• conservation management;
  • decorana;
  • Huisman;
  • multivariate analysis;
  • Olff and Fresco modeling;
  • post-fire succession;
  • species response curves;
  • variation partitioning


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

1. The role of prescribed burning of vegetation to manage fire risk is controversial in a variety of situations worldwide. It is becoming more topical (i) as a result of potential global warming where the risk of wildfire might increase and (ii) because fire might affect the various ecosystem services provided in a different way. Where prescribed fire is used, ecologists need to know the impact on biodiversity (post-fire recovery) and on provisioning and regulating services such as water collection and carbon sequestration. Here, we assess the effect of prescribed burning on plant community composition and its component species at the regional scale of the Peak District, where the moorland vegetation is severely degraded.

2. Species cover (%) was assessed on five moors with respect to elapsed time since prescribed burning and vegetation height. A stratified random method was used to choose burn patches covering a range of ages since burning; quadrats were then sampled randomly within these patches over a 3-year period. Detrended correspondence analysis was used to relate species composition to significant environmental variables, and variation partitioning was used to assess their relative contribution. Response curves were produced for the major species with respect to elapsed time since burning and vegetation height.

3. The species ordination produced two gradients, (i) a continuum from a graminoid-dominated vegetation to one dominated by Erica tetralix, Vaccinium myrtillus and Rubus chamaemorus and (ii) a post-fire growth response of the dominant species, Calluna vulgaris. Species composition was more highly correlated with vegetation height than elapsed time since burning. The environmental variables explained 15·2% of the variation.

4.Calluna vulgaris was the only species to show an increasing response after burning; all others showed an increase immediately after burning, but then they either decreased or showed a unimodal/skewed response. Most other species were restricted to vegetation <40 cm height and 20–25 years after burning.

5.Synthesis and applications. We found two major results of importance to policy makers and land managers: (i) that prescribed burning maintains species diversity in the immediate post-burn phase, and (ii) as the vegetation ages and increases in height, most species disappear and the vegetation becomes dominated by C. vulgaris. From a policy perspective, prescribed burning (or some other disturbance) is needed to maintain burning and a no-burn policy will result in a low-diversity, C. vulgaris-dominated vegetation. As vegetation height is the easiest measure for land managers to use in judging when to burn, we recommend moorland vegetation be burned before it reaches 25 cm in height to maintain the pre-burn complement of plant species. If the rotation allows the vegetation to become much taller (>40 cm), then most species will be lost and they will have to colonize after subsequent fires from the seedbank or from the surrounding area.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The upland moors in Great Britain (GB) have a high conservation value of international significance (Ratcliffe & Thompson 1988) with six moorland plant communities virtually confined to the British Isles (Thompson et al. 1995). Upland moors also provide habitat for bird species of conservation interest, e.g. curlew Numenius arquata (L.), golden plover Pluvialis apricaria (L.), black grouse Tetrao tetrix (L.), merlin Falco columbarius (L.), hen harrier Circus cyaneus (L.), short-eared owl Asio flammeus (Pontoppidan) and ring ouzel Turdus torquatus (L.) (Thompson et al. 1997; Robson 1998).

These moorlands are mostly cultural landscapes, dominated by dwarf shrubs [mainly Calluna vulgaris (L.) Hull], which were created and maintained by anthropogenic activity, which prevents succession to woodland and maintains a very nutrient-poor ecosystem (Gimingham 1972). Nomenclature follows Stace (1997) for higher plants, Atherton, Bosanquet & Lawley (2010) for bryophytes, and Dobson (2000) for lichens. Most moors are managed for sheep grazing and sporting interests (red grouse Lagopus lagopus scoticus Latham); hence, their wise management is important in terms of both the local economy and biodiversity. From an economic viewpoint, grouse shooting is worth 12% of the total UK shooting provision and 70 000 jobs (PACEC 2006). It has been estimated that between 2004 and 2006, the income from all game shooting in the UK was £750 million and had a Gross Value Added (GVA) of £1·6 billion (PACEC 2006). Assuming direct extrapolation, this implies that grouse shooting provided £90 million of direct income and £0·2 billion for GVA in these years. This is a minimum estimate of the economic value provided by moorlands as it does not include deer stalking, which provides considerable revenues in parts of the UK.

Burning was a major factor in the creation and maintenance of moorlands in north-west Europe and until recently was commonly used throughout the Continent (Ascoli et al. 2009). However, in many parts of Europe, burning has been replaced by other management techniques, e.g. cutting, grazing and sod cutting (Gimingham & de Smidt 1983). The reduction in burning on Continental heaths is partly because their dynamics are driven by either climatic extremes or insect attack, and usually regeneration is via the seed pathway (Marrs 1988). In oceanic areas, layering and vegetative regeneration from burnt/cut stems are more important regeneration pathways (Marrs 1988).

In the UK, burning has been part of moorland management for hundreds, perhaps thousands of years (Simmons 2003), and its use has increased in frequency over the last 200 years as a result of management for sheep and red grouse (Gimingham 1972). The usual management practice in the UK uplands is to burn on a rotation to produce a mosaic of stands in different stages of the burn-recovery cycle (Sotherton, Tapper & Smith 2009). The mosaic contains areas of short succulent Calluna shoots for the red grouse to feed on, and areas of older, taller vegetation for cover (Miller 1980). The aim of the prescribed burning is to remove woody plant growth, leaving charred stems. The burnt stems resprout to produce new growth, although regeneration reduces with plant age and falls off sharply after c. 15 years of age (Miller & Miles 1970). Where resprouting is slow, regeneration must come from seed, which is either buried in the soil or colonizes from outside the burn patch.

Recently, the use of burning has been reviewed (Tucker 2003), and this review stimulated a substantive debate on whether prescribed burning should be continued. The reason for this debate is that moorlands must now contribute to an increasing range of conservation objectives, ranging from the conservation of species and communities through to the provision of ecosystem services (carbon sequestration, water provision, recreation) (Marrs et al. 2007). Given the critical, and in some cases controversial, discussions surrounding these issues, it is surprising, given our increasing requirement to have evidence-based conservation (Pullin, Knight & Watkinson 2009), that so little is known about moorland post-fire succession. Some evidence is available from single-site studies in Scotland for higher plants (Gimingham, Hobbs & Mallik 1981; Hobbs & Gimingham 1984) and for lichens (Davies & Legg 2008). However, there is a lack of evidence on plant communities in England, where the debate over prescribed burning is most contentious. An experiment where different rotations were tested provided equivocal results (Rawes & Hobbs 1979); the overall conclusions suggested cessation of burning was ‘likely to be an acceptable management in the interests of conservation’, but this conclusion could not be fully justified from their results. They showed that burning was important in maintaining the graminoid:Calluna balance, and many species, including Sphagnum species, were at their maximum abundance in a 10-year rotation burn treatment compared to a 20-year one. This lack of reliable predictive information to inform conservation policy has been confirmed in a systematic review, which assessed the impact of burning as a conservation intervention on heaths and bogs; the outcome was that evidence was insufficient to generate robust management recommendations (Stewart, Coles & Pullin 2005).

Here, we assess the impact of prescribed burning in a multi-site study in the Peak District where we use a chronosequence approach to assess the changes in plant species community compositions on five moorlands. Multivariate analysis was used to assess community composition, and this was related to both the moorland site and a range of environmental factors that included both elapsed time since burning and vegetation height. The aim of this study was to determine the relative importance of environmental factors in explaining species community composition, and specifically the relative importance of the temporal aspects of recovery relative to other environmental variables. Thereafter, the realized niche of each species (Lawesson & Oksanen 2002; Smart et al. 2006) was calculated with respect to both elapsed time since burning and vegetation height using Huisman, Olff and Fresco (HOF) modelling to calculate response curves (Huisman, Olff & Fresco 1993). We hypothesized that the disturbance caused by prescribed burning would increase species richness. Here, we test this hypothesis by analysing the response of plant communities to a range of environmental factors, and through inspection of the individual species response curves describing their realized niche space.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Study Areas and Sampling Protocol

A space-for-time substitution study was carried out on five moorlands (Bamford, Broomhead, Howden, Midhope, and Snailsden moors) within the Peak District National Park, UK, over a 3-year period (Table 1). All of the burned patches on each moor were mapped using aerial photography taken in September 2005 and cross-referenced with land burning management maps provided by the land managers. This provided a series of burned patches of known age (elapsed time since burning) between 2 and 18 years; older patches were also identified where no burning had been carried out for at least 35 years. There were 79, 249, 103, 271 and 252 burn patches available on the five moors, respectively. An age-stratified random sampling procedure within each moor was used to select patches for sampling. Once selected, the georeferenced outlines of each of the sampled patches were digitized from aerial photographs, and their areas calculated within ArcGIS (2009). The number of 1 m2 quadrats available within each burn patch was counted, and a random selection was made for field sampling. This procedure was carried out in each of 3 years (Table 1). A pilot study was carried out in 2006 on two moors with 10 quadrats per burn patch; thereafter, in 2007, the survey was extended to the five moors, each with four quadrats per patch, and this was repeated in 2008 with 10 quadrats per patch. The older patches with no burning were sampled only in 2007 (n = 3/moor) and 2008 (n = 1/moor). There were 1010 quadrats sampled overall (2006 – 2 moors × 10 patches × 10 quadrats = 200; 2007 – 5 moors × 13 patches × 4 quadrats = 260; 2008 – 5 moors × 11 patches × 10 quadrats = 550, total n = 1010). The time since burning of the older stands is less precisely known, so they were corroborated by counting the growth rings of the Calluna stems (±3 years, Harris 2011).

Table 1.   Details of the five study moorlands in the Peak District National Park, England, where the post-fire vegetation succession was surveyed
Moorland siteLongitude & latitudeBritish National Grid squares digitized per moorElevation range (m)Age range (years)Estimated age of older patches (years)Sampling years
BamfordLatitude 53°21′N,SK 1993, 1994, 2093, 2094300–4203–14382006, 2007, 2008
Longitude 1°40′W
BroomheadLatitude 53°27′N,SK 2395, 2394, 2295 2294300–4602–15402007, 2008
Longitude 1°38′W
HowdenLatitude 53°28′N,SK 2184, 2185, 2284, 2285272–5402–1735–502006, 2007, 2008
Longitude 1°42′W
MidhopeLatitude 53°29′N,SK2198, 2197, 2099, 2098, 2097, 1999, 1998, 1997270–4803–15402007, 2008
Longitude 1°40′W
SnailsdenLatitude 53°30′N,SE1503, 1501, 1404, 1401, 1400, 1304350–4703–1840–502007, 2008
Longitude 1°44′W

Vegetation Survey

On each of the five moors, the patches and then the quadrat positions were located using GPS (eTrex Venture® HC, Garmin (Europe) Ltd, Romsey, UK). The cover (%) of all higher plants, bryophytes and lichens was recorded in a 1-m2 quadrat, as well as an assessment of the Calluna response to burning in two cover categories, either ‘Bush’ or ‘Stick’. These two categories represent management descriptions of burning response. The ‘Bush’ category reflects plants that recover normally from burning, and the ‘Stick’ category reflects older Calluna that has only been scorched, yet remains alive within a younger developing sward. Pilot studies showed that the 1-m2 quadrat approximated to the minimum sampling unit required to capture >95% of species present; increasing the sampling unit to 9 m2 did not increase species number. A range of environmental measurements were also taken (Table 2). All above-ground material was harvested from the central 0·25 m2 sorted into five fractions (dwarf shrubs, graminoids, bryophytes, litter and animal excrement), dried at 80 °C and weighed. On four moors, all samples had a peat depth of at least 50 cm; the exception was Midhope where some quadrats were on shallower peat (20–30 cm = 4·3%; 30–40 cm = 14·2%, 40–50 cm = 3·7%; >50 cm = 77·8%).

Table 2.   Environmental variables measured or derived for the study at each of the five moors in the Peak District National Park, England. The allocation of each variable to the environmental set for variation partitions, and the transformations used are presented
Variable nameDescriptionSourceTransformationCommentEnvironmental set used in variation partitioning
MoorMoor nameOrdinance Survey Maps Site
EastingNational Grid (km)Ordinance Survey MapsStandardized (mean = 0, s2 = 1);To reduce mean andvariance to same range as other variables 
NorthingNational Grid (km)Ordinance Survey MapsStandardized (mean = 0, s2 = 1)  
ElevationHeight above mean sea level (m)Ordinance Survey Mapsloge(x + 1)  
Burnt Calluna bushCover (%)Estimated in 1-m2 quadratsArcsin(sqrt(x/100)) Biotic
Burnt Calluna stickCover (%)Estimated in 1-m2 quadratsArcsin(sqrt(x/100))  
OthelittCover (%) of other litterEstimated in 1-m2 quadratsArcsin(sqrt(x/100))  
BaregrouCover (%) of bare groundEstimated in 1-m2 quadratsArcsin(sqrt(x/100))  
AnimexcrCover (%) of animal excrementEstimated in 1-m2 quadratsArcsin(sqrt(x/100))Surrogate measure of animal grazing intensity 
pHpH = −log10 [H+]Measured in laboratory (Allen 1989) Physical
LOIAmount of soil organic matter presentMeasured in laboratory (Ball 1964)Arcsin(sqrt(x/100))  
RockCover (%) of rockEstimated in 1-m2 quadratsArcsin(sqrt(x/100))  
AspectMeasured compass bearing (°)Magnetic compass (corrected for magnetic anomaly)   
FaspectFunctional transformation (F) of Aspect (a, degrees, the estimated site-wise mean), F = −sin (a/2)Derived from aspect Measure of southerly aspect 
SlopeMean gradient of survey site(degrees)ClinometerArcsin(sqrt(x/100))  
Open waterCover (%) of open waterEstimated in 1-m2 quadratsArcsin(sqrt(x/100))  
BiomassSampled at each quadrat (g m−2)Measured in laboratory (dry weight basis)loge(x + 1) Production
Elapsed time since last burn (Et)Elapsed time (year) since burningEstate management records   
Vegetation heightMeasurement (cm) of vegetation height within the quadrat, using pressure disc. Pressure of disc on vegetation P  =  Force/Area – (m × g)/(л × r2) = 27·76 PaMeasured with a pressure disc–diameter = 0·3 m, mass 0·2 kgsqrt(x + 0·5)Stewart, Bourn & Thomas (2001) 

Data Analysis

Multivariate analysis and HOF modelling were performed using the ‘vegan’ and ‘gravy’ packages, respectively (Oksanen 2004, 2010), within the R statistical environment (R Development Core Team 2010).

Multivariate analysis

Initially, the relationship between the species community data and the measured environmental factors was explored using Detrended Correspondence Analysis (DCA, ‘vegan’ function ‘decorana’); here, the species cover data were transformed [loge(y + 1)]. The environmental variables were fitted passively to this ordination using the function ‘envfit’ within ‘vegan’ and 10 000 permutations. The distribution of each moor was plotted on the DCA ordination as a bivariate ellipse (standard deviation +95% confidence limits) and the ellipse area calculated using ‘vegan’ function ‘ordiellipse’. Thereafter, variation partitioning was performed (function ‘varpart’, ‘vegan’, Oksanen 2010) to assess the relative contributions of the different environmental variables (allocated into four sets: Site properties, Physical factors, Biotic variables and Production-related variables, Table 2) in explaining the species composition (Peres-Neto et al. 2006). Significance of each of the testable fractions of the VP analysis was determined using redundancy analysis with a randomization test and 10 000 permutations restricted within moor and year.

Modelling post-burning species responses

There were several variables that could be used as the gradient in this analysis. The most obvious one would be elapsed time since burning, although vegetation height or vegetation biomass could also be used as surrogates. Here, elapsed time since burning and vegetation height were selected. The reasons for this were first, whilst every attempt was made to ensure that estimates of elapsed time since burning were accurate using historical information, there is some uncertainty about the oldest burned patches. Any inaccuracy here could skew model fitting. Secondly, the derived variable vegetation height × C. vulgaris cover and vegetation height alone were the most significant variables, and both were more or less coincident with elapsed time since burning in the ordination diagram. As vegetation height is easy to measure under field conditions, we concentrated on vegetation height in subsequent analyses.

The HOF modelling procedure was then used to fit species cover to vegetation height. This procedure fits five models in an hierarchical sequence: Model Type 1 is the null model, Type II has an increasing or decreasing response up to the maximum potential cover, Type III is similar to Type II but has an asymptote below the maximum possible, Type IV is an unimodal response and Type V is similar to Type IV but accommodates a skewed response (Huisman, Olff & Fresco 1993). Here, the HOF models were fitted with a Poisson error distribution, and the AIC statistic was used for model selection. The models for species cover were also compared using ΔD, i.e. the difference between the deviances of the null model and the one selected. The species niche optima and niche widths were estimated for the unimodal/skewed models (Lawesson & Oksanen 2002).


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The vegetation in this study was depauperate with only 20 species detected. The α-diversity (number of species per quadrat) varied both between moors (Bamford > Howden > Snailsden > Midhope > Broomhead) and with elapsed time since burning (Table 3). There was a marked reduction in α-diversity in the oldest age class.

Table 3.   Species number detected in a survey of five moors in the Peak District National Park, England according to elapsed time since burning. Mean values plus the range are shown
MoorAge class (years)
Bamford9·1 (2–16)8·1 (4–14)2·3 (1–5)
Broomhead4·2 (1–7)4·3 (2–8)3·1 (1–5)
Howden8·8 (3–16)8·3 (2–16)2·3 (1–4)
Midhope4·3 (1–8)3·9 (1–6)3·6 (2–5)
Snailsden5·1 (2–9)4·9 (2–8)2·8 (1–5)

Relating Species Community Composition to Environmental Variables

The DCA produced eigenvalues of 0·307, 0·136, 0·127 and 0·1263, and axis lengths of 3·40, 2·58, 2·52 and 3·26 for the first four axes. The species plot (Fig. 1a) shows the main dominant species C. vulgaris near the centre. The significant environmental variables (Table 4a, Fig. 1b) show a very strong diagonal gradient with all variables along it. The variables associated with the early stages of the post-fire period are plotted in the upper right quadrant (Stick and Bush Calluna, litter cover, slop, animal excrement, a surrogate measure of grazing intensity), whereas variables associated with later stages (elapsed time since burning, vegetation height, biomass and Calluna cover × height) are plotted in the lower left quadrant. This gradient suggests that much of the variation in species composition is associated with Calluna recovery and growth after burning (gradient G2, Fig. 1a). The distribution of other species within the biplot reflects inherent variability in moorland species composition (gradient G1, Fig. 1a) suggesting a transition from Eriophorum vaginatum and bryophyte-dominated communities in the upper left quadrant through to shrub-dominated communities with Erica tetralix, Vaccinium myrtillus with Rubus chamaemorus in the bottom right.


Figure 1.  Plots derived from the DCA analysis of plant species composition data and significant environmental variables. (a) Species plot, all species are illustrated (blue are most abundant), G1 and G2 highlight the major gradients; (b) Relationship with significant environmental variables (Table 4); (c) Quadrat plot identifying each moor; using bivariate SD-ellipses (95% confidence limits) superimposed. Codes are: (a) Species: most abundant species in blue; Ci = Campylopus introflexus; Cp = Campylopus pyriformis; Cv = Calluna vulgaris; Ea = Eriophorum angustifolium; Ev = E.vaginatum; En = Empetrum nigrum; Hj = Hypnum jutlandicum; Gs = Galium saxatile; Rc = Rubus chamaemorus; Vm = Vaccinium myrtillus; Less frequent species in red; Ac = Agrostis capillaris; Cci, Ccl, Cs, Csq = Cladonia coccifera, C. chloropea, C. squamosa, C. squammules, respectively; Cpl = Carex pilulifera; Df = Deschampsia flexuosa; Ds = Dicranum scoparium; Et = Erica tetralix; Je = Juncus effusus; Ns = Nardus stricta; Ps = Polytrichum commune; Rs = Rhytidiadelphus squarrosus; (b) Environmental variables see Table 2, and Biomass (g m−2); Bcalstck and Bcalbush refer to stick and bush C. vulgaris after burning (c) Moors: Bamford = blue; Broomhead = green; Howden = red; Midhope = purple; Snailsden = black.

Download figure to PowerPoint

Table 4.   The relationships between plant species composition and significant environmental variables from five moors in the Peak District National Park, England. (a) Correlations between the first two axes of the DCA ordination (Fig. 2) and measured or derived environmental variables (Table 2). The correlations were calculated using the ‘envfit’ function in ‘vegan’ with 10 000 permutations. (b) The relative contributions of these four sets of environmental variables in explaining species composition derived from variation partitioning. Sets are coded – S = Site, B = Biotic, Ph = Physical, Pr = Production. Pseudo-F was derived from a randomization test with 9999 permutations stratified within moor. The sets with vertical lines show the constrained variable left of the line and the covariables on the right; for these sets, the shared variation has been removed
VariableAxis 1Axis 2r2P
Calluna cover × height−0·773−0·6350·280·0001
Vegetation height−0·835−0·5500·270·0001
Calluna Stick cover0·7840·6210·220·0001
Litter cover0·8780·4790·130·0001
Elapsed time−0·759−0·6510·090·0001
Calluna Bush cover0·8010·5980·070·0001
Animal excrement0·7520·6580·020·0004
Rock cover−0·1000·0250·020·0018
Variable setd.f.Adjusted r2 (%)Pseudo-F (all < 0·001)
  1. DCA, Detrended Correspondence Analysis.


The five moors show much overlap within the ordination space, although they all centre on C. vulgaris (Fig. 1c). Three of the moors (Broomhead, Midhope and Snailsden) had relatively small ellipses (0·56–0·97 units) and occupy more or less the same ordination space, but Howden and Bamford occupy much larger areas, 2·29 units and 4·30 units, respectively.

The variation partitioning analysis explained 15·2% of the total variation in the data set (Table 4b). The four sets explained the following amounts of total variation in decreasing order: Site (9·3%), Production (8·3%), Biotic (8·0%), Physical (1·1%) and with 0·96% shared between them. When shared variation was removed, the amount of variation explained was reduced to Site (3·6%), Production (2·2%), Biotic (1·0%) and Physical (0·5%). All were significant (< 0·0001).

Relating Species Response to Elapsed Time Since Burning and Vegetation Height

Calluna vulgaris showed an increase in cover with both elapsed time since burning and vegetation height (Fig. 2). This species reached >90% cover on all moors after about 30 years. The overall model for vegetation height was a much better fit (reduction in deviance from the null model, ΔDev = 57·3%) than for elapsed time since burning (ΔDev = 25·7%).


Figure 2.  The response of Calluna vulgaris to (a) vegetation height (cm) and (b) elapsed time since burning (years) derived by Huisman, Olff and Fresco (HOF) modelling along a post-prescribed-fire succession on five moorlands in the Peak District, National Park, England. C. vulgaris exhibited a Type II HOF model.

Download figure to PowerPoint

All other species were reduced to a low cover as vegetation grew taller and with elapsed time since burning. Two types of response curve were found over the response ranges assessed: (i) a general decline (HOF model II) or a peak at the low end of the range with subsequent decline (HOF models IV and V). Those species where the fitted response reduced the deviance by more than 10% relative to the null model (HOF model I) are displayed (Figs 3 and 4), but the upper tolerance value for a wider range of species that exhibited a Type IV and V response are also shown (Table 5). Visual inspection of Figs 3 and 4 shows that most species had declined to zero cover between 20 and 40 cm in height and 20 years after burning, and the upper tolerance values for almost all species were below 40 cm in height and 22 years since burning. Two exceptions were Empetrum nigrum and E. vaginatum; both species reduced the deviance compared to the null model, but the reduction was <10%. E. nigrum had an upper tolerance value of 47 cm for vegetation height and 22 years since burning. E. vaginatum had an upper tolerance value of 37 cm for height, but the upper tolerance value for elapsed time since burning was not calculable as HOF model II was selected.


Figure 3.  The response of nine species to vegetation height (cm) derived by Huisman, Olff and Fresco modelling (model type is shown) along a post-prescribed-fire succession on five moorlands in the Peak District, National Park, England. Only those species that reduced deviance by at least 10% relative to the null model (Type = I) are illustrated. Species codes: Gs = Galium saxatile; Rc = Rubus chamaemorus; Cp = Campylopus pyriformis; Ac = Agrostis capillaris; Ds = Dicranum scoparium; Cpl = Carex pilulifera; Je = Juncus effusus; Cs = C. squamosa; Csq = C. squammules.

Download figure to PowerPoint


Figure 4.  The response of nine species to elapsed time since burning (years) derived by Huisman, Olff and Fresco modelling (model type is shown) along a post-prescribed-fire succession on five moorlands in the Peak District, National Park, England. Only those species that reduced deviance by at least 10% relative to the null model (Type = I) are illustrated. Species codes: Ea = Eriophorum angustifolium; Gs = Galium saxatile; Rc = Rubus chamaemorus; Ac = Agrostis capillaris; Cp = Campylopus pyriformis; Cpl = Carex pilulifera; Ds = Dicranum scoparium; Je = Juncus effusus; Cs = C. squamosa; Ns = Nardus stricta; Ci = Campylopus introflexus.

Download figure to PowerPoint

Table 5.   Upper tolerance limits of species with a fitted HOF model Type IV and V at each of the five moors in the Peak District, National Park, England
SpeciesVegetation height (cm)Elapsed time (year)
  1. nc, upper tolerance not calculated; HOF, Huisman, Olff and Fresco.

Agrostis capillaris14·615·4
Carex piluliferanc21·6
Cladonia chlorophea38·521·7
Campylopus introflexus20·220·3
Cladonia squamosa11·615·6
Deschampsia flexuosa26·411·0
Dicranum scoparium20·015·1
Erica tetralixnc23·3
Eriophorum angustifolium28·4nc
Eriophorum vaginatumnc36·7
Empetrum nigrum46·921·8
Hypnum jutlandicum37·220·3
Nardus stricta13·2nc
Polytrichum commune40·8nc


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Fire has been a major tool used to create and maintain heathland systems throughout Europe, and the few areas of heathland remaining have a high conservation value (Gimingham 1972). However, in recent years, the use of prescribed burning of heathlands/moorlands has declined (Ascoli et al. 2009). The long-term association of heathland species with fire was recently demonstrated in Norwegian heathlands through the detection of an ecophysiological response of C. vulgaris to smoke and smoke-derived solutions (Måren et al. 2010). Whilst there is no doubt that heathlands and moorlands can be managed using combinations of other techniques to prevent tree invasion and encourage C. vulgaris regeneration (Gimingham & de Smidt 1983), prescribed burning is a cost-effective tool for managing those ecosystems that contain such fire-adapted species. Whilst our study is on prescribed fire, there is considerable concern that the frequency of wildfire will increase in all heath/moor ecosystems in the future as a result of the drier summers produced under global warming (Worrall et al. 2010). If this occurs, then the use of prescribed burning may increase as a means of reducing fuel loads and increasing the resilience of these moorland systems cheaply. Our results, therefore, provide information to guide policies for the future sustainable management of European heaths and moors.

There is, however, a potential conflict between competing conservation objectives (Marrs et al. 2007) Apart from the conservation of species and communities found on moorlands, many moorlands in upland Britain occur on peaty soils that should sequester carbon and many are also drinking water catchments, where a reduced water quality (increased coloration) will increase purification costs For these latter ecosystem services, a reduction in burning frequency or a no-burn policy would be advised. Almost nothing is known about the impact of prescribed burning on carbon accounting (discussed in Worrall et al. 2009), but there is some evidence that burning may be associated with increased amounts of dissolved organic carbon and certain nutrients in moorland drainage (Yallop & Clutterbuck 2009). Therefore, there is a pressing need to assess the impact of prescribed burning on all aspects of ecosystem structure and function.

Here, we combined the use of a nested survey design and a chronosequence model to assess species and community response after prescribed burning at a regional scale within the British uplands. Such chronosequence studies can be criticized because temporal change was inferred from stitching together a sequence of sites with differing histories. However, our nested design was replicated on five moors; with burned patches with different elapsed times since burning chosen randomly, and then within-patch random sampling. The overall consistency of the responses using this approach suggests that the conclusions derived using this chronosequence approach are valid, although to be certain they need to be corroborated by long-term observational studies. We suggest that our sampling approach might have more general applicability for temporal studies of vegetation response.

Response of Peak District Moorlands to Prescribed Burning

The plant species diversity in these moors was very low, and this has been ascribed to past pollution (Tallis 1998), overgrazing (Anderson & Yalden 1981) and intermittent, damaging wildfires (Albertson et al. 2009). These moors also have a very limited species pool compared to potential target NVC communities (Rodwell 1991, 1992). Harris (2011) identified 28 NVC community classes/subclasses on these moors; the main upland classes were: H12, C. vulgaris–V. myrtillus heath; M19, C. vulgaris–E. vaginatum blanket bog; M20, E. vaginatum blanket/raised bog and U2 Deschampsia flexuosa grassland. Overall, 49 species were missing (mainly bryophytes) from these moorlands compared to their nearest NVC class after correcting for species that did not occur locally.

The community composition showed considerable overlap on the five moors around the centre of the DCA ordination; nevertheless, there were major differences in the overall species pool. This difference was evidenced by inspection of the ellipse areas; Bamford, with the lowest elevational range, had by far the greatest ellipse area, and the ellipse showed that this moor encompassed the graminoid-dominated communities; Howden was intermediate and at its higher elevational range had a more lichen-rich vegetation, whereas the remaining three moorlands (Broomhead, Midhope, Snailsen) had the smallest species pools, and were all of similar size. The size of these ellipses reflects the species richness at each site almost certainly caused by different past and current management practices.

The ordination analyses produced two gradients, the first relating to intrinsic variation in the plant communities ranging from Eriophorum-dominated communities to shrub-dominated ones, presumably reflecting local effects, including site history, management and climate, Most of the variation on this gradient was produced by the two moors with the larger species pools. A secondary gradient, which was centred on all sites, reflected post-fire succession and increasing cover of the major moorland species, C. vulgaris. The variation partitioning indicated that productivity variables, essentially associated with C. vulgaris growth, were almost as important as site variables in explaining community composition. The amount of variation explained was low, and this may be a result of the low species number, but they are of the same order of magnitude as those in other studies with larger data sets and with much greater species diversity (Corney et al. 2006; Marrs et al. 2011).

A further major finding was that in the analysis of post-prescribed-burning species response or succession, C. vulgaris was the only species to increase in cover throughout the time period investigated, producing more than 90% cover after the vegetation height reached 60 cm or after about 30 years. At this point C. vulgaris was the most dominant species, and this is reflected in the relatively large proportion of variation accounted for by the production variables, second only to site-level variables. All other species showed either a reduction or an unimodal response through the post-prescribed-burning succession. For the most part, these species had reduced to minimal levels by 22 years and 40 cm height with two exceptions: E. nigrum at 47 cm height and 22 years after burning, and E. vaginatum at 37 years after burning. The estimate limit for elapsed time since burning is in the range where there were no sample plots. Our age estimate has, therefore, some uncertainty associated with the shape of the curve in this region. However, the range of vegetation height, on the other hand, covers the entire height range up to 90 cm and should provide more reliable estimates. For this reason, and because vegetation height is easy to measure by land managers using the pressure disc approach used here, we recommend vegetation height as the preferred measure for assessing the optimal time to burn.

These results confirm that the disturbance caused by prescribed burning creates gaps, with enhanced light levels and perhaps a flush of nutrients, which allow plant species to colonize from seed and/or resprout from stems, or rhizomes (Mallik & Gimingham 1985) increasing diversity. Thereafter, C. vulgaris slowly becomes dominant, and there is a decline in the cover of almost all other species. Species richness declined with elapsed time since burning and vegetation height on these five moors, and similar results have been found in single-site studies for plant species diversity (Hobbs & Gimingham 1984) and terricolous lichens (Davies & Legg 2008).

These results confirm that a degree of disturbance brought about by prescribed burning maintains greater plant species diversity than when left undisturbed. The evidence suggests that if burning is not carried out on a regular basis, then the cover of all species detected here, except C. vulgaris, will decline. The species response curves presented here provide no evidence to support the hypothesis that late-successional (e.g. Betula spp.) or bog-forming species (Sphagnum spp.) will invade the moorlands if prescribed burning was not carried out. The results are consistent with the initial floristic model (Egler 1954) and the tolerance models of succession working together through time (Connell & Slatyer 1977). The initial floristic model is supported by the fact that all species were present at the very start of the succession, and there was an increase in cover of almost all species after gaps creation caused by the disturbance. The tolerance model explains the observed changes in vegetation cover by all species (including C. vulgaris) expanding immediately after disturbance, and thereafter all species other than C. vulgaris declining either as a result of their life-history strategies, or though competition from C. vulgaris as its biomass increases.

Implications for Practical Management

The moorland vegetation studied here is very species poor; however, the plant diversity that remains appears to require prescribed burning for its maintenance. Immediately after burning, there was an immediate increase in species and a reduction through time as vegetation height increased. Indeed, the oldest stands on these moors were composed of C. vulgaris, its litter, and very little else (Harris 2011). There was no evidence found in this study that C. vulgaris was exhibiting the 4-phasic-growth cycle reported elsewhere (Watt 1947; Gimingham 1972); here, the C. vulgaris appeared to grow upwards continually rooting in its own litter as has been reported for other upland moors in the north Pennines (Forrest 1971). We suggest therefore that (i) prescribed burning is required to maintain even the low, existing plant diversity, found on these moors, and (ii) a no-burning policy would result in monodominant stands of C. vulgaris of little conservation value. We do not know how the ecosystem would respond beyond the time period studied here. The seed bank of the moors is depauperate (Harris 2011), and it is unlikely that seed could establish in the very tall vegetation that develops.

Thus, it appears that prescribed burning maintains the dominance of C. vulgaris, a feature that is not always considered a high conservation priority (McVean & Ratcliffe 1962; Littlewood et al. 2010). Indeed, C. vulgaris might be considered a ‘Thug’ species (sensuMarrs et al. 2011), a native species that takes over and excludes other species. The issue in these moorlands is how to maintain biodiversity. On the moorland vegetation of the Peak District, a no-burn policy on its own will make this situation worse with (i) an enhanced cover and biomass of C. vulgaris and its litter and (ii) almost no other species. To enhance the species diversity, almost certainly some form of management to reduce C. vulgaris is needed complemented by the reintroduction of other species. Prescribed burning is the most economic treatment for maintaining the existing plant biodiversity.

We have shown that under a prescribed burning regime, most plant species persisted up to c. 20 years and a vegetation height of c. 40 cm. Therefore, to maximize diversity, prescribed burning will be most effective before any species disappear from the vegetation. As vegetation height is the easiest measure, we recommend that prescribed burning should be implemented before the vegetation reaches c. 25 cm with 40 cm as a maximum height threshold.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We thank Mark Osborne, the Heather Trust, Moors for the Future, the Moorland Association, the Royal Horticultural and Botanical Society of Manchester and the Northern Counties and the BiodivERsA FIREMAN program (NERC/Defra) for funding. Sandra Mather provided the illustrations. An anonymous reviewer and Dr Colin Legg made many helpful suggestions to improve the manuscript.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Albertson, K., Aylen, J., Cavan, G. & McMorrow, J. (2009) Forecasting the outbreak of moorland wildfires in the English Peak District. Journal of Environmental Management, 90, 26422651.
  • Allen, S.E., (ed.) (1989) Chemical Analysis of Ecological Materials. Blackwells, Oxford, UK.
  • Anderson, P. & Yalden, D.W. (1981) Increased sheep numbers and the loss of heather moorland in the Peak District, England. Biological Conservation, 20, 195213.
  • ArcGIS (2009) ArcGIS: The complete Geographic Information System. ESRI, Redlands, CA, USA. URL [accessed 26/1/2010].
  • Ascoli, D., Beghin, R., Ceccato, R., Gorlier, A., Lombardi, G., Lonati, M., Marzano, R., Bovio, G. & Cavallero, A. (2009) Developing an adaptive management approach to prescribed burning: a long-term heathland conservation experiment in north-west Italy. International Journal of Wildland Fire, 18, 727735.
  • Atherton, I., Bosanquet, S. & Lawley, M. (2010) Mosses and Liverworts of Brian and Ireland: A Field Guide. British Bryological Society, Plymouth, UK.
  • Ball, D.F. (1964) Loss-on-ignition as estimate of organic matter + organic carbon in non-calcareous soils. Journal of Soil Science, 15, 8492.
  • Connell, J.H. & Slatyer, R.D. (1977) Mechanisms of succession in natural communities and their role in community stability and organization. American Naturalist, 111, 11191144.
  • Corney, P.M., Le Duc, M.G., Smart, S.M., Kirby, K.J., Bunce, R.G.H. & Marrs, R.H. (2006) Relationships between the species composition of forest field-layer vegetation and environmental drivers assessed using a national scale survey. Journal of Ecology, 94, 383401.
  • Davies, G.M. & Legg, C.J. (2008) The effect of traditional management burning on lichen diversity. Applied Vegetation Science, 11, 529538.
  • Dobson, F. (2000) Lichens: An Illustrated Guide to the British and Irish Species. Richmond Publishing Company, Richmond, UK.
  • Egler, F.E. (1954) Vegetation science concepts, I. Initial floristic composition, a factor in old-field development. Vegetatio, 14, 412417.
  • Forrest, G.I. (1971) Structure and production of North Pennine blanket bog vegetation. Journal of Ecology, 59, 453479.
  • Gimingham, C.H. (1972) Ecology of Heathlands. Chapman & Hall, London.
  • Gimingham, C.H. & de Smidt, J.T. (1983) Heaths as natural and semi-natural vegetation. Man’s Impact on Vegetation (eds M.J.A. Werger & I. Ikusima), pp. 185199. Junk, The Hague.
  • Gimingham, C.H., Hobbs, R.J. & Mallik, A.U. (1981) Community dynamics in relation to management of heathland vegetation in Scotland. Vegetatio, 46–47, 149155.
  • Harris, M.P.K. (2011) The role of prescribed burning in moorland management in the Peak District. PhD thesis. University of Liverpool, Liverpool.
  • Hobbs, R.J. & Gimingham, C.H. (1984) Studies on fire in Scottish heathland communities. 2. Post-fire vegetation development. Journal of Ecology, 72, 585610.
  • Huisman, J., Olff, H. & Fresco, L.F.M. (1993) A hierarchical set of model for species response analysis. Journal of Vegetation Science, 4, 3746.
  • Lawesson, J.E. & Oksanen, J. (2002) Niche characteristics of Danish woody species as derived from coenoclines. Journal of Vegetation Science, 13, 279290.
  • Littlewood, N., Anderson, P., Artz, R., Bragg, O., Lunt, P. & Marrs, R. (2010) Peatland Biodiversity – A Technical Review for the IUCN Peatland Program. IUCN Peatland Commission, York, UK.
  • Mallik, A.U. & Gimingham, C.H. (1985) Ecological effects of heather burning 2. Effects on seed-germination and vegetative regeneration. Journal of Ecology, 73, 633644.
  • Måren, I.E., Janovský, Z., Spindelböck, J.P., Daws, M.I., Kaland, P.E. & Vandvik, V. (2010) Prescribed burning of northern heathlands: Calluna vulgaris germination cues and seed-bank dynamics. Plant Ecology, 207, 245256.
  • Marrs, R.H. (1988) Vegetation change on lowland heaths and its relevance for conservation management. Journal of Environmental Management, 26, 127149.
  • Marrs, R.H., Galtress, K., Tong, C., Blackbird, S.J., Heyes, T.J., Pakeman, R.J. & Le Duc, M.G. (2007) Conflicts between competing conservation goals, biodiversity or ecosystem services: quantification of element losses and species recruitment using a moorland-bracken model system. Journal of Environmental Management, 85, 10341047, doi:10.1016/jenvman.11.011.
  • Marrs, R.H., Le Duc, M.G., Smart, S.M., Kirby, K.J., Bunce, R.G.H. & Corney, P.M. (2011) Aliens or natives: who are the ‘thugs’ in British woods? Kew Bulletin, 65, 111. Still Online First.
  • McVean, D.N. & Ratcliffe, D.A. (1962) Plant Communities of the Scottish Highlands. HMSO, London, UK.
  • Miller, G.R. (1980) The burning of heather moorland for red grouse. Bulletin d’Ecologie, 11, 725733.
  • Miller, G.R. & Miles, J. (1970) Regeneration of heath (Calluna vulgaris) (L.) Hull) at different ages and seasons in north-east Scotland. Journal of Applied Ecology, 7, 5160.
  • Oksanen, J. (2004) Gradient Analysis of Vegetation (gravy) software, v.0.0-21. URL [accessed 26/5/2010].
  • Oksanen, J. (2010) Multivariate Analysis of Ecological Communities in R: vegan tutorial, v.1.11.0 for Windows. URL [accessed 26/5/2010].
  • PACEC (2006) The Economic and Environmental Impact of Sporting Shooting. A report prepared by Public and Corporate Economic Consultants (PACEC) on behalf of BASC, CA, and CLA and in association with GCT, Cambridge, UK.
  • Peres-Neto, P.R., Legendre, P., Dray, S. & Borcard, D. (2006) Variation partitioning of species data matrices: estimation and comparison of fractions. Ecology, 87, 26142625.
  • Pullin, A.S., Knight, T.M. & Watkinson, A.R. (2009) Linking reductionist science and holistic policy using systematic reviews: unpacking environmental policy questions to construct an evidence-based framework. Journal of Applied Ecology, 46, 970975.
  • R Development Core Team (2010) R: A Language and Environment for Statistical Computing. R Foundation for Statistical computing, Vienna, Austria. URL [accessed 25/5/2010].
  • Ratcliffe, D.A. & Thompson, D.B.A. (1988) The British uplands: their ecological character and international significance. Ecological Change in the Uplands (eds M.B. Usher & D.B.A. Thompson), pp. 936. Blackwells, Oxford.
  • Rawes, M. & Hobbs, R. (1979) Management of semi-natural blanket bog in the northern Pennines. Journal of Ecology, 67, 789807.
  • Robson, G. (1998) The breeding ecology of curlew Numenius arquata on North Pennine moorland. PhD Thesis. University of Sunderland, Sunderland, UK.
  • Rodwell, J.S. (1991) British Plant Communities Vol. 2: Mires and Heaths. CUP, Cambridge, UK.
  • Rodwell, J.S. (1992) British Plant Communities Vol. 3: Grasslands and Montane Communities. CUP, Cambridge, UK.
  • Simmons, I.G. (2003) The Moorlands of England and Wales: An Environmental History 8000 BC–AD 2000. Edinburgh University Press, Edinburgh, UK.
  • Smart, S.M., Thompson, K., Marrs, R.H., Le Duc, M.G., Maskell, L.C. & Firbank, L.G. (2006) Biodiversity loss and biotic homogenization across human-modified ecosystems. Proceedings of the Royal Society London B, 273, 26592665.
  • Sotherton, N., Tapper, S. & Smith, A. (2009) Hen harriers and red grouse: economic aspects of red grouse shooting and the implications for moorland conservation. Journal of Applied Ecology, 46, 955960.
  • Stace, C.A. (1997) New Flora of the British Isles, 2nd edn. CUP, Cambridge, UK.
  • Stewart, K.E.J., Bourn, N.A.D. & Thomas, J.A. (2001) An evaluation of three quick methods commonly used to assess sward height in ecology. Journal Applied Ecology, 38, 11481154.
  • Stewart, G.B., Coles, C.F. & Pullin, A.S. (2005) Applying evidence-based practice in conservation management: lessons from the first systematic review and dissemination projects. Biological Conservation, 126, 270278.
  • Tallis, J.H. (1998) Growth and degradation of British and Irish blanket mires. Environmental Review, 6, 81122.
  • 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, vegetation change and some objectives for nature conservation. Biological Conservation, 71, 163178.
  • Thompson, D.B.A., Gillings, S.D., Galbraith, C.A., Redpath, S.M. & Drewitt, J. (1997) The contribution of game management to biodiversity: a review of the importance of grouse moors for upland birds. Ecological Change in the Uplands (eds L.V. Fleming, A.C. Newton, J.A. Vickery & M.B. Usher), pp. 198212. HMSO, Edinburgh, UK.
  • Tucker, G.T. (2003) Review of the Impacts of Heather and Grassland Burning in the Uplands on Soils, Hydrology and Biodiversity. (English Nature Research Report No. 550). Natural England, Peterborough, UK.
  • Watt, A.S. (1947) Pattern and process in the plant community. Journal of Ecology, 35, 122.
  • Worrall, F., Evans, M.G., Bonn, A., Reed, M.S., Chapman, D. & Holden, J. (2009) Can carbon offsetting pay for upland ecological restoration. Science of the Total Environment, 408, 2636.
  • Worrall, F., Clay, G.D., Marrs, R. & Reed, M.S. (2010) IUCN Technical Review 5 – Impacts of Burning Management on Peatlands. IUCN Peatland Commission, York, UK.
  • Yallop, A.R. & Clutterbuck, B. (2009) Land management as a factor controlling dissolved organic carbon release from upland peat soils 1: spatial variation in DOC productivity. Science of the Total Environment, 407, 38033813.