AusCover Earth Observation Services and Data Cubes
Suzanne Prober_Multi-century changes in plant diversity and composition after fire in Eucalyptus salubris communities of the Great Western Woodlands
1. Multi-century changes in plant diversity and composition after fire in
Eucalyptus salubris communities of the Great Western Woodlands
Carl Gosper1,2, Suzanne Prober2 and Colin Yates1
1Department of Environment & Conservation; 2CSIRO Ecosystem Sciences
19 February 2013
2. Overview
• Background: Great Western
Woodlands
• Gimlet plots: early goals GWW Supersite
• Estimating time-since-fire
• Estimating age structure across
GWW
• Multi-century trends in floristic
diversity and PFT composition
• Implications for management
3. The Great Western
Woodlands
World’s largest extant Mediterranean-
climate woodland (16 M ha)
Largely intact, diverse
Mosaic of woodlands, mallee, scrub-
heath, ironstone and greenstone
ranges and salt lakes
Woodlands at as low as 220mm MAR
4. Fire in GWW
Fire is a key process driving vegetation composition, structure and recruitment
Poor understanding of historic fire regimes, including fire intervals
Recent decades have seen many large fires
Woodland communities fire sensitive
5. Gimlet (E. salubris) woodlands
•Serotinous non-resprouter
•Sparse understorey of sclerophyllous shrubs, chenopods, grasses and annuals: fire resistant
•Recruit after fire in even-aged stands, very slow growing
•Increasing fire a concern, especially with climate change
6. Aims:
1. Establish a series of permanently
marked plots in gimlet woodlands along
a time-since-fire gradient
2. Develop a method to estimate time
since fire of long-unburnt stands
3. Estimate the current stand age
structure of Eucalyptus woodlands
4. Assess how diversity indices and plant
functional type (PFT) composition
change with time since fire
7. Step 1: Estimating time since fire
Fire age (y) # plots
1. 72 plots stratified on fire age using Landsat
imagery from 1972 (located within 3 areas 2-10 17
of similar climate and sufficient fire-age 11-38 6
diversity)
38-60 13
>60 36
Fire layer developed from Landsat
Hollowed mature gimlet
8. Step1: Estimating time since fire
2. growth ring counts to estimate post-1910
age and develop size-age relationships
3. growth ring-plant size relationships to
estimate age in stands with hollowed trees
(pre-1910)
9. Step 1: Estimating time since fire 40 (Adj. r2 = 0.992, F1,55 = 7158, P < 0.0001)
Time since fire = -0.890 + 0.9695(growth rings)
Strong positive relationship between 95% Confidence Band
Expected relationship (y = x)
time since fire from Landsat and growth 30
Time since fire (years)
rings
20
Strong positive relationships between 10
growth rings and plant size (diameter at
base and/or tree height) 0
0 10 20 30 40
Mean growth rings per plant
Extrapolation using
Uncertainty over best model form 200
= trunks dated from growth ring counts
square-root
transformed diameter
beyond limits of growth ring record R2=0.882
Time since fire/growth rings
150
Extrapolation using
untransformed
Estimated maximum stand times since 100 diameter
R2=0.877
fire of 370 years (untransformed time) or
95% prediction interval
1460 years (square-root time) 50
0
0 5 10 15 20 25
Diameter at base (cm)
10. Step 1: Estimating time since fire
Key findings:
Gimlet tree rings and size can be used to estimate stand time since fire
Extends time since fire record well beyond that able to be discerned from
Landsat imagery
Allows the investigation of time since fire effects on ecological
processes, using relative age after 100 years
11. Step 2: Age-structure in GWW woodlands
•
Satellite image analysis
Understanding the age structure of GWW woodlands could 80
provide clues as to whether recent levels of woodland fire are
unprecedented
Percentage of woodland area
60
• To provide a crude assessment of the age structure of 40
woodlands across the GWW, we extrapolated the results from
gimlet woodlands and landsat fire mapping to the regional scale,
20
by assuming that the distribution in age classes older than 60
years is proportional to random samples from E. salubris
woodlands 0
0-60 61+
Age class (years since fire)
Area mapped
for fire age
GWW boundary
12. Step 2: Age-structure in GWW woodlands
Key findings:
Estimated age-class distributions can be compared to theoretical distributions to
determine fire management interventions.
Irrespective of model used, recently-burnt vegetation is over-represented on a fixed
proportion basis but less so compared to a negative exponential function.
We are working on improving these estimates
80 80
Untransformed growth rings
Percentage of woodland area
Percentage of woodland area
predicted by diameter and location Square-root growth rings predicted by diameter and location
60 Negative exponential theoretical 60
Negative exponential theoretical age-class distribution
age-class distribution Fixed proportion theoretical age-class distribution
Fixed proportion theoretical
age-class distribution
40 40
20 20
0 0
0 60 100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 100 150 200 250 300 350 400 450 500 550 600 650 700 750 751+
50 0 5060 800
Age class (years since fire) Age class (years since fire)
Untransformed time Square-root time
13. Step 3: Multi-century changes in plant diversity
Hypotheses:
Initial floristic composition model:
declining diversity with time
Intermediate disturbance
hypothesis: highest diversity at
intermediate time since fire
Both hypotheses imply a need for
recurrent fire to maintain
diversity, e.g. GWW shrubland
communities have decreasing
diversity with time since fire
14. Step 3: Multi-century changes in plant diversity
•72 50 x 50 m plots sampled for species composition and cover
•PFTs defined according to plant growth form (7 categories) and seed dispersal
distance potential (2 categories) to form 12 PFTs
Mature
15. Young Mature
4.0
y = 3.313 - 0.194x + 0.011x2
2
Adj. r = 0.304, F2,69 = 16.5, P < 0.0001
3.5
Shannon Diversity
3.0
2.5
2.0
1.5
1.0
0 5 10 15 20 25
Square-root (years since fire)
Results (1):
Intermediate
•globally rare ‘U’-shaped
•graph shows linear model for relationship between diversity
stand age and time since fire is likely to
be driven by dominant trees
and shrubs having maximum
competitive influence at
intermediate times since fire
16. Step 3: Multi-century changes in plant diversity
Results (2):
• Richness and abundance of many PFTs changed with time since fire.
• Dominant trees with short-distance dispersal potential (e.g. gimlet) had peak cover at
intermediate times since fire.
• Shrubs with short-distance dispersal potential had lower cover in mature vegetation.
•Native annual and perennial herbs & grasses with long-distance dispersal potential increased in
mature vegetation
50 a 12
Trees - short dispersal
45 a
Shrubs - short dispersal 10
40
b Annuals - long dispersal
b a
35 a Perennial herbs & grasses - long dispersal
8
Percentage cover
Percentage cover
30 ab
25 6
b b
20 a
4
15 b
b
10
2
5
0 0
Young (< 18) Intermediate (38-120) Mature (> 140) Young (< 18) Intermediate (38-120) Mature (> 140)
Time since fire age class (years) Time since fire age class (years)
17. Step 3: Multi-century changes in plant diversity
Management implications
• As diversity was highest in mature woodlands,
there is no support for gimlet woodlands
requiring recurrent fire to maintain plant
diversity, at least within 400+ year timeframes
• Intense stand-replacing fires at intervals of <
200 yrs would have adverse implications for
biodiversity conservation. Species diversity
would not increase to the community maximum
and the mature vegetation community that
appears distinct in PFT composition would not
develop.
Next steps
•Evaluate time-since-fire impacts on fuels, soil
processes and a range of other organisms
•Improve estimates of woodland age structure
18. Further information:
Parsons & Gosper (2011) International Journal of Wildland Fire 20, 184-194
Prober et al. (2012) Climatic Change 110, 227-248
Gosper et al. (in press) Australian Journal of Botany doi: 10.1071/BT12212
Thank you
Carl Gosper, Suzanne Prober and Colin Yates
t (08) 9333 6442
e carl.gosper@csiro.au
CSIRO ECOSYSTEM SCIENCES AND DEPARTMENT OF ENVIRONMENT & CONSERVATION
19. Further information:
Parsons & Gosper (2011) International Journal of Wildland Fire 20, 184-194
Prober et al. (2012) Climatic Change 110, 227-248
Gosper et al. (in press) Australian Journal of Botany doi: 10.1071/BT12212
Thank you
Carl Gosper, Suzanne Prober and Colin Yates
t (08) 9333 6442
e carl.gosper@csiro.au
CSIRO ECOSYSTEM SCIENCES AND DEPARTMENT OF ENVIRONMENT & CONSERVATION
Notas do Editor
200-400 mm, higher proportion woodlands than at equivalent rainfall in other Mediterranean regions fires propagate through the canopy due to the discontinuity of litter and near-surface fuels, combined with regular severe fire weather in summer Suzanne – could mention Ngadju project here (poor understanding ....) if you wished
* Serotinous non-resprouter, so gimlets are killed by complete canopy scorch. * Recruit after fire and other disturbances in dense, even-aged stands
* We used a combination of methods to determine the time since fire of our sample stands: (i) using fire scars mapped and dated through analysis of Landsat imagery. As the oldest image was taken in 1972, only fires more recent than 1972 can be precisely dated. (ii) growth ring counts in basal stem sections in E. salubris. (iii) using relationships between growth rings and plant size, we estimated the time since fire of stands based on E. salubris size* Transformations of time were used to test for the possibility than plant growth rates may decline with increasing time since fire
* We used a combination of methods to determine the time since fire of our sample stands: (i) using fire scars mapped and dated through analysis of Landsat imagery. As the oldest image was taken in 1972, only fires more recent than 1972 can be precisely dated. (ii) growth ring counts in basal stem sections in E. salubris. (iii) using relationships between growth rings and plant size, we estimated the time since fire of stands based on E. salubris size* Transformations of time were used to test for the possibility than plant growth rates may decline with increasing time since fire
Times since fire were known from Landsat imagery, so this relationship was tested over the time since fire range of 2 to 38 years since fire. The number of growth rings approximates time since fire, although there was a tendency for slightly more growth rings than years since fire, presumably due to E. salubris occasionally developing false rings in addition to annual ones Stem diameter is a useful proxy for plant age Models predicting both untransformed and square-root transformed growth rings from trunk diameter at the base performed well over the period of complete growth ring records. However there is uncertainty over best model form beyond limits of growth ring record There were large differences in the estimated times since fire for the longest-unburnt stands between these two model transformation options
Age-class predictions are based of the assumption that the distribution in age classes older than 38 years is proportional to our random samples of E. salubris woodlands whose time since fire was determined
Age-class predictions are based of the assumption that the distribution in age classes older than 38 years is proportional to our random samples of E. salubris woodlands whose time since fire was determined
The post-hoc division of plots into groups resulted in three groups of similar size and with divisions reflecting gaps in the time since fire range sampled The traits used to define PFTs were selected on the basis of competition and recruitment over the period between fires being important processes driving vegetation change
The post-hoc division of plots into groups resulted in three groups of similar size and with divisions reflecting gaps in the time since fire range sampled The traits used to define PFTs were selected on the basis of competition and recruitment over the period between fires being important processes driving vegetation change
e.g. Of shrub with short dispersal potential: Acacia hemitelese.g. Of annual with long dispersal potential: Calotis hispidulae.g. Of perennial herb and grass with long dispersal potential: Austrostipa elegantissima There were general trends for ground-layer PFTs to increase with increasing time since fire while tree and shrub-layer PFTs decreased; and short dispersal potential PFTs decreased with increasing time since fire while long dispersal potential PFTs increased.
This contrasts some adjoining communities, such as mallee-heath, where recurrent fire does appear important for maintaining plant community vigour and diversity It has been speculated that fires initiate community change from Eucalyptus woodlands to mallee or Acacia shrublands, but we found no evidence for this