Controlled burning is used world-wide for vegetation management, but there are serious concerns about its environmental implications (Freckleton 2004). In the UK, fire has been used to control upland vegetation since 7700–6300 BC (Goodfellow 1998), but over the last 150 years many upland landscapes have been subjected to controlled rotational burning regimes (Davies 2008). Rotational burning usually occurs on patches of c. 400 m2, and burning cycles vary from 8 to 25 years (Davies 2008; Grant et al. 2012) depending on productivity, habitat type, grazing level, traditional burning schedules or government body instigated management prescriptions. Thus, the catchment of an individual stream will have dozens of burning patches of different ages. Typically, burning will take place within the catchment most years, but each year, a different set of patches will be burned so that on average an individual patch will be burned once every 8–25 years. Across burned peatland, there will therefore be patches that have been very recently burned (i.e. within the last 12 months) and those that have not been burned for many years, thereby creating a mosaic. Rotational burning on peatlands is practised to remove ageing dwarf shrubs [e.g. Molinia caerulea (L.) Moench and Calluna vulgaris (L.) Hull] and allow regeneration of younger, palatable shoots. This is deemed to be suitable for increasing red grouse populations (Harris et al. 2011; Worrall et al. 2011). Annually, in England and Wales alone, grouse shooting is worth more than £10 million to land owners (Ward et al. 2007) and contributes some £192 million to the UK upland economy indirectly (P.A.C.E.C 2006).
Open upland moors consist of a variety of vegetation and soil types including deep blanket bog, wet heath and dry heath. In England and Wales, there is a Code (Defra 2007) that anyone burning vegetation is expected to follow. This burning code includes a presumption against burning on blanket bog. Undoubtedly, however, a large amount of burning takes place on blanket bog, often with permission of regulatory authorities. Previous work from Yallop, Thacker & Clutterbuck (2006a) has suggested that there was an increase in c. 20% of upland heath and bog that had been burnt recently, implying an increase in rotation frequency. Defra (2010) estimated that 18% of UK peatlands have been subjected to managed burning, which is c. 3150 km2. Although there are large economic benefits with sport shooting (see report by P.A.C.E.C 2006), more research is needed to understand fully the environmental impacts of rotational vegetation burning (Sutherland et al. 2006).
A conservation status assessment made by English Nature (2003) reported that 24% of the area of upland Sites of Special Scientific Interest (SSSI) in England was in an unfavourable condition due to rotational burning. Rotational burning can cause alterations to the terrestrial environment (e.g. vegetation, soil structural, physical and chemical alterations, Maltby, Legg & Proctor 1990; Laubhan 1995), increase sediment erosion and transfer to stream systems (e.g. Imeson 1971; Arnett 1980), increase saturation-excess overland flow through higher water-tables as there is less plant transpiration (e.g. Clay et al. 2009a) and perhaps induce changes to stream chemistry [e.g. dissolved organic carbon (DOC), Mitchell & McDonald 1992; Clay, Worrall & Fraser 2009b; Clay, Worrall & Fraser 2010]. While there are multiple drivers of increased water discoloration (associated with DOC production) in peatland streams over the past 40 years (Worrall et al. 2004; Evans et al. 2006a; Chapman et al. 2010), there is evidence to suggest that prescribed burning is an additional factor, although further work is required to establish causal mechanisms (Holden et al. 2012).
Despite the recent increase in attention on the effects of rotational vegetation burning on aquatic systems, there remains a lack of knowledge about impacts on stream biota (Ramchunder, Brown & Holden 2009; Worrall et al. 2010). Ramchunder, Brown & Holden (2012) documented that increases in fine particulate organic matter (FPOM) and suspended sediment concentrations (SSC) following peatland drainage were associated with decreased abundance of some mayfly and stonefly species, but increases in Ephemera danica (Müller) (Ephemeroptera), Chironomidae and Simuliidae abundances. Comparable responses of the stream ecosystem can be hypothesized for systems affected by vegetation burning because the alterations caused to the terrestrial environment could potentially deliver elevated sediment loads to nearby water courses (Ramchunder, Brown & Holden 2009). Similar effects have been observed in stream ecosystems affected by forest fires (e.g. Minshall, Robinson & Lawrence 1997; Vieira et al. 2004).
Macroinvertebrates constitute an important part of animal production within freshwaters and are integral to the structure and functioning of these ecosystems (Allan & Castillo 2007). The categorization of stream macroinvertebrates into functional feeding groups (FFG) is a reliable tool for assessing the dynamics of lotic communities (Allan & Castillo 2007). Postwildfire studies in US forests have shown shredder biomass decreases due to the loss of riparian vegetation inputs, while algal biomass increases following the opening of the canopy and nutrient release led to more scrapers (Minshall 2003). To date, there have been no studies investigating macroinvertebrate community responses following rotational vegetation burning on UK peatland ecosystems or elsewhere.
This study investigated stream macroinvertebrate communities from ten headwater peatland catchments (five intact and five burned). The aim was to provide a detailed evaluation of how controlled vegetation burning on peatland influences stream macroinvertebrate communities. Based on knowledge from previous studies of peatland drainage and burning, it was hypothesized that (H1) streams in burned catchments would have higher SSC and benthic FPOM compared with intact catchments (Maltby, Legg & Proctor 1990; Tucker 2003). Previous work by Ramchunder, Brown & Holden (2012) suggested that increases in FPOM and SSC in artificially drained catchments altered individual species abundance, but had no discernible effect on community richness, Simpson's diversity, dominance and total abundance. Therefore, (H2) similar biological responses were expected in burned catchments. However, (H3) alterations in the stream environment due to burning were expected to result in macroinvertebrate communities containing higher abundance of taxa associated with in-stream fine sediment deposition and benthic particulate organic matter, with increases in filtering-collectors (linked to FPOM supply from burned catchments), but negative effects on herbivore and predator abundance (e.g. Mihuc & Minshall 1995; Vieira et al. 2004). The findings of this study are considered subsequently in the context of more general literature on rotational vegetation burning effects on peatland stream ecosystems, and some implications for upland policy makers and landowners are discussed.
Materials and methods
This study comprised of: (i) a seasonal study of three burned sites and three unburned sites (hereafter 3v3 survey) located in Upper Teesdale, Wensleydale and Geltsdale in northern England and (ii) a broader, single occasion survey, comparing five burned sites and five unburned sites (hereafter 5v5 survey), with the data sets from (a) augmented by sampling at additional sites in the north Peak District (Table 1).
Table 1. Catchment information for the ten stream study sites
|Moss Burn (Teesdale)||Intact||Blanket peat||2·15||54°41′1″N2°27′0″W|
|Unnamed second-order tributary of River Tees (Teesdale)||Intact||Blanket peat, stagnogley, stagnohumic gley, humic gley, fine loam, alluvial gley||2·23||54°15′7″N21°6′1″W|
|Snaizeholme (Wensleydale)||Intact||Blanket peat, alluvial floodplain||1·12||54°41′8″N2°26′8″W|
|Crowden Little Brook (Peak District)b||Intact||Blanket peat, fine sandy loam||2·11||53°30′8″N2°53′4″W|
|Short Grain (Peak District)b||Intact||Blanket peat, fine sandy loam||1·49||53°34′2″N2°55′9″W|
|Great Eggleshope Beck (Teesdale)||Burned||Blanket peat, stagnogley||4·10||54°40′5″N2°3′8″W|
|Eller Beck (Teesdale)||Burned||Blanket peat, stagnogley, fine loam||1·67||54°29′2″N2°0′9″W|
|New Water (Geltsdale)||Burned||Blanket peat, stagnogley||2·18||54°50′8″N2°37′1″W|
|Ashop Clough (Peak District)b||Burned||Blanket peat, stagnogley, fine sandy loam||1·82||53°24′8″N2°53′0″W|
|Thickwoods Brook (Peak District)b||Burned||Blanket peat, fine sandy loam||1·61||53°29′2″N2°41′5″W|
Potential study catchments were identified as those having second-order streams based on 1 : 25000 Ordnance Survey maps, and candidate burned sites were identified from aerial photographs. Sites were selected randomly with no confounding effects of recent wildfire, mining, major erosion or forest cover. At each catchment outlet, a representative 15-m reach was selected randomly for study with subsequent sampling undertaken in riffle areas of those reaches.
All sites had blanket peat cover, with vegetation dominated by Eriophorum spp. and C. vulgaris (L.) Hull, and there was Sphagnum spp. cover at all sites, but this was less abundant in the Peak District. Although data were not available for all sites, mean annual precipitation of 2012 mm (1951–1980; 1991–2006) occurs at Moor House, Teesdale (Holden & Rose 2011). Mean annual air temperature at Moor House is 5·3 °C (1931–2006; Holden & Rose 2011). Annual rainfall varies considerably across the Peak District, ranging from 1000–1584 mm (Evans, Warburton & Yang 2006b; Shotbolt, Rothwell & Lawlor 2008). The climate is cool with mean monthly temperatures ranging from 2–14 °C (Evans 2005).
For the 3v3 survey, streams were sampled seasonally across 3–4 days per quarter (2007: September 11–13, December 19–21; 2008: March 4–7, June 10–13, September 16–18). The 5v5 survey was concurrent with the September 2008 survey. During each site visit, 16 stream environmental variables were measured to provide contextual habitat information (Table 2). Water temperature, pH and electrical conductivity (EC) were measured using MP120 and MP126 handheld probes (Mettler-Toledo Ltd, Leicester, UK). Dissolved oxygen (DO) concentration was measured using a HI9412 probe (Hanna Instruments Ltd, Bedfordshire, UK). Additionally, 120 mL of stream water was passed through a 0·45-μm filter and subsequently analysed in the laboratory for chloride (Cl), sulphate (SO4) and nitrate (NO3), dissolved organic carbon (DOC), aluminium (Al) and iron (Fe). A further 500 mL of unfiltered stream water was collected for the determination of SSC by filtration. Streambed sediments were characterized by sampling 100 clasts randomly, measuring b-axis lengths and calculating the median grain size (D50). To provide a relative indication of flow differences between sites and over time, stream discharge (Q) was measured at the time of sampling using an open channel flow meter (Valeport, Devon, UK) and the velocity–area method.
Table 2. Descriptive statistics and rm-anova and one-way anova results for the physicochemical variables measured during the 3v3 and 5v5 surveys, respectively. Bold text indicates significance at P < 0.05
|Min||2·33||< 0·01||2·54||< 0·01||0·01||4·56||4·90||36·80||4·86||1·00||2·0||0·02||0·21||0·25||1·5||0·01|
|Season (F4,29)||F = 8·16P = 0·033||F = 14·87P = 0·011||F = 2·18P = 0·234||F = 1·40P = 0·375||F = 1·20P = 0·431||F = 0·85P = 0·562||F = 1·89P = 0·276||F = 0·16P = 0·949||F = 1·17P = 0·442||F = 0·82 P = 0·575||No replicates||F = 2·79P = 0·172||F = 6·25P = 0·052||F = 0·89P = 0·544||F = 1·04P = 0·484||F = 2·99P = 0·157|
|Land management (F1,29)||F = 21·00P = 0·010||F = 14·41P = 0·019||F = 25·41P = 0·007||F = 14·87P = 0·018||F = 968·60P < 0·001||F = 45·87P = 0·002||F = 0·08P = 0·791||F = 4·55P = 0·100||F = 7·18P = 0·055||F = 146·71P < 0·001||F = 19·88P = 0·011||F = 27·28P = 0·006||F = 50·71P = 0·002||F = 25·36P = 0·007||F = 0·21P = 0·671||F = 1·94P = 0·236|
|Season*Land management (F4,29)||F = 0·47P = 0·754||F = 1·68P = 0·194||F = 0·86P = 0·506||F = 1·86P = 0·156||F = 2·09P = 0·120||F = 0·87P = 0·500||F = 2·35P = 0·090||F = 2·29P = 0·096||F = 1·41P = 0·266||F = 0·35P = 0·838||No replicates||F = 0·66P = 0·629||F = 0·26P = 0·901||F = 0·44P = 0·778||F = 2·92P = 0·047||F = 2·86P = 0·050|
|Land management (F1,9)||F = 0·58P = 0·469||F = 0·15P = 0·710||F = 0·05P = 0·838||F = 17·41P = 0·003||F = 4·09P = 0·078||F = 9·91P = 0·014||F = 0·55P = 0·480||F = 0·53P = 0·487||F = 0·08P = 0·784||F = 19·38P = 0·002||F = 22·40P = 0·001||F = 0·85P = 0·386||F = 2·51P = 0·152||F = 1·90P = 0·206||F = 1·37P = 0·275||F = 0·02P = 0·886|
Five replicate benthic macroinvertebrate samples were collected randomly on each site visit from riffle habitats using a modified 0·05-m2 Surber sampler (250-μm mesh) and were preserved immediately in 70% ethanol. After sorting in the laboratory, macroinvertebrates were identified to species level (where possible) under a light microscope (×40 magnification), but some taxa were identified to higher levels [e.g. Diptera (Family/Genus), Oligochaeta (Class)] using standard keys (see Pawley, Dobson & Fletcher 2011 and references therein). Particulate organic matter (POM) retained in each sample was sorted into fine (<1 mm; FPOM) and coarse fractions [>1 mm; Coarse Particulate Organic Matter (CPOM)], then ashed to determine ash-free dry mass.
Repeated-measures (RM) anova (season as repeated measure) with Bonferroni correction was used to ascertain whether there were significant differences in stream environmental variables as a function of land management. Land management was fixed and season was random. Sites were selected randomly as a ‘representative reach’ for each treatment type, and because the focus of the study was on effects of burning, inter-site comparisons were not considered in detail. One-way anova was used for the single occasion 5v5 survey to determine whether there were differences in stream environmental variables as a function of management type.
Macroinvertebrate community structure was summarized using five measures: (i) log10(total abundance+1) expressed as the total number of individuals per m2; (ii) taxonomic richness; (iii) relative abundance of FFGs assigned following Hynes (1977), Elliott, Humpesch & Macan (1988), Edington & Hildrew (1995) and Wallace, Wallace & Philipson (2003); (iv) 1/Simpson's diversity index (1/S): (Simpson 1949); and (v) taxonomic dominance (D): estimated using the Berger–Parker index:
where Nmax is the number of individuals in the most abundant species and N is total abundance.
Repeated-measures anova and one-way anova were repeated for the macroinvertebrate community metrics for the 3v3 and 5v5 surveys, respectively, using the same methods outlined above for environmental variables. All environmental and macroinvertebrate data sets were tested for normality and, where necessary, 1og10-, arcsin- or square-root-transformed to improve normality and homogeneity of variance prior to statistical tests. All tests were undertaken in SPSS v17.0 (IBM Corporation, Armonk, New York, USA) or Minitab v15.0 (Minitab Inc., State College, PA, USA) and considered significant where P < 0·05. Mauchly's test of sphericity was not violated throughout the rm-anovas.
Taxon–habitat relationships were assessed for both the 3v3 and 5v5 surveys separately using redundancy analysis (RDA) in CANOCO v4.5 (Plant Research International, Wageningen, The Netherlands; Lepš & Šmilauer 2003). Invertebrate abundance data were Hellinger-transformed following Legendre & Gallagher (2001). Forward selection was used to determine which of the stream environmental variables accounted for a significant proportion of the species variance. An initial RDA on the 3v3 survey included a dummy variable ‘Time’ (no. days from start of sampling) to determine whether there were significant seasonal dynamics within the stream macroinvertebrate communities. Following this, a partial RDA (pRDA) was carried out to remove the variance accounted by Time, providing a better indication of the land management and between stream components of the data set (Borcard, Legendre & Drapeau 1992). A standard RDA was conducted on the 5v5 survey as samples were collected only in September 2008.
One-way analysis of similarity (anosim) tested the null hypothesis that differences in stream macroinvertebrate taxa abundance between burned and unburned peatlands were not different to those within the two land management types. anosim was not undertaken to test for seasonal effects in the 3v3 survey owing to the small number of replicates per quarterly sample collection, and because spatial dynamics (linked to management type) were the central focus of this study. anosim was undertaken using the Jaccard's coefficient of similarity (based on taxa presence/absence), with 10 000 permutations and Bonferroni corrections using PAST v2.05 (Hammer, Harper & Ryan 2001).
Rotational vegetation burning effects on stream environmental variables
This study has provided a detailed insight into the spatial and seasonal dynamics of stream environmental variables and macroinvertebrate communities in UK upland rivers influenced by rotational vegetation burning. Both the 3v3 and the 5v5 surveys showed that burning was linked to changes in several stream environmental variables (e.g. increases in SSC, FPOM, Al, SO4, NO3, DOC and smaller D50) allowing H1 to be upheld. These findings are supported in part by evidence from other studies, where the removal of the vegetation cover and litter layer by fire, coupled with wind and rain, can increase vulnerability of the soil to physical erosion, resulting in higher sediment yields being deposited into streams (Tallis 1987; Tucker 2003). Charred peat after burning can also form loose crusts that are broken down easily and washed into streams in overland flow (Tucker 2003).
Higher concentrations of SO4 were found in burned catchment streams compared with the intact sites. Burning removes ‘blocks’ of vegetation, and thus, the exposed peat can be subjected to enhanced drying and oxidation (Maltby, Legg & Proctor 1990; Tucker 2003). The oxidation of reduced sulphur stored in the peat and the mineralization of organic sulphur to dissociated sulphuric acid may explain the observed higher levels of SO4 in this study (e.g. Bottrell et al. 2004; Clark et al. 2005). These findings of increased SO4 in this study were similar to those from artificially drained peatland catchments (Ramchunder, Brown & Holden 2012).
In this study, significantly higher concentrations of DOC were observed in catchments managed via burning compared with intact catchments. Although numerous drivers of increased DOC production have been proposed [e.g. water-table drawdown via drainage (Wallage, Holden & McDonald 2006), warmer temperatures (Tranvik & Jansson 2002) or a reduction in SO4 deposition (Evans et al. 2006a)], this study adds weight to the mounting (but not entirely unequivocal) evidence that burning may be a local driving factor in DOC production operating alongside larger-scale factors. While it should be recognized that we only conducted seasonal spot sampling, intensive sampling by Yallop & Clutterbuck (2009) also documented an increase in DOC concentrations with the greater exposure of peat surface following burning. Furthermore, this relationship was observed for both ‘microscale’ (<3 km2) catchments and larger catchments. Additionally, Yallop, Clutterbuck & Thacker (2012) working in three South Pennine catchments documented elevated humic DOC in catchments with a high proportion of new burns. However, further work is required as data from plot-scale studies to date are not able to account for these catchment-scale patterns (Holden et al. 2012).
Rotational vegetation burning effects on stream macroinvertebrate communities
Both the 3v3 survey and the 5v5 survey revealed significant differences in community richness, 1/S and dominance, and therefore, we rejected H2. This was in contrast to the findings of Ramchunder, Brown & Holden (2012) where artificial drainage had no discernible effect on stream macroinvertebrate community metrics, and from previous forest wildfire research by Minshall, Robinson & Lawrence (1997) and Minshall (2003). Nevertheless, similar findings have been documented by Minshall, Royer & Robinson (2001) and by Vieira et al. (2004) where the authors documented less resistance and resilience to postfire spates. Indeed, the loss of terrestrial vegetation and postfire flooding could have altered the physical properties in the stream channels of the burned catchments in this study. However, studies across a larger number of burned and unburned streams may be necessary to provide a more conclusive insight into burning effects on stream macroinvertebrate community structure.
Stream ecosystem functional group responses following rotational burning are poorly understood, but our results show lower abundance of herbivores and predators in the burned sites partly supporting H3. Furthermore, the ordination analysis demonstrated a shift in the stream macroinvertebrate community from one dominated by mayflies and large predatory stoneflies at the intact sites, to a community dominated by dipterans and smaller stoneflies at burned sites. Individual taxa respond differently to the various physical changes and shifts in food resource, and opportunistic species appear to favour streams impacted by fire (Mihuc & Minshall 1995; Minshall, Royer & Robinson 2001; Minshall 2003). The increase in Chironomidae relative abundance following rotational burning could be related to the elevated organic SSC (e.g. Vieira et al. 2004), or it could be a response to the reduction in predator abundance. Vuori & Joensuu (1996) and Ramchunder, Brown & Holden (2012) found that artificial drainage of peatlands encouraged increased Chironomidae and Simuliidae abundance, suggesting synergies between the stress imparted on stream ecosystems by seemingly disparate artificial drainage and vegetation burning management techniques.
The greater abundance of Amphinemura spp. in the burned catchments from both the 3v3 and the 5v5 surveys suggests that nemourids are more resilient to the effects of rotational burning. These findings are supported by wildfire and postwildfire work by Vieira, Barnes & Mitchell (2011) and Mihuc & Minshall (1995) in the Guaje Canyon, New Mexico and Yellowstone, respectively. Dietary flexibility, life-history strategy (univoltine) and small-body size (therefore able to utilize refugia in microhabitats) may explain the higher abundance of nemourids at the rotationally burned catchments in our study. Both the 3v3 and the 5v5 surveys showed a lower abundance of herbivores, while the 3v3 survey showed a lower abundance of predators in the burned sites, suggesting a strong influence of land use on FFGs. The fine sediment can limit oxygen availability by reducing flow velocities in clogged interstices, reduce interstitial water exchange and constrict the movement of these invertebrates in the substrata (Bo et al. 2007). At present, it is unclear whether burning altered producer biomass, thus depressing herbivore abundance (Vieira et al. 2004), or whether changes in the stream environment were more important for influencing herbivores directly. There is some evidence for the latter because scraper/grazer feeding can be quickly impaired on sediment-smothered surfaces (Larsen & Ormerod 2010).
Implications for peatland and moorland management
In many regions of the world, the biodiversity and ecosystem services of headwater streams have been compromised due to catchment degradation (Harding et al. 1998; Allan 2004). This study suggests that rotational vegetation burning leads to alterations to peatland stream ecosystems, perhaps necessitating focused efforts to restore impacted systems. Although the catchments investigated in this study were <10 km2, and therefore ‘under the radar’ of major management efforts being undertaken as part of the EU Water Framework Directive, the results suggest that a lack of detailed consideration of small headwater systems could be providing inaccurate estimates of the number of watercourses in the different ecological status classes. Structural alterations of macroinvertebrate communities can also influence ecosystem functional processes, and this study suggests that upland managers need to consider ways of reducing the extent or rotation frequency of burning to reduce effects on river ecosystems. There also needs to be more routine monitoring of upland systems such as those that we studied, both to characterize effects of contemporary land management and to monitor whether streams will recover if or when upland management changes are implemented.
Currently, there is a growing focus on the effects of peatland vegetation burning on peat carbon stores and DOC release (Worrall, Armstrong & Adamson 2007; Clay et al. 2009a), while the impacts of burning on stream ecosystems have hitherto remained unknown. This is the first study to document the impacts of peatland vegetation burning on the relationships between physical, chemical and biological communities in river ecosystems and has therefore added significantly to the current knowledge and understanding of rotational burning. It may be that prescribed burning also affects other aquatic organism groups (e.g. algae, microbes, fish) and there is a clear need for more work in this area, particularly given the apparent recent increase in burn frequency and encroachment of prescribed burning onto larger areas of blanket bog (Yallop et al. 2006b). We focused solely on headwater second-order streams and therefore need to examine the effects of upland prescribed burning further downstream to determine the spatial extent of burning impacts (Meyer & Wallace 2001). The generality of the results is difficult to determine at this stage because there have been no other published studies into stream ecosystem responses to heather burning, but ongoing research at different study sites across northern England appears to confirm the findings of this work. The similarities to findings from studies of wildfire in other locations suggest some common effects of vegetation burning and catchment disturbance for stream ecosystems (e.g. Minshall, Robinson & Lawrence 1997; Minshall 2003; Vieira et al. 2004; Mihuc 2005).
The enactment of recommendations and regulations surrounding burning needs to be done with sensitivity to the views of both grouse moor owners and managers and the wider array of groups with interests in upland ecosystems. In particular, we need to improve knowledge exchange between government agencies, managers or upland stakeholders and scientists (Brown et al. 2010). Such exchanges will be important in developing appropriate moorland management regimes to deliver multiple ecosystem services and not just burning heather in rotation to maximize red grouse yields. Peatland fires occur at a global scale (Kuhry & Turunen 2006), and our results suggest that trade-offs are needed to satisfy both economic and ecological facets of the combined social–ecological systems in such areas, especially if fire is implemented as a management tool.
This research was funded by a NERC studentship (NER/S/A/2006/14151) with CASE support from Yorkshire Water, and additional funding from the North Pennines AONB Peatscapes project (ED1113347) and Natural England (SAE03-02-051). Three anonymous reviewers and David Angeler provided insightful comments on the manuscript. Views expressed within this paper are those of the authors and not necessarily those of the agencies who funded the research.