Pattern & Process of Tree Mortality Waves in the Mountains of the Southwestern United States [Alison Macalady]


Published on

Pattern & Process of Tree Mortality Waves in the Mountains of the Southwestern United States. Presented by Alison Macalady at the "Perth II: Global Change and the World's Mountains" conference in Perth, Scotland in September 2010.

Published in: Education
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Pattern & Process of Tree Mortality Waves in the Mountains of the Southwestern United States [Alison Macalady]

  1. 1. Pattern & Process of Tree Mortality Waves in theMountains of the Southwestern United States Alison Macalady1 & Harald Bugmann2,1 1 Laboratory of Tree-Ring Research, University of Arizona 2 Forest Ecology, ETH Zürich, Switzerland Photo: Craig Allen
  2. 2. Photo: Craig Allen
  3. 3. Mortality in the 1950s and 2000s 1950s 2000sAllen and Breshears (1998), PNAS Breshears et al. (2005), PNAS
  4. 4. Mortality mechanisms McDowell et al. (2008), New Phyto
  5. 5. Growth-mortality models 1MortalityProbability ? 0 Index based on radial growth CCR >80%, (e.g. Bigler & Bugmann 2004, Ecol Appl) – growth level over past few years – growth trend over past years to decades – growth sensitivity
  6. 6. Research questions Can the probability of piñon mortality under drought be accurately modeled using indices derived from diameter growth? What do growth-mortality models reveal about the drivers of tree mortality through space and time?
  7. 7. Field sites
  8. 8. Sampling design
  9. 9. Tree growth – typical patterns Large Low growth SEV 2000s release/recovery before death of L trees! Divergence of L TRP 2000s and D trees incited by 1950s drought
  10. 10. Fitting mortality models: one siteSevilleta, 1950s Internal validation: 60% fitting, 40% testing 500 simulations
  11. 11. Fitting mortality models: all sitesSite/period Variable AU ROC CCR meanSEV 1950s 0.89 78.7% sensitivity 50 meanBNM 1950s 0.92 82.0% sensitivity 25 recentSEV 2000s 0.83 75.3% growth 3BNM 2000s – – – growthTRP 2000s 0.67 59.6% difference 15
  12. 12. Validating mortality models Calibration data [shown is CCR]Validation SEV 1950s BNM 1950s SEV 2000sSEV 1950s – 73.1 77.4BNM 1950s 77.4 – 60.0SEV 2000s 55.9 61.7 –BNM 2000s 31.6 16.7 14.3TRP 2000s 53.4 55.9 52.5
  13. 13. What’s going on?High model accuracies associated with 1950’s and SEV2000’s data reflect a chronic stress signal associated withmortality risk •Best predictors reflect the resource status of the trees over different time periods. •Supports carbon starvation mechanism of mortalityLack of fit in N 2000’s models suggests other processes. •Acute drought stress •Increased temps driving accelerated bark beetle/fungi dynamics? •Carbon allocation to defensive compounds (Kane and Kolb 2010, Oikos)?
  14. 14. ConclusionsStrong influence of acutedrought stress and/or barkbeetle/fungi dynamics atnorthern sites in the 2000’sDifferences in space and timean early indicator of globalchange?Challenges of predictingmortality under drought
  15. 15. Acknowledgements…AcknowledgementsCraig Allen, Julio Betancourt, Tom Swetnam, Dave Breshears,Kay Beeley, Collin Haffey, Greg Pederson, Derek Murrow, ChrisBaisan, Rex Adams, Alex Arizpe, Christof BiglerFinancial supportScience Foundation Arizona, US DOE GREF (AM)ETH Zürich, UA Lab. Tree-Ring Research, Haury Fellowship(HB)