.C. 1995. Upland land cover change in post-war Scotland. In (eds.) Thompson, D.B.A, Hester, A.J. & Usher, M.B. Heaths and moorlands, pp. 28-42. HMSO Edinburgh.
Watt, A.S. 1995. Bracken vs heather, a study in plant sociology. J. Ecol. 35: 105-112.
Welch, D. 1984a Studies in the grazing of heather moorland in the north-east of Scotland I. Site descriptions and patterns of utilization. J, Appl. Ecol. 21: 179-195
Welch, D. 1984b Studies in the grazing of heather moorland in the north-east of Scotland II Response of heather J, Appl. Ecol. 21: 197-207.
39
Table 1. Characteristics of the different plant groups identified on Areas 1 and 2 at ADAS Redesdale (groups 1-5) and on Areas 3 and 4 at ADAS Pwllpeiran (groups 6 & 7).
GROUP CHARACTERISTICS
1. Calluna-dominated heath
Calluna-dominated vegetation with constant Carex nigra and frequent Molinia and Eriophorum vaginatum but at relatively low covers. Closest NVC community is M15, Scirpus cespitosus – Erica tetralix wet heath
2. Molinia - dominated heath
Dominant Molinia with frequent Calluna and C. nigra. Closest NVC community is M15, Scirpus cespitosus - Erica tetralix wet heath
3. Nardus - dominated heath
Predominantly grassy vegetation dominated by Nardus and characterised by species typical of dry grassy vegetation such as Agrostis spp., Anthoxanthum odoratum, Festuca ovina and Potentilla erecta. There is frequent Molinia but at low cover. Closest NVC community is U5, Nardus stricta - Galium saxatile grassland.
4. Calluna / Eriophorum heath
A Calluna-dominated group characterised by species typical of wet conditions such as E. vaginatum, E. angustifolium and Sphagnum spp. This community has affinities with M15 and M19, Calluna vulgaris-Eriophorum vaginatum blanket mire.
5. Mixed grassland No single dominant species although C. nigra is more common than other species. The composition is similar to group 3 but Nardus is not dominant, merely frequent at low covers. Juncus effusus frequency is also high but again at low covers. Closest NVC community is U5, with strong affinity to M15 and M23 - Juncus efusus/acutiflorus-Galium palustre rush pasture.
6. Vaccinium / Nardus heath
Dominated by grasses and constant Vaccinium myrtillis. Nardus stricta, Deschampsia flexuosa, Festuca sp. and Galium saxatile all have high constancy values
7. Calluna/Nardus /Vaccinium heath
Dominated by Calluna, Vaccinium and Nardus and including species from both wet (e.g. Scirpus cespitosus) and dry (e.g. D. flexuosa) heath vegetation.
40
Figure legends
Fig. 1. Detrended Correspondence Analysis of quadrat samples for Areas 1 (1.5
sheep /ha) and 2 (2.1 sheep/ha) at ADAS Redesdale. Eigenvalues for axes 1 and 2 are
0.54 and 0.23 respectively. The positions of species that are characteristic for each
plant community group are marked with a solid square and the species name. The
positions of the five plant groups identified from fuzzy clustering are marked with a
solid diamond and the group number (1-5). 1 is the Calluna-dominated heath group, 2
- the Molinia-dominated heath group, 3 - the Nardus-dominated heath group, 4 -
Calluna-Eriophorum dominated heath and 5 - mixed grassland.
Fig. 2. Mean percentage frequency of Calluna and Molinia in a) Calluna-dominated
heath, b) Molinia-dominated heath and c) Nardus-dominated heath on Area 1 (1.5
sheep/ha) at ADAS Redesdale. P-values indicate change in Molinia frequency
(Tukey HSD tests) between the year, above which the P-value is positioned, and the
baseline year of 1990. Results for Calluna were all not significant
41
Fig.1.
43
0
10
20
30
40
50
60
70
90 92 94 97Year
% fr
eque
ncy
Calluna Molinia
p<0.01ns
ns
a)
ns
Fig.2
44
0
10
20
30
40
50
60
70
90 92 94 97Year
% fr
eque
ncy
Calluna Molinia p<0.01 p<0.01p<0.01
b)
Fig.2.
45
0
10
20
30
40
50
60
70
90 92 94 97Year
% fr
eque
ncy
Calluna Molinia
p<0.01 p<0.01 p<0.01
c)
Fig.2
46
Techniques for the control of Molinia caerulea on wet heath after burning
S Y ROSS, S HARVEY, H F ADAMSON and A E MOON
ADAS Redesdale, Rochester, Otterburn, Newcastle-upon-Tyne NE19 1SB, UK
SUMMARY
The effectiveness of three treatments to control the competitive vigour of Molinia caerulea were compared on a Calluna vulgaris -
Molinia dominated wet heathland. The treatments were burning, burning with cutting or burning with herbicide application, along with
an unburned control. The effect of the treatment was dependent on the dominant species present in each of the vegetation types sampled.
Where Molinia was the dominant species, the burning and herbicide treatment reduced Molinia cover in the short term, but the effect
were not sustained. Other treatments showed little effect. Burning stimulated seedling germination of Calluna where it was a prominent
component of the vegetation, and additional cutting or herbicide application increased germination in the mixed heathland vegetation
only. This effect was, however, also short-lived. Repeated post-burn herbicide application may have produced more sustained responses
in Molinia and in the regeneration of Calluna.
INTRODUCTION
47
There has been a reduction in the amount and quality of heather moorland across the UK over the last 50 years. The associated decrease
in Calluna and other dwarf shrubs has often been accompanied by increase in competitive grass species such as Molinia caerulea. In
response to these changes a key objective of the Ministry of Agriculture, Fisheries and Food (MAFF) agri-environment schemes is to
reduce grazing levels and to encourage the maintenance and enhancement of heather on upland moors. Burning management is often
included within these schemes to rejuvenate Calluna, but burning may also increase the competitive advantage of M. caerulea. This
experiment aims to investigate the effectiveness of different management methods in reducing the cover and competitive vigour of
Molinia whilst maintaining, and potentially enhancing, the cover of Calluna.
METHODS
Replicated plots were established in an area of wet heath dominated by Calluna and Molinia, under a stocking rate of 1.5 ewes/ha within
ESA Tier 1 stocking levels. Each plot allocated a different treatment, as follows: (i) burning and cutting to a height of 8 cm, (ii) burning
only, (iii) burning and ‘Fusilade’ herbicide application, and (iv) a control.
In July 1995 five 1m2 permanent quadrats were established in each plot prior to treatment application. Each quadrat was divided into 100
10cm2 cells and three baseline floristic assessments carried out: (i) first hits by cross-wires (ADAS, 1999), (ii) frequency of Molinia and
Calluna, and (iii) dominant plant species. These assessments were repeated in July 1997, 1998 and 1999, after treatments had been
applied. Post-burn regeneration of Calluna was assessed by counting the number of seedlings in 20 10 cm2 cells per quadrat, twice yearly
between June 1996 (before cutting or herbicide treatments) and May 1998.
48
RESULTS
Changes in species composition
DCA ordination and fuzzy clustering analysis (Equhia 1990) highlighted three main vegetation groups from the 1995 baseline vegetation
data (Fig. 1). Group 1 showed a higher proportion of grass species, dominated by Molinia, group 2 was dominated by Calluna, while
group 3 showed a mixture of heathland species. The response of these vegetation groups to the treatments applied was assessed from
1997 to 1999 using the change in DCA ordination axis 1 scores. Overall, there was a significant difference in axis 1 scores between the
four treatments (F(3) = 5.775; P = 0.0443), and trends could be identified within each vegetation group. In 1997 for all treatments, except
the control, the Calluna-dominated group 2 quadrats moved toward the Molinia and Deschampsia flexuosa ordination points. The mixed
heath quadrats (group 3) showed a similar but smaller shift. In all treatments the Molinia-dominated vegetation (group 1) showed very
little change in ordination space. In 1998 and 1999 there was some indication that vegetation groups 2 and 3 moved slightly to the right
of the ordination suggesting a change back toward Calluna-dominated vegetation.
Change in frequency and dominance of Calluna and Molinia
There was no significant difference in the frequency of Molinia between the four treatments (F(3) = 0.804; P = 0.5430), with this species
recorded in at least 70 of the 100 cells. There was, however, some indication of a decline in Molinia in the Molinia-dominated vegetation
type directly after herbicide application (Fig. 2 treatment 3 1997), although this was not statistically significant. There was an overall
49
significant reduction in Calluna between the treatments (F(5) = 63.566; P = 0.0002). This decline was seen in the three treatments after
burning for both the Calluna-dominated and the mixed heath vegetation types (Fig. 3).
Overall, there were few significant changes in species dominance within the three vegetation types. A significant change from Calluna to
Deschampsia dominance was observed in the Calluna-dominated vegetation group, between 1995 and 1998 (H(22) = 8.729; P = 0.0331).
A significant change from Calluna to Carex nigra was seen for the same year for the mixed heath vegetation type (H(18) = 8.826; P =
0.0317). There was some change in dominance from Calluna to Molinia in the mixed heath vegetation, but this was not statistically
significant (H(28) = 7.465; P = 0.0585).
Regeneration of Calluna
Regeneration was greatest in the Calluna and mixed heath vegetation types, while the Molinia-dominated vegetation group showed
reduced seedling germination overall. Within the Calluna vegetation type regeneration was highest after burning and herbicide
application (treatment 3) in the few months directly after treatment application, although this difference was not statistically significant.
For the mixed heath vegetation type regeneration under burning plus cutting, and burning plus herbicide treatments were significantly
greater than for burning alone (H(120) = 7.080; P = 0.0290) (Fig. 4).
DISCUSSION
50
The Calluna canopy was reduced in all three treatments of burning, or burning with additional cutting or herbicide application, as much
of the aboveground biomass was removed by the burn. The trend was similar for all three treatments indicating burning had the greatest
impact on Calluna with little additional effect of cutting or herbicide treatments.
Molinia was not significantly reduced in any of the treatments over the five years. There was, however, some indication that burning
with herbicide application reduced Molinia in the Molinia-dominated vegetation, but effect was not sustained suggesting that repeated
application of the herbicide may need to be considered for longer term control.
Burning has been shown to increase the competitiveness of M. caerulea in heathland due to nutrient release and reducing competition for
light (Heil & Bruggink, 1987; Alonsi & Hartley, 1998). Our results suggest burning alone did not increase the frequency of Molinia.
There was, however, some increase in its dominance in the mixed heath vegetation for two years after treatment application, but this was
not sustained. Species composition data also suggested that although Calluna-dominated and mixed heath areas initially become more
similar to Molinia-dominated areas after treatments, they begin to move back toward a more Calluna-rich species composition after three
years.
Heather regeneration after burning is typically highly variable, and dependant on many factors including plant age, site characteristics
and intensity of the burn. Burning stimulated seedling germination where Calluna was initially present as more than 40 % of the
vegetation canopy. Calluna is not expected to achieve dominance until several years after a burn (Gimingham, 1958; McFerran,
McAdam & Montgomery, 1995), and as Calluna was still widely found in the sward giving potential for re-establishment as a dominant.
Burning with cutting or herbicide application increased the numbers of seedlings, with the greatest effect being seen for the mixed heath
vegetation. The effect was short term, and all treatments showed lower seedling survival in the second year (1997). This may be 51
attributed to seedling die back over the winter between October 1996 and May 1997, or increased grazing/trampling on the Calluna
seedlings during the winter months (Welch, 1984). Where Calluna cover was low, and Molinia dominated the vegetation, seedling
germination was much reduced under all treatments, and it is unlikely that Calluna would re-establish as a dominant.
ACKNOWLEDGEMENTS
The authors would like to thank Jeff Byrne, Anna Gundry, Fiona Kennedy and David Oatway for help collecting data, and Dr Sarah
Gardner for comments on the manuscript. The work was funded by MAFF, as part of project BD1218.
REFERENCES
ADAS 1999. STATISTICAL ASSESSMENT OF THE TECHNIQUES FOR MONITORING SPECIES COMPOSITION IN
UPLAND PLANT COMMUNITIES. UNPUBLISHED REPORT TO MAFF. 46PP.
ALONSI I, HARTLEY S E. 1998. EFFECTS OF NUTRIENT SUPPLY, LIGHT AVAILABILITY AND HERBIVORY ON THE
GROWTH OF HEATHER AND THREE COMPETING GRASS SPECIES. PLANT ECOLOGY 137: 203-212.
DIEMONT W H, LINTHORST HOMAN H D M. 1989. RE-ESTABLISHMENT OF DOMINANCE BY DWARF SHRUBS ON
GRASS HEATHS. VEGETATIO 85: 13-19.
EQUHIA M. 1990. FUZZY CLUSTERING OF ECOLOGICAL DATA. JOURNAL OF ECOLOGY 78: 519-534.
52
GIMINGHAM C H. 1958. BIOLOGICAL FLORA OF THE BRITISH ISLES: CALLUNA VULGARIS (L.) HULL. JOURNAL OF
ECOLOGY 48: 455-483.
HEIL G W, BRUGGINK M. 1987. COMPETITION FOR NUTRIENTS BETWEEN CALLUNA VULGARIS AND MOLINIA
CAERULEA L. OECOLOGIA (BERLIN) 73: 105-107.
MCFERRAN D M, MCADAM J H, MONTGOMERY W I.. 1995. THE IMPACT OF BURNING AND GRAZING OF
HEATHLAND PLANTS AND INVERTEBRATES IN COUNTY ANTRIM. BIOLOGY AND ENVIRONMENT - PROCEEDINGS
OF THE ROYAL IRISH ACADEMY 95B: 1-17.
WELCH D. 1984. STUDIES IN THE GRAZING OF HEATHER MOORLAND IN NORTH-EAST SCOTLAND. I. SITE
DESCRIPTIONS AND PATTERNS OF UTILISATION. JOURNAL OF APPLIED ECOLOGY 21: 179-195.
Fig. 1 Detrended Correspondence Analysis ordination plot for pre-treatment (1995) 1st hits data, showing the three vegetation groups
identified through fuzzy clustering techniques in relation to plant species ordination. The partition coefficient is 0.4983. Species as
follows; Cv - Calluna vulgaris, Et - Erica tetralix, Df - Deschampsia flexuosa, Ns - Nardus stricta, Mc - Molinia caerulea, Vm -
Vaccinium myrtillus, Cn - Carex nigra
Fig. 2 The mean (+/- SE) frequency of Molinia for the Molinia-dominated vegetation group, over time for each treatment as follows; 1 -
burning and cutting, 2 - burning only, 3 - burning and herbicide application, 4 - control.
53
Fig. 3 The mean (+/- SE) frequency of Calluna for (A) the Calluna-dominated and (B) the mixed heath vegetation groups, over time for
each treatment, as follows; 1 - burning and cutting, 2 - burning only, 3 - burning and herbicide application, 4 - control.
Fig. 4 The mean (+/- SE) germination of Calluna for the mixed heath vegetation group over time for each treatment as follows; 1 -
burning and cutting, 2 - burning only, 3 - burning and herbicide application.
FIG 1
54
FIG 2
55
FIG 3
56
FIG 4
57
TECHNIQUES FOR THE CONTROL OF PURPLE MOOR-GRASS
S.Y. ROSS AND H.F. ADAMSON
ADAS Redesdale, Otterburn, Newcastle-upon-Tyne, NE19 1SB, UK
INTRODUCTION
There has been a reduction in UK heather moorland over the last 50 years (Thompson et al., 1995), which includes both a loss of area,
and a decline in quality (Bardgett et al., 1995). The decrease in heather (Calluna vulgaris) can be accompanied by increases in purple
moor-grass (Molinia caerulea), particularly after burning (e.g. Welch & Scott, 1995). Such a change in the balance of species affects
both heathland grazing quality and its biodiversity. This experiment investigates the effectiveness of three management techniques in
reducing the frequency and dominance of Molinia on heathland, under two stocking rates that fall within Environmentally Sensitive Area
(ESA) prescriptions for upland areas.
MATERIALS AND METHODS
Experimental plots were established under stocking rates of 1.5 ewes/ha (Area 1: ESA Tier 1 prescription), and 0.66 ewes/ha (Area 2:
ESA Tier 2 prescription). On each area, three replicated blocks were split into four 10 m 2 plots, with a 2 m discard, and each plot
randomly allocated a different treatment. The treatments were (1) burning + cutting to 8 cm high, (2) burning, (3) burning + herbicide,
and (4) a control receiving no treatment. Burning was carried out in April 1996, and additional cutting was applied to plot 1, and
58
fluazifop-P-butyl herbicide (‘Fusilade’ 250EW, 1.5 litres/ha) applied to plot 3 in July 1996. In July 1995 five 1m 2 permanent quadrats
were established in each plot prior to treatment application. They encompassed three vegetation types; Molinia-dominated, Calluna-
dominated and mixed Calluna/Molinia vegetation. Quadrats were divided into 100 10 cm2 cells, and baseline data of Molinia frequency
(presence/absence) and dominance were recorded in each cell. These assessments were repeated annually from 1996 to 1999 after
treatments had been applied. Frequency data were arcsine transformed and analysed using repeated measures ANOVA, with vegetation
group entered as a covariate, while dominance data were analysed separately for each vegetation group using Kruskal-Wallis tests.
RESULTS
Changes in Molinia frequency
For all three vegetation types there was no significant difference in Molinia frequency between the treatments (P = 0.5430) for Area 1
(1.5 ewes/ha). There was, however, a decline in Molinia in the Molinia-dominated vegetation type directly after herbicide application in
1997, although this was not statistically significant. For the lower stocking rate (Area 2: 0.66 ewes/ha), there was no significant
difference in the frequency of Molinia between treatments (P = 0.8896), and no indication of a 1997 decline in Molinia under the
burning plus herbicide treatment.
Changes in Molinia dominance
59
Overall, Molinia dominance across all three vegetation types was not significantly different (P = 0.2927) between treatments, for higher
stocking rates (Area 1). There was, however, some reduction in Molinia dominance in the Molinia-dominated vegetation, and this
reduction was sustained until 1999 (Fig. 1). Under lower stocking rates (Area 2) there was a significant increase in Molinia dominance
compared with the control (P = 0.0017), and this was particularly evident for the Molinia-dominated vegetation.
60
Fig. 1. Change in the mean dominance (+/-SEM) of Molinia over time, in the Molinia-dominated vegetation, for Area 1 (1.5
ewes/ha). Treatments are 1 – burning and cutting; 2 – burning only; 3 – burning and herbicide; 4 – control.
0.00
20.00
40.00
60.00
80.00
100.00
1 2 3 4
Treatment
% d
omin
ance
of M
olinia
1995
1998
1999
61
DISCUSSION AND CONCLUSIONS
The management treatments showed little effect on reducing Molinia frequency in either
Area 1 (1.5 ewes/ha) or Area 2 (0.66 ewes/ha). Most management techniques increased
Molinia dominance under both stocking rates, particularly burning alone. There was,
however, some indication that under 1.5 ewes/ha, burning followed by herbicide
application reduced both Molinia dominance and frequency in Molinia-dominated
vegetation. The reduction in frequency was only sustained for the year after herbicide
application, although dominance remained lower until 1999. This suggests that repeated
application of herbicide would need to be considered for longer-term control of Molinia,
although long-term effects on heathland are not known and require further research. The
effectiveness of this management technique strongly depends on both the dominance of
Molinia in the vegetation prior to treatment, and stocking rates applied after treatment.
REFERENCES
Bardgett R.D, Marsden J.H, and Howard J.H. (1995) The extent and condition of heather on moorland in the uplands of England and Wales. Biological Conservation, 71, 155-161.Thompson D.B.A., MacDonald A.J., Marsden J.H., and Galbraith C.A. (1995) Upland heather moorland in Great Britain: a review of international importance, vegetation change and some objectives for nature conservation. Biological Conservation, 71, 163-178.Welch D. and Scott D. (1995) Studies in the grazing of heather moorland in north-east Scotland. VI. 20-year trends in botanical composition. Journal of Applied Ecology, 32, 596-611.
62
The effects of summer-only grazing on the post-burn recovery of a wet heath,
The effectiveness of thirteen techniques for monitoring change in vegetation condition and composition on upland heather moor is evaluated. The evaluation is based on standardised criteria assessing the type of information collected, the objectivity and reliability of the measurements, and the resource requirements for each technique. Two main approaches are currently used for monitoring change in moorland vegetation, namely, the assessment of grazing pressure or of botanical composition. Techniques measuring grazing pressure provide information on the impact of one of the drivers of change on upland moor but little information on the response of the vegetation. Techniques measuring botanical composition provide information on the vegetation response, but few clues to pinpoint the key drivers of change. Ideally, monitoring programmes should therefore aim to include both approaches. Techniques using a proportionate measure of grazed shoots and a frequency count of species presence within fixed sampling unit, were seen to provide the most objective and reliable measurements of vegetation change. These two criteria are of particular importance in large-scale and/or long-term monitoring programmes that employ several field staff and aim to assess the effectiveness of management and/or policy initiatives to conserve upland moor.Keywords: heather moorland, ecological monitoring, grazing pressure, environmentally sensitive areas
95
IntroductionUpland heather moorland is a UK habitat of international conservation importance (Thompson, et al.,
1995), which has declined in area and condition since 1945 (Bardgett et al., 1995; Tudor and Mackey,
1995). A primary reason for the decline has been overgrazing, particularly by sheep (Anderson & Yalden,
1981; Sydes and Miller, 1988; Thompson et al. 1995). Consequently, the habitat has become the subject
of a large number of conservation initiatives, both local and national. These include the implementation of
national agri-environment schemes, such as the Environmentally Sensitive Areas (ESAs) scheme, arising
from EC regulation 2078/92, which aim to promote environmentally sustainable farming practices.
In these schemes and other moorland restoration projects there is increasing pressure on policy makers,
land managers and conservation agencies to use survey and monitoring techniques to ensure correct
management plans are put in place, and to measure the extent to which conservation objectives are being
met. Here we use the precise definitions of Hellawell (1992) for survey and monitoring, in which the key
difference between the two is that monitoring is undertaken with the aim of determining the extent to
which a value/ range of values deviates from a predetermined norm. Survey simply involves the
measurement of values without any preconception of what the findings will be.
Monitoring and survey techniques for heather moorland have evolved quite rapidly since the 1960s when
productivity of vegetation was a major research topic as evidenced by the International Biological
Programme (IBP). Then studies concentrated on heather as a source of forage and measured standing crop
and the differences between grazed and ungrazed plots (Moss, 1969; Grant, 1971). Subsequent grazing
studies focused on individual heather shoots and the proportion of shoot biomass removed by grazing
(Grant et al., 1978; Grant et al., 1981). Succession from heather moorland to acid grassland was becoming
increasingly obvious in the 1970s and 80s (Bunce, 1989; Felton and Marsden, 1990), and hence there was
a need to determine the extent to which the condition of heather was declining and graminoids were
spreading. In response to this, since 1990, there has been a proliferation of techniques for monitoring
trends in heather condition and extent. These include methods for monitoring end of winter grazing
pressure and utilisation (e.g. Welch, 1984a; Henderson et al., 1996), methods based on heather form or
structure (e.g. Gill and Scott, 1995; Rae, 1995) and monitoring of heather cover and species composition
(e.g. Welch, 1984b; Welch and Scott, 1995; English Nature, 1996). In addition to this there has been
development of site ‘condition assessment’ or ‘indicator’ methodologies which aim to obtain an holistic
96
assessment of moorland condition from a single, relatively rapid survey (Macdonald et al., 1997; Jerram
and Drewitt, 1997).
The aim of this paper is to evaluate the range of techniques available to conservation practitioners, policy
makers and researchers in order to inform the development of monitoring activities within upland
restoration projects. Previously no overall review of upland monitoring techniques has been undertaken,
although there has been a comparison of methods for estimating levels of grazing on heather (Armstrong
and Macdonald, 1992). There have also been studies comparing different methods for estimating botanical
composition in other habitats (e.g. Friedel and Shaw, 1987; Everson and Clarke, 1987; Brakenhielm and
Quighong, 1995) but none have addressed heather moorland. This paper will address these issues by a)
ranking the wide range of techniques available against standardised evaluation criteria and b) testing the
sensitivity of methods for estimating botanical composition in a replicated using data collected from field
experiments. In doing so we hope to provide scientists and policy makers with the information they need
to identify the methods most appropriate for use in large-scale heather monitoring programmes or field
trials.
97
Methodology
The study was conducted in three stages. Firstly, information was collated on current field techniques and
the main aspects of their application summarised. The techniques were then evaluated using a ranking
procedure against a number of clearly defined criteria. Finally, a subset of methods for estimating species
composition were compared statistically with respect to their ability to detect change in species
abundance using data from existing field experiments.
Review of techniques
Information on field techniques was extracted from the scientific literature and relevant research and
monitoring project reports. The review was concerned with all field survey techniques that could
practicably be applied at the scale of the moorland management unit (area 30-300 ha). Hence detailed
physiological measurements and other more labour intensive methods were excluded, as were remote
sensing techniques. The key aspects to each methodology are summarised, including the variables
measured, typical size of area assessed, and the size of the individual sampling units used (e.g. quadrats,
plots, management unit etc.).
Evaluation of techniques
As several of the techniques measure different aspects of the moorland vegetation, there was no
experimental way of directly comparing the techniques. A ranking procedure was therefore devised to
enable the comparison of techniques. This procedure included a number of pre-defined and standardised
criteria to enable provision of a transparent and robust evaluation of the different techniques.
Each technique was ranked according to the pre-defined criteria (see Table 1). These included the plant
attributes measured, the precision (and hence to some degree, sensitivity) of the data collected, the
objectivity of the measurements, the amount of training/quality control required to ensure consistent and
accurate use by field staff, and the time required for a single measurement.
In any large-scale monitoring exercise the availability of resources is a key factor influencing the choice
of methods adopted. Therefore, in order for judgements to be made as to the cost-effectiveness of the
techniques, a standardised field ‘scenario’ was developed which allowed judgements to be made as to the
quantity of information collectable within a defined time period. The scenario used in this exercise was
based on a 500 ha management unit of 50:50 heather/grass covered moor, with moderate grazing pressure.
98
For each technique, a random sampling process was assumed and a maximum of two staff days (16 hours
survey time) was permitted.
The ranks for each criterion were not summed in order to identify a single technique as the best. Rather
the ranking system provided information against specific criteria so that once the most important criteria
for a particular study have been identified, it would be possible to select the appropriate technique.
Statistical comparison of techniques for measuring species composition
Existing data from a grazing experiment conducted in the English and Welsh Uplands (Rushton et al.,
1996; ADAS, 1998) were used to compare four different methods that measure change in botanical
composition. The methods were applied to permanent quadrats placed in three contrasting vegetation
types situated within different management units. Each quadrat was sampled in 1995 and again in 1997.
The vegetation types. sampled were Calluna-Molinia vegetation, similar to the M15 Scirpus cespitosus-
Erica tetralix wet heath of the National Vegetation Classification (NVC, Rodwell, 1992), at ADAS
Redesdale (30 quadrats) in the Northumberland National Park, Calluna-Nardus vegetation (NVC H12
Calluna-Vaccinium heath) at ADAS Pwllpeiran (30 quadrats) in the Cambrian Mountains ESA, and
Festuca-Nardus-Vaccinium vegetation, similar to NVC U4e - the Vaccinium-Deschampsia sub-
community of Festuca-Agrostis-Galium grassland and U5 - Nardus-Galium grassland, also at Pwllpeiran
(24 quadrats). Quadrats were 4m², each comprising four 1 m x1 m quadrants. Each quadrant was further
subdivided into a cellular grid of one hundred 100 cm² cells. Each sampling technique tested, utilised the
cellular grid in each quadrant and data were amalgamated to the 4 m² scale. All analyses were conducted
at this scale. The four sampling techniques tested are described below:
i) Frequency of key species: The presence (rooted or canopy) or absence of above-ground vegetation for
each of seven key species was recorded in each 100 cm² cell. For this study, the key species chosen were:
moorland in Great Britain: A review of international importance, vegetation change and some
objectives for nature conservation. Biological Conservation 71, 163-178.
Tudor, G. J. and Mackey, E. C. (1995). Upland land cover change in post-war Scotland. In: Heaths and
moorland: cultural landscapes (D.B.A. Thompson, A.J. Hester and M.B. Usher, eds.), pp. 28-42.
Scottish Natural Heritage / HMSO, Edinburgh,.
112
Welch, D. (1984a). Studies in the grazing of heather moorland in north-east Scotland. II Response of
heather. Journal of Applied Ecology 21, 209-225.
Welch, D. (1984b). Studies in the grazing of heather moorland in north-east Scotland. III Floristics.
Journal of Applied Ecology 21, 197-207.
Welch, D. and Scott, D. (1995). Studies in the grazing of heather moorland in north-east Scotland. VI. 20-
year trends in botanical composition. Journal of Applied Ecology 32, 596-611.
Welch, D., Scott, D. and Staines, B. W. (1996). Study on effects of wintering red deer on heather moorland.
Report on work done 1995, and analysis of data collected 1992-1995. Report to Scottish Natural
Heritage, Edinburgh.
113
Table 1 Criteria used for assessing heather monitoring techniques
Criterion Definition Ranking system
Plant attribute The feature of the plant that is measured e.g. height, cover, form, utilisation
N / A
Season Time of year when technique should be applied N / A
Area Area over which individual measurements are taken e.g. 1m2 quadrat, individual points / plants / shoots
N / A
Time Time needed to record one measurement 1
2
3
> 30 min
5-30 min
< 5 min
Precision of information
The type of information measured:
1 Nominal exclusive categories e.g. presence/absence, male/female
2 Ordinal - a rank order with unequal intervals between the scale e.g. DAFOR, DOMIN
3 Discontinuous interval e.g. % cover
4 Continuous interval e.g. height
1
2
3
4
categorical
poor sensitivity
quite sensitive
most sensitivea
Objectivity 1. An estimate requiring an observer judgement of quantity e.g. % cover
2. A quantitative measurement based on a standardised & universal scale e.g. individual counts, height in cm
1
2
less objective
more objective
Training requirement
Time taken to learn techniqueb 1
2
3
> 2 days
1-2 days
< 1 day
Quality Assurance
Time needed to verify consistency and standards of operation between observersb.
1
2
3
> 2 days
1-2 days
< 1 daya Sensitivity to quantitative change in abundance or extentb Rank estimates based on authors’ experience
114
Table 2 Techniques of assessing heather moorland condition and examples of use
Code Attribute Season Area Reference
Methods measuring grazing pressure/utilisation of heather
U1 % shoots grazed End of winter
1m² Welch and Scott (1995)*, Welch et al. (1996)*
U2 % shoots grazed + index of biomass lost
End of winter
1m² Henderson et al. (1996)*
U3 % shoots grazed + biomass utilisation
End of winter
0.5m² ADAS (1997a)*
U4 Annual off-take Summer & winter
0.25m² Moss (1969), Moss and Miller (1976)
Methods measuring heather growth form, age or structure
F1 Growth form in response to grazing
Typically Summer
25ha Gill and Scott (1995), Rae (1995)
F2 Cover and height index Typically Summer
100ha Bardgett et al. (1995)
F3 Old/young shoot counts Typically Summer
<0.1m² Smith et al. (1997)*
Monitoring botanical composition
GCH1 Cover, frequency or biomass-based estimates of botanical composition
Summer Typically 0.5-4m²
Welch and Scott (1995)*, Welch et al. (1996)*, ADAS (1997b), Critchley and Poulton (1998)
GF1 % cover and growth form
Summer 4m² English Nature (1996), FRCA (1997)
* authors also use other monitoring techniques
115
Table 3 Results of ranking assessment of heather monitoring techniques (see Table 1 for interpretation of the ranking system)
Method Time No. of measures
Precision Objectivity Training QA
U1 3 100 3 1 3 2
U2 2 60 3 1 3 2
U3 2 100 3 2 2 2
U4 1 40 4 2 3 2
F1 1 1 survey 2 1 2 2
F2 1 1 survey 2 1 2 1
F3 2 100 4 2 3 2
GCH1 (i)b 2 50 2 1 3 2
GCH1 (ii) 2 50 3 1 3 2
GCH1 (iii) 2 50 3 2 3 3
GCH1 (iv) 3 200 points 1 2 3 3
GCH1 (v) 1 1 3 2 3 3
GF1 1 1 survey 2 1 2 1a Area refers to the size of the sampling unit not the total area over which the technique may be appliedb GCH1 includes several options and each is evaluated in turn, i) percentage cover - Domin scale, ii) percentage cover, iii) frequency - first hits in a gridded quadrat, iv) frequency of point touches, v) presence or absence in nested quadrats or ADAS stand.
116
Table 4 Summary of the effectiveness of each assessment technique in predicting the variation in the standard datasets for each species. Variation accounted for is represented as *** >80%, **>70%, *>60%, ---- <50%
Species Vegetation type First Hit Domin Dominant species
Calluna vulgaris C. vulgaris-N. stricta
C. vulgaris-M. caerulea
***
**
***
***
***
***
Empetrum nigrum C. vulgaris-N. stricta
C. vulgaris-M. caerulea
**
***
----
**
----
*
Erica tetralix C. vulgaris-N. stricta
C. vulgaris-M. caerulea
-----
***
----
**
----
***
Molinia caerulea C. vulgaris-N. stricta
C. vulgaris-M. caerulea
N. stricta-V. myrtillis
*
*
---
----
*
----
----
----
----
Nardus stricta C. vulgaris-N. stricta
C. vulgaris-M. caerulea
N. stricta-V. myrtillis
***
***
**
*
**
***
**
**
**
Vaccinium myrtillis C. vulgaris-N. stricta
C. vulgaris-M. caerulea
N. stricta-V. myrtillis
----
*
*
*
***
***
----
----
----
V. oxycoocus C. vulgaris-M. caerulea *** ---- ***
No. of species with > 60% variation accounted for in each vegetation type
C. vulgaris-N. stricta
C. vulgaris-M. caerulea
N. stricta-V. myrtillis
4/6
7/7
2/3
3/6
6/7
2/3
2/6
5/7
1/3
Overall 13/16 11/16 8/16
117
Table 5. Summary of results of repeated measures ANOVA for different species within contrasting vegetation types using three different field methods. CM – Calluna-Molinia vegetation, CN – Calluna-Nardus vegetation, NV-Nardus-Vaccinium vegetation. Figures are P values. ns-non-significant at P<0.05. n/a – data not available. ‘Sensitivity’ relates to proportion of significant results from all tests for each method. Vegetation Type