GUARDING AGAINST 
FALSE DISCOVERY 
IN LARGE-SCALE 
DENDRO 
CHRONOLOGY 
Sco! St. George 
University of Minnesota 
Understanding Uncertainties in Paleoclimate and Paleoecology: Age Models, Proxy Processes, and Beyond | December 18, 2014
3ASSERTIONS 
BASED ON 
TREE-RING DATA
COSMIC 
RAYS
“ … THERE WAS A CONSISTANT 
” 
AND SIGNIFICANT RELATIONSHIP 
BETWEEN 
GROWTH OF THE TREES 
AND THE FLUX DENSITY OF 
GALACTIC COSMIC RADIATION. 
Dengel et al., 2009 
New Phytologist
SOLAR 
CYCLE
“ SPECTRAL ANALYSIS ALSO SHOW 
” 
THAT SIGNIFICANT PERIODICITIES 
OF ~11 YEAR, 54 YEAR, AND 204 YEAR PEAKS 
[IN TREE-RING DENSITY RECORDS] 
RESPOND TO 
THE SCHWABE CYCLE, 
THE FOURTH HARMONIC OF THE SUESS CYCLE, 
AND THE SUESS CYCLE, 
RESPECTIVELY. 
Duan and Zhang, 2014 
Journal of Geophysical Research - Atmospheres
ATLANTIC 
MULTIDECADAL 
OSCILLATION
“ TUPELO HONEY YIELD-PER-HIVE 
” 
[RECONSTRUCTED FROM TREE-RING WIDTHS] 
HAS OSCILLATED BETWEEN 
MULTIDECADAL-LENGTH PERIODS OF LOW AND HIGH YIELDS 
DURING THE PAST 211 YEARS 
AND IS ASSOCIATED WITH 
THE ATLANTIC MULTIDECADAL OSCILLATION. 
Maxwell et al., 2013 
Agricultural and Forest Meteorology
LARGE-SCALE 
GEOPHYSICAL 
PHENOMENA 
SMALL SETS 
OF TREE-RING WIDTH 
DATA
BOTTOM-UP 
APPROACH
JUNE TEMPERATURE 
JULY TEMPERATURE 
AUGUST TEMPERATURE 
SEPTEMBER TEMPERATURE 
OCTOBER TEMPERATURE 
NOVEMBER TEMPERATURE 
DECEMBER TEMPERATURE 
JANUARY TEMPERATURE 
FEBRUARY TEMPERATURE 
MARCH TEMPERATURE 
APRIL TEMPERATURE 
MAY TEMPERATURE 
JUNE TEMPERATURE 
JULY TEMPERATURE 
AUGUST TEMPERATURE 
SEPTEMBER TEMPERATURE
p > 0.05
TYPE 1 
ERROR
There are more than 3,200 publicly-available tree-ring records 
Pinus 
Picea 
Quercus 
Pseudotsuga 
Larix 
Nothofagus 
Austrocedris 
Phyllocladus 
Agathis 
Source: St. George, PAGES News, 2014 
(and many more held by individual investigators).
FALSE 
DISCOVERY 
RATE
“ … THE EXPECTED PROPORTION 
” 
OF REJECTED LOCAL NULL HYPOTHESIS 
THAT ARE 
ACTUALLY TRUE. 
Wilks, 2006 
Journal of Applied Meteorology and Climatology
SYNOPTIC 
APPROACH
Pinus 
Picea 
Pseudotsuga 
Quercus 
Larix 
A 
This map illustrates the reasonably-large subset 
of the North American tree-ring width network 
C 
36ºN 
18ºN 
records 
records 
B 
54ºN 
72ºN 
90ºN 
180ºW 120ºW 60ºW 0º 60ºE 120ºE 180ºW 
2000 
1000 
12% 
6% 
0º 
Source: St. George et al., Geophysical Research Le!ers, 2013 
analyzed by this study.
STRONG 
WIDESPREAD 
DIRECT
-0.8 -0.6 -0.4 -0.2 +0.2 +0.4 +0.6 +0.8 
correlation coefficient 
(ring-width vs. climate)
(a) Winter precipitation 
WINTER PRECIPITATION 
Source: St. George, Quaternary Science Reviews, 2014
Fri!s, H.C. Growth-rings of trees: their correlation with climate. Science 154, 973-975 (1966).
EL NIÑO SOUTHERN OSCILLATION
ENSO 
Source: St. George, Quaternary Science Reviews, 2014
WEAK(ER) 
UNCOMMON 
INDIRECT
ATLANTIC 
MULTIDECADAL 
OSCILLATION
Source: St. George, Quaternary Science Reviews, 2014 
ATLANTIC MULTIDECADAL 
OSCILLATION
PACIFIC/ 
NORTH AMERICAN 
PATTERN
(d) Pacific/North-American PACIFIC/NORTH AMERICAN 
PATTERN 
Source: St. George, Quaternary Science Reviews, 2014
ENSO PNA 
475/2270 
139/2270 
records 
records
p > 0.05
TYPE 1 
ERROR
3 POTENTIALLY-USEFUL 
APPROACHES 
TO ADDRESS 
THIS CHALLENGE
STA1TISTICAL 
SIMULATIONS
Dr. Dave Meko University of Arizona
EXACT 
SIMULATION
+200 
0 
precipitation anomaly (mm) 
-200 
decadal target 
decadal ‘noise’ 
1900 1920 1940 1960 1980 2000 
Source: St. George and Ault, Journal of Geophysical Research - Atmospheres, 2011
FO2RWARD 
MODELLING
Dr. Eugene Vaganov Siberian Federal University
Source: Vaganov et al., In Dendroclimatology, 2010
Synthetic tree-ring records generated by a forward model (and modern climate data) 
Can we predict how local trees should respond to weak or remote climate influences? 
oW 
120exhibit largely the same spatial pa!erns in climate response as real trees. 
60oW 
0o 60oE 
120oE 
75oN 
90oN 
SOIL MOISTURE : TEMPERATURE 
Source: Xiaolu (Grace) Li, Cornell University
SE3LECTION 
CRITERIA
Pinus 
Picea 
Quercus 
Pseudotsuga 
Larix 
Nothofagus 
Austrocedris 
Phyllocladus 
Agathis 
Source: St. George, PAGES News, 2014 
Because the global ring-width network is now so large, 
it is crucial to ensure our understanding of tree-environment relations is not influenced 
by decisions to include or exclude particular records.
Guarding against false discovery in large-scale dendroclimatology

Guarding against false discovery in large-scale dendroclimatology

  • 1.
    GUARDING AGAINST FALSEDISCOVERY IN LARGE-SCALE DENDRO CHRONOLOGY Sco! St. George University of Minnesota Understanding Uncertainties in Paleoclimate and Paleoecology: Age Models, Proxy Processes, and Beyond | December 18, 2014
  • 2.
    3ASSERTIONS BASED ON TREE-RING DATA
  • 3.
  • 4.
    “ … THEREWAS A CONSISTANT ” AND SIGNIFICANT RELATIONSHIP BETWEEN GROWTH OF THE TREES AND THE FLUX DENSITY OF GALACTIC COSMIC RADIATION. Dengel et al., 2009 New Phytologist
  • 5.
  • 6.
    “ SPECTRAL ANALYSISALSO SHOW ” THAT SIGNIFICANT PERIODICITIES OF ~11 YEAR, 54 YEAR, AND 204 YEAR PEAKS [IN TREE-RING DENSITY RECORDS] RESPOND TO THE SCHWABE CYCLE, THE FOURTH HARMONIC OF THE SUESS CYCLE, AND THE SUESS CYCLE, RESPECTIVELY. Duan and Zhang, 2014 Journal of Geophysical Research - Atmospheres
  • 7.
  • 8.
    “ TUPELO HONEYYIELD-PER-HIVE ” [RECONSTRUCTED FROM TREE-RING WIDTHS] HAS OSCILLATED BETWEEN MULTIDECADAL-LENGTH PERIODS OF LOW AND HIGH YIELDS DURING THE PAST 211 YEARS AND IS ASSOCIATED WITH THE ATLANTIC MULTIDECADAL OSCILLATION. Maxwell et al., 2013 Agricultural and Forest Meteorology
  • 9.
    LARGE-SCALE GEOPHYSICAL PHENOMENA SMALL SETS OF TREE-RING WIDTH DATA
  • 10.
  • 11.
    JUNE TEMPERATURE JULYTEMPERATURE AUGUST TEMPERATURE SEPTEMBER TEMPERATURE OCTOBER TEMPERATURE NOVEMBER TEMPERATURE DECEMBER TEMPERATURE JANUARY TEMPERATURE FEBRUARY TEMPERATURE MARCH TEMPERATURE APRIL TEMPERATURE MAY TEMPERATURE JUNE TEMPERATURE JULY TEMPERATURE AUGUST TEMPERATURE SEPTEMBER TEMPERATURE
  • 12.
  • 13.
  • 14.
    There are morethan 3,200 publicly-available tree-ring records Pinus Picea Quercus Pseudotsuga Larix Nothofagus Austrocedris Phyllocladus Agathis Source: St. George, PAGES News, 2014 (and many more held by individual investigators).
  • 15.
  • 16.
    “ … THEEXPECTED PROPORTION ” OF REJECTED LOCAL NULL HYPOTHESIS THAT ARE ACTUALLY TRUE. Wilks, 2006 Journal of Applied Meteorology and Climatology
  • 17.
  • 19.
    Pinus Picea Pseudotsuga Quercus Larix A This map illustrates the reasonably-large subset of the North American tree-ring width network C 36ºN 18ºN records records B 54ºN 72ºN 90ºN 180ºW 120ºW 60ºW 0º 60ºE 120ºE 180ºW 2000 1000 12% 6% 0º Source: St. George et al., Geophysical Research Le!ers, 2013 analyzed by this study.
  • 20.
  • 21.
    -0.8 -0.6 -0.4-0.2 +0.2 +0.4 +0.6 +0.8 correlation coefficient (ring-width vs. climate)
  • 23.
    (a) Winter precipitation WINTER PRECIPITATION Source: St. George, Quaternary Science Reviews, 2014
  • 25.
    Fri!s, H.C. Growth-ringsof trees: their correlation with climate. Science 154, 973-975 (1966).
  • 26.
    EL NIÑO SOUTHERNOSCILLATION
  • 27.
    ENSO Source: St.George, Quaternary Science Reviews, 2014
  • 28.
  • 29.
  • 30.
    Source: St. George,Quaternary Science Reviews, 2014 ATLANTIC MULTIDECADAL OSCILLATION
  • 31.
  • 32.
    (d) Pacific/North-American PACIFIC/NORTHAMERICAN PATTERN Source: St. George, Quaternary Science Reviews, 2014
  • 33.
    ENSO PNA 475/2270 139/2270 records records
  • 34.
  • 35.
  • 36.
    3 POTENTIALLY-USEFUL APPROACHES TO ADDRESS THIS CHALLENGE
  • 37.
  • 38.
    Dr. Dave MekoUniversity of Arizona
  • 40.
  • 41.
    +200 0 precipitationanomaly (mm) -200 decadal target decadal ‘noise’ 1900 1920 1940 1960 1980 2000 Source: St. George and Ault, Journal of Geophysical Research - Atmospheres, 2011
  • 42.
  • 43.
    Dr. Eugene VaganovSiberian Federal University
  • 44.
    Source: Vaganov etal., In Dendroclimatology, 2010
  • 45.
    Synthetic tree-ring recordsgenerated by a forward model (and modern climate data) Can we predict how local trees should respond to weak or remote climate influences? oW 120exhibit largely the same spatial pa!erns in climate response as real trees. 60oW 0o 60oE 120oE 75oN 90oN SOIL MOISTURE : TEMPERATURE Source: Xiaolu (Grace) Li, Cornell University
  • 46.
  • 47.
    Pinus Picea Quercus Pseudotsuga Larix Nothofagus Austrocedris Phyllocladus Agathis Source: St. George, PAGES News, 2014 Because the global ring-width network is now so large, it is crucial to ensure our understanding of tree-environment relations is not influenced by decisions to include or exclude particular records.