SEARCH

SEARCH BY CITATION

Keywords:

  • fire potential;
  • fuel;
  • heathy forest;
  • hummock sedgeland;
  • macropods;
  • tussock grassland

Summary

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

1. Natural area managers use fire and grazing to achieve nature conservation/production goals and to prevent the loss of life and property. Yet, little is known of the effects of post-fire grazing on fuel load and the proportion of days on which fire can be sustained (fire potential). This knowledge could help managers in planning interventions to achieve their goals.

2. At seven sites in Tasmania, Australia, including sedgeland, heathy forest and grassland, fire potential and fuel load were measured before, and for 2 years after fire. Measurements were made in burning, fencing and burning plus fencing treatments, and in control quadrats.

3. Burning followed by grazing, largely by native vertebrates, resulted in lower fuel loads than either grazing by itself or burning by itself. A new steady state was established in two grasslands. Fire potential at the oligotrophic sites was largely a function of time elapsed since the last fire, while at grassy sites was increased by grazing without fire, but depressed or slightly increased by grazing after burning.

4.Synthesis and applications. Effects of grazing after burning on flammability are not predictable from the single or additive effects of grazing and burning, varying between vegetation type and environment. In highland grassy ecosystems fire potential can be reduced by excluding grazing animals after fire, while in scleromorphic ecosystems grazing after fire does not affect fuel or fire potential. Intense grazing after fire can cause an, often desirable, shift from tussock to lawn grassland. Burning and subsequent grazing of tussock grassland vegetation in the lowlands may reduce the chance of wildfire damaging property and conservation/production values, while in highland tussock grassland burning followed by grazing will be largely ineffective in reducing the already low chance of such damaging fire.


Introduction

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

Understanding the influence of fire and grazing on ecosystems is important for both conservation and production (Bond & Archibald 2003; Kirkpatrick, Bridle & Leith 2007; Fuhlendorf et al. 2008). An important component of this understanding is the influence of grazing after burning on vegetation flammability, important in planning interventions for wildfire prevention, nature conservation and stock production. There is a large and rapidly growing literature on the effects of fire regimes on flammability (Marsden-Smedley 2009), a smaller body of work that addresses the effects of grazing regimes on the potential for fire to spread from ignition (Knapp et al. 1999; Blackmore & Vitousek 2000; Valderrabano & Torrano 2000; Williams et al. 2006; Savadogo et al. 2007; Leonard, Kirkpatrick & Marsden-Smedley 2010) and a paucity of information on the influence of grazing after fire on post-fire flammability. Although it is often deduced that grazing following fire will result in a lesser potential for fire than fire by itself (Jauregui et al. 2009; Celaya et al. 2010), there appears to be no direct test of this proposition.

Fire is less discriminating in its consumption of biomass than grazing animals (Bond 2005; Bond & Keeley 2005). Post-fire regeneration is usually more attractive to herbivores than the pre-fire vegetation (Gureja & Owen-Smith 2002; Archibald & Bond 2004; Styger et al. 2010) as a result of the destruction of unpalatable biomass, the widespread tendency for grasses and herbs to be more prominent soon after fire than in long-unburned vegetation (Sachro, Stronga & Gates 2005), the better nutritional quality of the shoots of resprouting grasses than old shoots (Hayes 1985; Heckathorn & DeLucia 1996) and the tendency for resprouts and seedlings of many shrub species to be palatable (Henderson & Keith 2002; Meers & Adams 2003).

Intense herbivory can maintain vegetation in a palatable state (McNaughton 1984; Cromsight & Olff 2008) and highly palatable vegetation that is grazed tends to have low fire potential (Blackmore & Vitousek 2000; Leonard, Kirkpatrick & Marsden-Smedley 2010). Therefore, it could be expected that grazing following burning should reduce flammability more than either grazing or burning on their own. However, this supposition assumes that there is always a pool of grazing animals sufficient to consume any excess forage created by a burn, that grazing cannot increase fire potential and that there are no regenerating plants that are ignored by herbivores. The first assumption may be erroneous in some situations. In southern African savannas, an increase in the proportion of the landscape burned results in a reduction in grazing intensity per unit burned area (Archibald & Bond 2004), indicating that the pool of animals is insufficient to fully consume all of the attractive fodder in the burned areas. The second assumption may not always be true, as grazing of tussock (bunch) grasses can increase the proportion of dead foliage, one of the major controllers of fire potential (Leonard, Kirkpatrick & Marsden-Smedley 2010). The third assumption may also be incorrect in some situations, such as on oligotrophic soils, where slow-growing plants can develop defences against herbivory that may be effective even in the seedling or resprout stage (Turner 1994). It therefore seems that, at the oligotrophic end of a soil fertility continuum, there might be too few grazing animals and too great a regenerating biomass of unpalatable, flammable plant species, for grazing to affect flammability after fire (Kirkpatrick 2007). Thus, one key to predicting the impact of grazing animals on vegetation flammability after fire may lie in the influence of the amount of forage in the pre-fire condition on their populations.

Our study examines the effect of grazing after fire, largely by native animals, on the flammability of tussock grassland, hummock sedgeland and heathy open-forest in Tasmania, Australia. We build on previous work that has determined the effects of grazing on unburnt Tasmanian tussock and lawn grasslands (Leonard, Kirkpatrick & Marsden-Smedley 2010) and the effects of burning of Tasmanian hummock sedgeland and heathy open-forest on macropod densities (Styger et al. 2010). The vegetation types that are covered by the present article are widely subjected to both planned burning and manipulation of the number of herbivores, to reduce the likelihood of loss of adjacent assets, to maintain native biodiversity or to increase economic productivity. As in the rest of the world, this is currently carried out without any experimental understanding of the effects of the combinations of grazing and burning regimes on the chance of further fire.

We test the hypotheses that: (i) grazing after burning decreases fire potential more than grazing alone and burning alone; (ii) grazing after burning decreases fuel more than grazing alone and burning alone and (iii) the magnitude and direction of the relative effects on fire potential and fuel of grazing alone and grazing after fire are caused by vertebrate herbivore responses to the availability of forage in different vegetation types. We then discuss the implications of the results for the achievement of wildfire prevention and conservation/production goals.

Materials and methods

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

Site selection and characteristics

The criteria for the selection of the seven sites (Table 1) were native vertebrate herbivores present, domestic livestock absent, vegetation composed of native species, representation of lowland and highland tussock grassland, lowland heathy open-forest and Gymnoschoenus sphaerocephalus (R. Br.) Hook. f. hummock sedgeland, experimental burning legal and feasible, and, site not burned in the last decade.

Table 1.   Site characteristics
 TG†THLHPPMPIGIB
  1. †TG, Tom Gibson tussock grassland; TH, Tom Gibson heathy open-forest; LH, Lime Bay heathy open-forest; PP, Paradise plains tussock grassland; MP, Mathinna p††lains hummock sedgeland; IG, Iris tussock grassland; IB, Iris hummock sedgeland.

  2. §hf, Eucalyptus amygdalina Labill. heathy open-forest; hs, Gymnoschoenus sphaerocephalus hummock sedgeland; tt, Themeda triandra Forsskal tussock grassland; tpl, Poa labillardierei Steudel tussock grassland; tph, Poa hiemata Vickery tussock grassland.

Latitude (°′S)41 4641 4642 5841 2041 1941 3341 32
Longitude (°′E)147 18147 17147 52147 42147 45145 55145 54
Slope (°)43323313
Aspect (°)216270203311318141322
Altitude (m)20522060810800870900
Annual rainfall (mm)5875987221358131722832322
Daily maximum temperature January (°C)23·323·219·917·617·716·316·1
Daily max temp July (°C)10·410·311·46·26·45·25·0
Daily minimum temperature January (°C)8·88·810·77·17·25·45·3
Daily minimum temperature July (°C)0·50·54·20·40·60·20·1
Loss on ignition (%)5·87·014·820·249·829·083·4
pH5·85·74·35·04·24·03·7
Burn size (ha)0·10·1104·220·7321·484·59·8
Byram intensity (kW m−1)‡17261615121651529771858
Macro scats no burn (ha per day)650490705906058010
Macro scats burn (ha per day)390140120580220670150
Vegetation type§tthfhftplhstphhs
Tree basal area (m2 ha−1)0·722·625·20·00·00·00·0
Grass cover (%)59·211·20·030·47·368·80·4
Herb cover (%)7·60·50·38·64·824·70·5
Sedge, rush and lily cover (%)5·40·12·90·589·51·291·9
Shrub cover (%)9·224·219·421·312·91·917·9
Bracken cover (%)0·00·018·50·00·00·00·0
Cryptogam cover (%)18·35·40·353·13·931·60·5
Litter cover (%)7·766·096·16·20·50·011·1
Bare ground (%)6·03·62·30·13·50·00·0
Rock cover (%)1·80·00·00·00·00·00·0

The sites varied in climate, soil fertility, their use by macropods Marsupialia: Macropodidae, and cover of different plant lifeforms (Table 1). The most numerous vertebrate herbivore at all sites was the Bennett’s wallaby Macropus rufogriseus (Desmarest). The pademelon Thylogale billardierii (Desmarest) and the wombat Vombatus ursinus (Shaw) were also observed at all sites. The common brush-tailed possum Trichosurus vulpecula (Kerr) was present at sites with trees. Rabbits Oryctolagus cuniculus (L.) were present in low numbers at the grassy sites. Sheep Ovis aries (L.) were present in low numbers for about 3 months between one and a half and 2 years after treatment at site TG.

Experimental design

The overall design was fencing nested within burning nested within site. Twenty pairs of 1 × 1 m quadrats were located randomly in the vegetation typical of each site. One of each pair was randomly allocated to be fenced. Half of the pairs were randomly allocated to be burned. Thus, there were three treatments and a control. The control was grazed and unburned (GU). The three treatments were grazed and burned (GB); ungrazed and unburned (UU); and ungrazed and burned (UB). Species cover and height data collected before treatment from the 40 quadrats were subjected to oneway anova to ensure that there were no significant differences between the quadrats allocated to each treatment in vegetation attributes. Burning and fencing took place in spring, except for site LH, which was fenced and burned in autumn. Burning always encompassed more than one pair of quadrats and usually involved protecting those designated not to be burnt from the flames. At site IB, four quadrats that were not allocated to burning were accidentally burned. Four other quadrats were randomly placed in the unburned part of the site and the new configuration checked to ensure similarity between treatments. The above procedures ensured that differences in the outcomes of the four treatments could not be attributed to pre-treatment variation.

The fenced quadrats were enclosed by a 2 × 2 m fence, 1·5 m tall. The fences were constructed of chicken wire with a floppy top and buried or weighted wire on the outer margin. The burns were of low to moderate intensity, and varied in size (Table 1). Variation in the inherent flammability of the different vegetation types made it impossible to achieve constancy in fire intensity and size. However, achievement of constancy was not desirable as inherent variations in burn intensity, fire size and vertebrate herbivore numbers between ecosystems need to be recognised in any realistic appraisal of spatial variation in the effects of fire and grazing exclusion on fire potential and fuels.

Field data collection

Data were collected from the quadrats before burning and fencing, and at 6, 12, 18 and 24 months after treatment. Except for site LH, data were also collected at 1 month after treatment. A quadrat frame with 100 equal divisions was used as an aid to estimate the cover of each discernible vascular plant taxon, cryptogams, litter, rock and bare ground. Rock and bare ground were estimated as plan cover, while the rest were estimated as overlapping cover. The percentage of dead foliage was estimated. The mean height of each taxon and litter in each quadrat was estimated with the aid of multiple measurements. Inflorescences that extended above the foliage were not included in height measurements. One or both of JBK and JMS were involved in all estimates to ensure consistency.

Scats (herbivore dung pellets) were counted and cleared from the unfenced quadrats every 6 weeks. They were grouped as macropod, wombat, rabbit, sheep or possum pellets. Of these, only macropod and sheep pellets gave a reliable indication of grazing activity, as wombats avoid defaecating where they eat, rabbits have communal scat deposition areas and possums feed in trees, as well as on the ground. Almost all pellets at all sites were from macropods, so the macropod scat counts between 1 and 2 years after treatment were used as an indicator of relative grazing intensity.

Data analysis

Total fuel (t ha−1) was estimated from the quadrat cover and height estimates, using the relationships derived by Leonard (2009) and Marsden-Smedley & Catchpole (1995a). All Poaceae and dicotyledonous non-woody plants were assumed to be forage. These are the preferred feed of macropods and wombats (Sprent & McArthur 2002; Evans, Macgregor & Jarman 2006).

Fire potential was defined as the percentage of days on which fire could sustain spread from an ignition point. This was determined for each quadrat at mid afternoon for all days between 12 and 24 months after the experimental burns took place. The percentage of the days on which fire could spread was then calculated. Daily climatic data for the period were collected on site, or from nearby stations for input to the calculations. Models for predicting the threshold between sustaining and non-sustaining fires are available for Tasmanian native grassland (Leonard 2009) and hummock sedgeland (Marsden-Smedley, Catchpole & Pyrke 2001). These models are based on a large number of experimental fires in known weather and fuel conditions. No suitable models for fire potential are available for heathy open-forest. The model for predicting sustaining versus non-sustaining fires in buttongrass moorlands was designed for use in operational fire management. Its inputs are wind speed, fuel moisture, time since fire and site productivity, not the live to dead fuel ratios and fuel loads that were particular to our sites and treatments.

To address the above problems with the direct calculation of fire potential we used flame height as a surrogate. Flame height relates to the ability of fires to sustain across fuel discontinuities and to keep burning with high fuel moistures and low wind speed (Marsden-Smedley, Catchpole & Pyrke 2001). For the grassland quadrats, flame height was predicted using the model of Byram (1959). For the hummock sedgeland quadrats, flame height was predicted using the models in Marsden-Smedley & Catchpole (1995a,b, 2001). In the heathy open-forest sites, the in-forest wind speed was adjusted using a formula derived from the fire prediction spreadsheet developed by fire management agencies in southern Australia (Tolhurst, personal communication.), which in turn was based on relationships derived by McArthur (1967). The formula for calculating the wind speed was:

  • image

In the above formula, the TH site had a vegetation class of 3 and the LH site had a vegetation class of 4. The flame height in the heathy open-forest sites was then predicted using the model of Byram (1959).

The relationship between the percentage of days on which flame height would have exceeded 50 cm was strongly related to fire potential as defined by the models of Leonard (2009) and Marsden-Smedley & Catchpole (2001) in the hummock sedgeland and grassland sites (fire potential = −6·107 + 0·9855 percentage of days that flames exceed 50 cm, r2 = 93·7%, < 0·001). Henceforth in this article, the percentage of days on which flame height could exceed 50 cm is called fire potential.

The general linear model procedure in MINITAB 14 was used for each of fire potential between 12 and 24 months, total fuel at 24 months, the percentage of dead grass at 24 months (for grassy sites only), forage at 24 months and macropod scats between treatment and 24 months, to determine variation between sites and treatments. Fence was nested within burn and site, and burn was nested within site. To give an indication of the relative effects of fire, grazing and grazing after fire on fire potential, percentage of dead grass and total fuel, the effect of fire was approximated as treatment codes UB-UU, the effect of grazing as GU-UU, and the effect of grazing following fire as GB-UU. The significance of the difference between each pair of these values equals the significance of the differences between each pair of GB, GU and UB in the multiple comparisons at the fence (burn site) level in the general linear models. In all general linear models, pairwise comparisons between means were undertaken using the Tukey method at < 0·05.

The percentage of forage consumed on unburned ground by site was approximated by UU-GU and for burned ground by UB-GB. A general linear model in which burn was nested within site was used to determine whether there was variation between sites and variation, and burned and unburned ground for UU-GU and UB-GB.

Results

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

Between 1 and 2 years after fencing and burning, fire potential was strongly affected by site, burning and fencing (Table 2). The hummock sedgeland and heathy open-forest sites could have burned on most days, while the grassland sites seldom would have burned (Table 3). There was a significant negative effect of burning on fire potential at one grassland site (TG) and at all hummock sedgeland and heathy open-forest sites (Table 4). At site TG, fire potential was higher in the grazed and unburned treatment than ungrazed and burned treatment, while at the hummock sedgeland and heathy open-forest sites there were significant differences in the fence level of the model that reflected the burn effect (Table 4). Fire potential was increased by grazing at the grassland sites and decreased by burning at all sites (Table 4). There was no indication of any synergism between burning and grazing effects (Table 4).

Table 2.   Attributes of general linear models for fire potential, total fuel, forage, macropod scat numbers and percentage of dead fuel in grasses (DG %)
Model attributes (site)PotentialFuelForageMacroDG %
F215·7134·264·214·843·6
P<0·001<0·001<0·001<0·001<0·001
F burn74·4122·88·63·118·6
P burn<0·001<0·001<0·0010·004<0·001
F graze (burn)2·30·616·313·36·0
P graze (burn)0·0060·827<0·001<0·001<0·001
r2 (%)87·4186·1972·1953·3262·97
Table 3.   Site means for fire potential (%), total fuel (t ha−1), forage (t ha−1), macropod scats (ha per day) and percentage of dead fuel in grasses (DG %), showing significant differences between means (< 0·05) between sites
SitePotentialFuelForageMacropodDG %
  1. na in the DG% column indicates extremely low levels of grass cover present in that site.

  2. Sites that share any letter in a column are statistically identical.

TG14·46b3·35d0·773ab685·0ab39·4c
TH75·35a11·28c0·100cd388·0bcd27·4d
LH76·93a25·50a0·002d130·2cdna
PP12·54b2·99d0·617b461·7bc63·5a
MP73·42a18·09b0·279c151·7cdna
IG6·86b2·79d0·901a801·7a53·9b
IB61·61a16·96bc0·022d51·4dna
Table 4.   Fire potential (the percentage of days that flame height would have exceed 50 cm at 3 pm), 12–24 months after fire, total fuel t ha−1 at 24 months and percentage dead grass at 24 months in burned treatments [grazed and burned (GB); ungrazed and burned (UB)] and, in the same periods, in the unburned and ungrazed treatment [ungrazed and unburned (UU))] and control [grazed and unburned (GU)]; and the grazing effect (G, GU-UU), the burn effect (B, UB-UU) and the burning and grazing effect (BG, GB-UU), by site
SiteBurn (Site)Fence (Burn site)BGBG
GUGBUUUB
  1. ***< 0·001, **< 0·01.

  2. Site means for different fence (burn site) treatments and effects are statistically identical (> 0·05) if they share a letter.

Fire potential
TG***29·0a5·8ab23·0ab0·1b−22·9b6·0a−17·2ab
TH***83·6a51·9b93·0a72·9b−20·1b−9·4a−41·1b
LH***97·7a53·7b96·9a59·5b−37·4b0·8a−42·2b
PPNS25·013·78·62·8−5·816·45·1
MP***100·0a45·9b100·0a47·8b−52·2b0·0a−54·1b
IGNS15·89·81·80·1a−1·714·08·0
IB***100·0a34·4b100·0a34·4b−65·6b0·0a−65·6b
Total fuel
TGNS2·82a0·69a6·25a3·63a−2·62a−3·43a−5·56a
TH**13·90a6·81a15·00a9·42a−5·58a−1·10a−8·19a
LH***41·05a11·13b37·63a12·18b−25·45b3·42a−26·50b
PPNS3·88a1·70a3·90a2·49a−1·41a−0·02a−2·20a
MP***29·60a6·65b29·21a6·89b−22·32b0·39a−22·56b
IGNS2·23a2·01a3·81a3·10a−0·71a−1·58a−1·80a
IB***30·63a5·27b33·15a5·49b−27·66b−2·52a−27·88b
Dead grass
TG***53a24b53a28b−25b0a−29b
TH***44a11b39a15a−24b5a−28b
PPNS75a66a58a55a−3a17a8a
IGNS72a62ab48bc34c−14b24a14a

The percentage of dead grass at 2 years differed significantly between all the four sites with more than 10% grass cover (Table 2), being higher at the highland than the lowland sites (Table 3). The burned quadrats on the lowland sites had much lower values than the unburned quadrats, while there was no response to burning on the highland sites (Table 4). The percentage of dead grass was increased by grazing at three of these four sites and was unaffected at the other (Table 4). At the lowland sites, grazing following burning had the same negative effect as burning, whereas at PP there was no difference between the burn, graze or burn followed by grazing effects, and at IG the burn effect was negative and significantly different from the positive grazing and grazing following burning effects (Table 4).

Fuel load at 24 months was strongly affected by site and burning, while being little affected by grazing (Table 2). Fuel loads were highest at the sites with the least grass (MP, IB, LH, Table 3). Burning reduced 24 month fuel load at all except the three grassiest sites (TG, PP, IG, Table 4). At all sites there was no difference between the effect of burning and the effect of burning followed by grazing, while the latter value was higher than the former (Table 4). However, at LH and MP grazing increased fuel loads and had a significantly different effect to burning followed by grazing, which markedly reduced them. At IB, the reduction in fuel loads due to grazing was much smaller than the reduction caused by either burning or grazing following burning (Table 4).

At the three grassland sites fuel levels in the grazed and unburned quadrats did not consistently increase during the experimental period (Fig. 1 and Figs S1 and S2 Supporting information), whereas increase was more consistent at the other sites (Fig. 2 and Figs S3–S5 Supporting information). At both TG and IG fuel levels in the ungrazed and unburned quadrats strongly increased relative to the grazed and unburned quadrats, while at the other sites there was little or no difference in tendency between the two. At the same two grasslands, the fuel in the burned and ungrazed quadrats rapidly increased to be greater than that in both the unburned and grazed quadrats and the burned and grazed quadrats. At the other grassland site, PP, the fuel in the ungrazed and burned quadrats did not overtake that in the grazed and unburned quadrats, but it did increase relative to that in the grazed and burned quadrats. At all other sites, fencing did not increase fuel accumulation in burned areas. At two of the grassland sites, TG and PP, grazing in the burned area appeared to prevent fuel levels converging with those in the grazed unburned area, whereas this convergence took place at the other grassland site. At the two forest sites tendencies towards convergence reversed in the last 6 months. At the two hummock sedgeland sites there was a very slow convergence.

image

Figure 1.  Total fuel in the three treatments and control in the tussock grassland at TG showing 95% confidence intervals at the measurement times. Time 1, pretreatment; 2, 1 month after treatment; 3, 6 months after treatment; 4, 1 year after treatment; 5, one and a half years after treatment; 6, 2 years after treatment.

Download figure to PowerPoint

image

Figure 2.  Total fuel in the three treatments and control in the hummock sedgeland at IB showing 95% confidence intervals at the measurement times. Time 1, pretreatment; 2, 1 month after treatment; 3, 6 months after treatment; 4, 1 year after treatment; 5, one and a half years after treatment; 6, 2 years after treatment.

Download figure to PowerPoint

The mean number of macropod scats was highest in the grassland sites and lowest at the hummock sedgeland sites and LH (Table 3). There were no significant differences between burned and unburned quadrats within sites in the model for macropod scats.

Forage was strongly influenced by all factors (Table 2). There was a large amount of forage at the three grassiest sites, a moderate amount at MP and TH and negligible amounts at IB and LH (Table 3). There was a site effect on the percentage of forage consumed (= 10·91, < 0·001), but only a weak burn effect (= 2·03, = 0·081), in a model that explained 42% of the variance. TG and TH had significantly higher values than all other sites. At all sites the mean percentage of forage consumed was higher in the burned quadrats than in the unburned quadrats (Fig. 3). The sites with high amounts of forage tended to also have high macropod scat numbers (Table 3, Fig. 4).

image

Figure 3.  Means and 95% confidence intervals for the percentage of herbs and grasses consumed in the burned and unburned quadrats by site.

Download figure to PowerPoint

image

Figure 4.  Mean and 95% confidence intervals for macropod scat numbers (ha per day) (mac) and forage at 2 years (t ha−1*1000) (for) by site.

Download figure to PowerPoint

Discussion

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

The results of the experiment are inconsistent with a general rule that grazing after burning decreases fire potential more than each of grazing and fire alone (Table 4). A dramatic increase in the proportion of dead grass, and therefore fire potential, as a result of selective grazing within tussocks (Leonard, Kirkpatrick & Marsden-Smedley 2010) did not occur as strongly with grazing after burning as with grazing by itself, being neutralised by burning at the lowland grassy sites and moderated by burning at the highland grassy sites (Table 4). Leaf death in highland tussock grasslands occurred in the winter after the burns, making tussocks susceptible to an increase in flammability through the selective removal of green shoots that sprout in spring and summer. The lowland tussocks remained predominantly green for the 2 years after the burn (Table 4), so could not be selectively grazed into substantially greater flammability.

These results imply that frequent burning followed by grazing in tussock vegetation in the lowlands may be desirable to reduce fire potential, where such a reduction is needed, while grazing after burning of highland tussock grassland will be ineffective in reducing fire potential. Burning by itself is as effective as grazing after burning in reducing fire potential in lowland grasslands, while resulting in only a weak reduction of fire potential in highland areas (Table 4).

There being little edible material in the vegetation of the hummock sedgeland, either before or after fire, selective grazing of green shoots in grass tussocks had a negligible effect on fire potential. Grazing after burning had a greater effect than the sum of grazing and burning effects in heathy open-forest, with its greater component of edible species. There was no substantial effect of grazing on fuel levels, another driver of fire potential, in these scleromorphic vegetation types.

The dominance of scleromorphic shrubs, sedges and rushes in scleromorphic vegetation (Table 1) resulted in time elapsed since fire being the overwhelming influence on fire potential and fuel. The hummock sedgeland and heathy open-forest belong to the ‘black world’ of Bond (2005), in which fire plays the major role in biomass reduction. As in many other vegetation types, fuel rapidly accumulates after fire and keeps on accumulating for decades (Marsden-Smedley 2009), whereas in grass-dominated vegetation, while initial accumulation can be rapid in the absence of grazing, equilibrium is attained within a few years (e.g. Groves 1965; Schwinning & Parsons 1999; Archibald 2008).

Our results further confirm the importance of the highly pyrogenic nature of scleromorphic vegetation (Jackson 1968; Odion, Moritz & DellaSala 2010). Fire potential cannot be mitigated by grazing, either by itself or after fire, in vegetation that almost totally consists of species adapted to resist herbivory. Even the effects of fire in reducing fire potential are short-lived. However, any substantial grass component in vegetation that is dominated by scleromorphic plants makes both grazing and grazing after burning effective in reducing fire potential, as at TH. The much lower fire potential of grasslands than scleromorphic vegetation indicates a potential for grassland vegetation to be used as a firebreak in planned burning directed at hummock sedgeland or heathy vegetation.

As well as contributing to fire potential (Marsden-Smedley & Catchpole 2001; Leonard 2009), total fuel is a strong predictor of the intensity of fire (Cheney 1981; De Luis et al. 2004). The results for total fuel were consistent with hypothesis two, with all sites having a reduction caused by burning followed by grazing greater than either of the reductions caused by grazing or burning, although the difference between the effect of burning and grazing following burning was consistently non-significant. There were significant differences between the effect of grazing and grazing following burning only for the three sites with the most scleromorphic vegetation. These were also the only sites where fuel reduction caused by grazing was significantly less than the fuel reduction caused by burning. The general management implication of these results is that grazing after fire can be more effective than fire itself in reducing fuel levels, but not to any great degree in vegetation dominated by scleromorphic species.

At some grassy sites, there was some indication that the effects on fuel of grazing following burning were persistent (Fig. 1, Fig. S2), conforming to the hypothesis for Tasmania that grazing after a fire can convert flammable tussock-dominated vegetation into non-flammable lawns (Kirkpatrick 2007; Leonard, Kirkpatrick & Marsden-Smedley 2010). This was the opposite outcome to that observed in South Africa (Archibald & Bond 2004), where lawns are more common in unburned savanna than in burned savanna (Archibald 2008). This opposite outcome could possibly be explained by differences in landscape context. Tasmanian grasslands exist as small patches in a matrix of vegetation that produces much less forage than the grasslands themselves, whereas the South African savannas are extensive. In Tasmania, grasslands, whether burned or unburned, are concentration points for herbivores, and extensive fires are likely to reduce overall forage availability at the landscape scale rather than increasing it. In South Africa, animals are diverted away from existing lawns to the nutritious growth produced by fires. As a result, the lawns disappear. In Tasmania, there is no great possibility for diversion, because most of the area burned in extensive fires consists of other vegetation types, which do not produce much forage (Table 3). Grasslands experience more grazing pressure after fire than before fire, because the amount of forage available after a fire, although of higher quality than in the unburned condition, is lower in volume. Thus, at 2 years after fire, burned grasslands supported fewer macropods than unburned vegetation at the same sites (Table 1).

The relative effects on fire potential of grazing alone and grazing after fire seem to be caused by vertebrate herbivore responses to the availability of forage. The higher percentage of forage consumed in the burned quadrats than the unburned quadrats at all sites (Fig. 3) is consistent with the generalisation that regrowth after fire is more attractive to grazing animals than unburned vegetation. This role of fire in producing attractive ‘green pick’ is well-known among land managers (e.g. Cubit 1996) and ecologists (e.g. Moe, Wegge & Kapela 1990; Murphy & Bowman 2007). The positive relationship of herbivore numbers with forage was expected and realised, consistent with a large amount of previous work (e.g. Sachro, Stronga & Gates 2005; Ritchie et al. 2008; Leonard, Kirkpatrick & Marsden-Smedley 2010).

The major application of our work lies in the planning of interventions using fire and/or grazing that best fit protection, production and conservation goals. For instance, we have shown that grazing after burning can be used to help achieve flammability reduction in some vegetation types and environments, but not others. It seems likely that at least some of the results of the present study may have relevance to vegetation elsewhere in the world, as the tussock (bunch) grass form is globally widespread and the ability of grazing animals to remove live leaves from such grasses may be common. However, the presence of bulk feeders, such as cattle Bos primigenius (L.), in a grazing system will reduce the effect, as will burning. The general lesson from the results of our study is that the effects of grazing after burning are not necessarily predictable from the single or additive effects of grazing or burning, or prima facies logic, and need to be explored for each vegetation type and each environment that supports vertebrate herbivores.

Acknowledgements

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

This research was supported by an Australian Research Council grant (DP0665083 ‘Grazing-fire interactions and vegetation dynamics’). Jen Styger, Jen Calder, Vera Markgraf and Kath Dickinson assisted with some of the fieldwork. We thank the Tasmanian Parks and Wildlife Service and Forestry Tasmania for permission to set up experiments on their land and their help with burning.

References

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

Supporting Information

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

Fig. S1. Total fuel in the three treatments and control in the tussock grassland at PP showing 95% confidence intervals at the measurement times.

Fig. S2. Total fuel in the three treatments and control in the tussock grassland at IG showing 95% confidence intervals at the measurement times.

Fig. S3. Total fuel in the three treatments and control in the heathy forest at TH showing 95% confidence intervals at the measurement times.

Fig. S4. Total fuel in the three treatments and control in the heathy forest at LH showing 95% confidence intervals at the measurement times.

Fig. S5. Total fuel in the three treatments and control in the hummock sedgeland at MP showing 95% confidence intervals at the measurement times.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

FilenameFormatSizeDescription
JPE_1962_sm_FigS1-5.doc292KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.