Presentation on theme: "Disturbance, Spatial Heterogeneity and Ecosystem Function: Effects of Fire in Yellowstone National Park Monica G. Turner University of Wisconsin-Madison."— Presentation transcript:
Disturbance, Spatial Heterogeneity and Ecosystem Function: Effects of Fire in Yellowstone National Park Monica G. Turner University of Wisconsin-Madison Funded by NSF, USDA, National Geographic The Andrew W. Mellon Foundation
Established in 1872 as the world’s first national park Considered the “crown jewel” of the US park system Encompasses ~900,000 ha Yellowstone National Park
Yellowstone is well known for its many unique natural features…
… and diversity and abundance of native wildlife populations.
The 1988 Yellowstone Fires Burned under conditions of severe drought and high winds Affected ~40% of the park Burned in all ages of forest Stopped by snow in mid September
Fire History in Yellowstone (Romme 1982; Romme and Knight 1982; Romme and Despain 1989) Extensive subalpine forests dominated by lodgepole pine (Pinus contorta) Stand-replacing fires have occurred at 100 to 500 yr intervals throughout the Holocene
Context Heterogeneity and Disturbance Disturbance–key source of spatial and temporal heterogeneity in many ecosystems Large, severe, infrequent disturbances not well understood Many temperate and boreal coniferous forests characterized by infrequent, severe, stand-replacing fire
Outline What are the causes and consequences of postfire heterogeneity? 1.Burn severity 2.Postfire succession 3.Ecosystem function
1. Burn Severity
Postfire Mosaic of Burn Severity Fire spread largely determined by weather Burned through stands of all age Prefire heterogeneity had some but little influence Historic fire suppression (1945-1972) had little effect
Postfire Mosaic of Burn Severity– Summary The 1988 fires produced a spatially complex landscape with patches of varying size, shape and severity. –Of the area affected by crown fire > 50% was within 50 m of a green edge > 75% was within 200 m of a green edge ~16,000 ha > 200 m of green edge How does this landscape mosaic influence postfire vegetation?
2. Postfire Succession Oct. 1988 Same area, July 1989
Postfire Succession Tested hypotheses about effects of the burn mosaic (variation in patch size and fire severity) and environmental variation Nine crown-fire patches of varying size studied since 1990, >700 permanent plots –Small (1 ha) –Moderate (75-200 ha) –Large (500-2700 ha)
Burn severity, patch size, geographic location affected early succession (Turner et al. 1997, Ecol. Monogr., Turner et al. 2003, Frontiers) Species Richness, 1991-2000
Herbs, graminoids and shrubs resprouted in 1989 and flowered profusely in 1990. Seedling recruitment peaked in 1991. Less flowering and few seedlings observed since 1992. Surprise: Survivors dominated.
Surprise: Wide variation in postfire stand structure and serotiny
Pine sapling density, 1990-2000
Burn severity Higher postfire pine densities in severe surface burn than in similar areas of crown fire. Why ? Serotiny Higher postfire pine densities in areas of high prefire serotiny.
Lodgepole Pine Density and Serotiny SitePrefire serotiny (% of stand) 1993 Pine Density Cougar Creek65%21.1/m 2 (211,000/ha) Fern Cascades10%0.23/m 2 (2,300/ha) Yellowstone Lake <1%0.06/m 2 (600/ha)
What explains variation in serotiny? (Schoennagel et al. 2003, Ecology) Serotiny is generally low at high elevations –Fire interval ~300 yrs Serotiny varies with stand age at low elevations –Fire interval ~180 yrs –Young trees (< 70 yr) have low probability of being serotinous
Spatial Variation in Serotiny
How spatially variable is postfire lodgepole pine density across the entire burned landscape, and what explains that variation? 90 plots (0.25 ha) sampled in 1999 for stand structure and function
Variation in Pine Sapling Density >50,000 stems/ha 1,000 stems/ha 0 stems/ha 1999 densities spanned 6 orders of magnitude! Range: 0 – 535,000/ha Mean: 29,380/ha Median; 3,100/ha (n = 90 0.25-ha plots) (Turner et al. 2004, Ecosystems)
Explaining the Variation ANOVA: 36% of variability in PICO sapling density explained by elevation (r = -0.61) and distance to unburned forest –Understanding controls on serotiny (strongly correlated with elevation) is critical link for predicting postfire pine density
1:30,000 color infrared aerial photographs obtained in August 1998 Photos scanned, georectified to produce orthophotos, and classified Best map –Supervised classification using the 90 pts, similar to procedure used with satellite imagery (Kashian et al. 2004, CJFR) –Pine sapling density mapped with 76% accuracy using 5 density classes. Mapping Pine Sapling Density
Densities are < 5,000/ha over ~55% of the burned landscape, but 20% of the landscape has densities > 20,000/ha.
Density of mature stands (60 - 90 years old) 11,000 stems/ha 3,000 stems/ha 1,100 stems/ha What happens over time?
Stand age class 100 0 20 40 60 80 50-100125-175200-250300-350 Coefficient of Variation (%) a a b b 0 Density (trees/ha) 5000 2000 3000 4000 1000 (Kashian et al. In press, Ecosystems)
0 5 10 15 20 25 30 35 40 45 50 0246810 Self-thinning stand Age: 126 years Density: 5,400 stems/ha Serotiny: 53.3% (Kashian et al. In press Ecology)
Filling-in stand Age: 125 years Density: 1,020 stems/ha Serotiny: 0% (Kashian et al. In press Ecology)
Long-term Spatial Heterogeneity Spatial heterogeneity diminishes through successional time –Highest variance in stand structure (and growth rates) in younger age classes –Stand density (and growth rates) converged by 200 years Self-thinning and infilling both contributed to convergence (Kashian et al. In press Ecology and In ptess Ecosystems)
Postfire Succession– Summary Fire severity and patch size influenced plant cover and species richness Postfire landscape is a very heterogeneous mosaic of widely variable stand densities –Serotiny, fire severity especially important The legacy of the spatial variation in stand density created by the 1988 fires may persist for 200 years.
3. Ecosystem Processes How does the spatial variation in postfire vegetation influence ecosystem function? –Aboveground net primary production (ANPP) and leaf area index (LAI) Important indicators of carbon dynamics –Nitrogen (N) cycling System reported to be N limited
A Broader Context Understanding patterns, causes and consequences of spatial heterogeneity in ecosystem processes–a frontier in both ecosystem and landscape ecology –No spatially explicit theory of ecosystem function & few empirical studies –Ecosystem ecology usually focuses on mean rates and change through time –Landscape ecology focuses on spatial variability but not of process rates
ANPP and LAI Developed our own allometric relationships for lodgepole pine and dominant herbaceous species and applied them to vegetation measurements from the 90 0.25-ha plots 1999 ANPP (11-yr old stands) Lodgepole pine: 0 to 14.5 Mg/ha/yr (mean = 1.8)fr –ANPP increased with PICO density and decreased with elevation (ANOVA, r 2 = 0.86, P = 0.0001) Total: 0.04 to 15.0 Mg/ha/yr (mean = 2.9) –Total ANPP increased with PICO density, also influenced by elevation and soil classes (ANOVA, r 2 = 0.80)
ANPP and LAI Pine, herbaceous and total ANPP and LAI influenced by lodgepole pine sapling density –Tree and total positively related to PICO density –Herbaceous negatively related to PICO density Explanatory power –High for tree and total ANPP and LAI –Low for herbaceous ANPP and LAI
Extrapolating to the Landscape If lodgepole pine tree density is known, then ANPP and LAI can be predicted reasonably well across the burned landscape –Pine density and elevation are key predictors
ANPP is already > 2 Mg/ha/yr across 45% of the overall burned landscape, with about 15% of the landscape > 4 Mg/ha/yr.
Nitrogen (N) Cycling Most knowledge of postfire N cycling derived from low-severity fire or other ecosystems (Wan et al. 2001; Smithwick et al. In press, Ecosystems) Spatial/temporal variability in net N mineralization rates following natural, stand- replacing wildfire not well understood No evidence of N loss after the 1988 fires Fires during 2000 burned ~3,000 ha
Fire alters vegetation, productivity and nutrient status… Questions Among stands (broad scale): –How does net N mineralization vary following stand-replacing fire, and what explains this variation? Expected results: initial NH 4 pulse followed by NO 3 pulse; positive correlation between net N mineralization and herbaceous cover
Questions, cont’d Within stands (fine scale): –How variable is net N mineralization within a stand, and what explains this variation? –Is variation in net N mineralization spatially structured? –Is the spatial structure of variability in net N mineralization coincident with that of aboveground cover? Expected results: Little initial spatial structure, then similar scales of spatial autocorrelation in net N mineralization rates and herbaceous cover
Methods Field studies initiated in 2001 in areas burned during summer 2000 –Glade Fire (just S of Yellowstone) 1,280 ha fire in 150-yr old lodgepole pine –Moran Fire (in Grand Tetons) 840 ha fire in >200-yr old Engelmann spruce, subalpine fir, and lodgepole pine Established ten 0.25-ha plots –Five in each fire –All in stand-replacing burns, both crown and severe-surface fire severity
Plot Layout and Sampling Design Within-stand Variation (4 plots) Among-stand Variation (6 plots) n = 20 cores/plotn = 81 cores/plot minimum separation of 2 m
20022003 Glade Moran
Field Measurements Percent cover recorded annually 2001-04 in 0.25-m 2 circular frames (n = 20 or 81 per plot) Net N mineralization measured in situ using resin cores –One-yr incubations –15-cm depth –5-cm diameter PVC cores –Extracted within 24 hr –Analyzed for NO 3 and NH 4
Data Analyses for NO 3, NH 4 and net N mineralization Broad-scale variation among stands –Repeated measures ANOVA with mean percent cover variables for each stand (n = 10 stands) Fine-scale variation within stands –Repeated measures ANOVA (n = 444 cores) Aboveground percent cover around each core –Semivariogram analysis (n = 4 stands, 81 cores/stand) Exponential and spherical models estimated for NO 3, NH 4, net N mineralization, and all percent cover Nugget, sill and range estimated
Results Broad-scale Variation among Stands (N = 10 stands, 95% CI) NO 3 about 10x greater than in similar but older forests NO 3 explained by charred litter (+) coarse wood (-) graminoids (-) (r 2 =0.82)
Results Fine-scale Variation Within Stands
Variation Within Stands: Repeated Measures ANOVA (n = 444 individual cores over 2 years) NO 3 (r 2 = 0.17, P < 0.0001) –Significant variation by year and site –Current year % litter (-) and graminoids NH 4 (r 2 = 0.05, P = 0.0033) –No effect of year and site –% litter (+)
Is variation in net nitrogen mineralization spatially structured? Semivariograms, Glade A, 2001-02 2.1 3.83.5
Fine-scale Variation Within Stands: Spatial Structure (Range) for N Stand2001-022002-03 Glade A NO 3 2.1 m NH 4 3.8 m Net 3.5 m None Glade BNone Moran ANone NO 3 2.1 m NH 4 1.6 m Net 1.8 m Moran BNoneNH 4 12 m
Spatial Structure in Aboveground Cover? Semivariograms, Glade A, Graminoids (Note similar range but increasing sill) 4.8 4.95.4
Within-stand variation: Aboveground Cover (Range in m) Glade Site Stand200120022003 Glade A Charred litter 1.5 Mineral soil 2.6 Rock 2.5 Graminoids 4.8 Charred litter 3.0 Mineral soil 1.6 Litter 1.4 Graminoids 4.9 Charred litter 1.8 Mineral soil 1.7 CWD 2.1 Graminoids 5.4 Lupine 1.4 Glade B Graminoids 1.5 Lupine 1.8 Charred litter 1.9 Rock 2.4 Litter 11.0 Graminoids 2.8 Charred litter 3.8 Mineral soil 3.4 Graminoids 1.3 Lupine 4.4 Forbs 2.1
Are there scale-dependent relationships between cover and net N mineralization? Graminoid patches in Glade A, 2002
Exploratory Multi-scale Analysis Glade A Data Averaged by 4, 6 or 9 Cells
Multi-scale Correlations (r), Glade A 2003 Cover and 2002-03 Net NO 3 Availability Analysis scale % Graminoids (range 5.4 m) % Charred litter (range 1.8 m) Individual core (0.25 m 2, n=81) -0.120.07 4-core mean (2m x 2m, n=9) -0.17 0.29 6-core mean (2m x 4m, n=9) -0.220.01 9-core mean (4m x 6m, n=9) -0.43 0.11
Fine-scale Variation Within Stands N mineralization rates among individual cores not strongly coupled with aboveground cover (low r 2 values) When spatial structure was present, autocorrelation was generally at short distances (< 6 m) –Scales of variation (ranges) in N and cover were similar in magnitude and through time
Fine-scale Variation Within Stands Exploratory analyses suggest scale-dependent relationships between aboveground cover variables and net NO 3 availability –Strongest correlations detected at the scale of spatial structure in the predictor These relationships may become stronger in time
Caveats Mechanisms need further study –microbial communities and activity? carbon? belowground resources? Organic N may also be important Stand-replacing fires create significant spatial heterogeneity at multiple scales –It may be necessary to account for scale- dependent relationships to explain fine-scale variation in ecosystem process rates
Ecosystem Function-Summary Striking spatial variation in postfire landscape structure, ANPP and LAI –ANPP (and LAI) controlled primarily by postfire pine sapling density, with secondary effects of elevation and soils –RECALL! Pine sapling density is a contingent response determined by prefire levels of serotiny and by fire severity and size System appears to conserve nitrogen…stay tuned! Spatial variability in ecosystem function across landscapes is not well understood; our studies contribute to this growing knowledge.
Causes and Consequences of Postfire Heterogeneity Burn severity: fire spread and pattern controlled primarily by weather and produced complex mosaic of burn severities Succession: Fires produce complex mosaic of burn severity and patterns of succession, even in these “relatively simple” systems. –Rapid recovery of community composition and structure –Detectable effects of the postfire mosaic on succession –Legacy in stand structure may persist for 175-200 years
Causes and Consequences of Postfire Heterogeneity Ecosystem process: The spatial variability in structure in turn influences– perhaps even dominates–ecosystem function across the landscape. –Spatial variability after fire is of similar magnitude to variability through successional time –Evidence for functional legacy up to 200 yrs after fire –Intriguing scale dependence bears further investigation