SO-EM-06-04
TITLE: Regional oak decline-related and regular oak mortality predictions of extent and severity: including
factor identification, and probability modeling.
LOCATION: Ozark Highlands in Arkansas and Missouri (Ecological section 222A: Ozark highlands)
DURATION: 2 years
PROJECT LEADER:
FUNDING SOURCE: Base
Martin A. Spetich
Research Forester
Southern Research Station
USDA Forest Service
PO Box 1270
Hot Springs, AR 71902
mspetich@fs.fed.us
Zhaofei Fan (Co-principal Investigator)
Quantitative Ecologist
Department of Forestry
University of Missouri
Columbia, MO 65211
(573) 875-5341
fanzha@missouri.edu
COOPERATORS:
Stephen Shifley and John Kabrick
Research Foresters
North Central Research Stn.
USDA Forest Service
202 ABNR, Univ. of Missouri
Columbia, MO 65211 (573) 875-5341
W. Keith Moser
Research Forester
North Central Research Stn.
USDA Forest Service
1992 Folwell Avenue
St. Paul, MN 55108 (651)649-5155
Bruce Moltzan and Rob Lawrence
Resource Scientists
Missouri Dept. of Conservation
1110 S. College Ave.
Columbia, MO 65201
(573) 882-9880
PROJECT OBJECTIVES:
ƒ Quantify historical rates of black, scarlet and northern red oak mortality in the Ozark Highlands (Ecological
section 222A) as a baseline against which to evaluate current levels of oak mortality.
ƒ Use classification and regression tree (CART) analysis and/or related classification techniques to determine the
tree and plot factors that are the best indicators of oak mortality.
ƒ Determine the relative importance of factors that are determined to be statistically significant indicators of oak
mortality.
ƒ Apply the CART model using the latest available data from the Arkansas and Missouri FIA state-wide
inventories (Cycle 5) to identify risk levels (severity).
ƒ Use the CART model to identify FIA plots measured in the most recent Arkansas and Missouri state-wide
inventories that appear to show high risk (severity) of future oak mortality & map predicted extent.
JUSTIFICATION: In the Ozark Highlands of northern Arkansas and southeast Missouri there is considerable
concern about oak decline (Oak et al. 2004, Starkey et al. 2004, Spetich 2004 and 2006). In the Ozark National
Forest of Arkansas alone, at least 300,000 acres have been severely impacted (Starkey et al. 2004). Red oak
group species (Quercus spp. L.; subgenus Erythrobalanus) appear to be particularly susceptible, especially those
that are large or physiologically mature, in dense stands, and growing on droughty sites (Law and Gott 1987;
Starkey and Oak 1989, Starkey et al. 2004, Johnson et al. 2002). A number of related agents such as the red oak
borer are also involved in the decline and mortality complex (Voelker and Muzika 2004). Forests of the Ozark
Highland ecoregion (Ecological section 222A) (Keys et al. 1995) are highly susceptible to oak decline in the coming
decades (Lawrence et al. 2002). Approximately 98 percent of all scarlet oak volume, 83 percent of all black oak
volume, 54 percent of all northern red oak volume in Missouri and Arkansas occurs in this ecoregion. There are
statistical techniques and data available to identify the factors associated with oak decline. Those factors can be
used to develop a range of models (simple to complex) that can be applied with FIA/FHM data (or other sources of
data) to identify tree, stand, or site conditions and geographic subregions of concern for future mortality. The same
suite of models can be applied by silviculturists and forest managers to mitigate oak mortality by identifying trees
and sites where future mortality is likely to be unacceptably high.
DESCRIPTION:
a. Background: Recent research on oak mortality from the Missouri Ozark Forest Ecosystem Project (MOFEP)
(Brookshire and Shifley 1997, Shifley and Brookshire 2000, Shifley and Kabrick 2002) has demonstrated that
species, crown class, and tree size are significant indicators of the probability of oak mortality (Kabrick et al. 2004).
Those results are based on a time period covering approximately 1991 to 2002. Oak decline is a periodic event
and recent oak mortality rates may or may not be consistent for earlier decades. However there are historical data
that can be used to test the temporal consistency of oak mortality rates. A number of analytical and modeling
techniques can be used to identify the factors most associated with future oak mortality. Classification and
regression tree analysis or CART (Breiman et al. 1984) has been shown to be particularly useful at identification of
factors associated with oak Mortality (Kabrick et al. 2004). Additional unpublished analyses give preliminary
indications that oak diameter growth rates are a particularly good indicator of future mortality rates. This is a logical
outcome and consistent with some past models of tree survival (e.g. Buchman et al. 1983). For most forest
management inventories, individual-tree diameter growth rates are not available (nor cost-effective to measure).
However, diameter growth rates are regularly observed on FIA plots. Analyses based on a variety of data sources
1
SO-EM-06-04
should identify threshold diameter growth rates that, in conjunction with other tree, plot, and/or climate variables,
are indicative of high rates of future oak mortality. When such relationships are verified, they can be applied with
current inventory data (e.g., FIA) to identify plots and regions that appear to be at comparatively high risk of future
oak mortality.
b. Methods: Use CART and other statistical analysis tools to identify hierarchical relationships among tree, stand
and/or site factors that are associated with varying levels of oak mortality.
ƒ Evaluate data collected at different points in time from 1957 through 2002 to determine if the patterns are
consistent over time.
ƒ Specifically evaluate the relationship of a tree’s recent diameter growth rate to its subsequent survival.
ƒ Use CART analyses to create a probability based model of oak survival. Previous work indicates that models
will most likely be based on tree species, crown class, and diameter growth rate (and possibly other factors).
We will explore links between mortality and drought indices over the various plot remeasurement intervals
found in the available data.
ƒ Apply the model to current FIA data to identify (and possibly map) plots that have a high probability of future
oak mortality.
c. Products:
ƒ A ranked list of the factors that are the best indicators of oak mortality (or survival).
ƒ An associated model of the mortality probability expected for each level (or value) of each significant factor.
ƒ An evaluation of the model performance for historical data datasets with initial measurements dating back as
far as 1952.
ƒ Application of the best model(s) to current (cycle 5) FIA data for the Arkansas and Missouri Ozarks to identify
and map regions (extent) where future oak mortality is expected by severity (risk) class.
d. Schedule of Activities:
Based on time from date of award receipt
Year 1
Award + 6 months
Data assembly and initial summary
Award + 12 months
Initial model development and factor identification
Year 2
Award + 4 months
Model testing across data sets
Award + 7 months
Application to Arkansas and Missouri cycle 5 FIA data
Award + 12 months
Mapping, final report Revision and draft publication.
e. Progress/Accomplishments: We analyzed data from both the managed (Pioneer Forest) and unmanaged
(MOFEP sites) stands in Missouri. We built the CART models to identify major risk factors related to oak decline
and mortality and predict tree mortality based on these risk factors. With declining trees/stands, we also explored
the distribution of crown dieback class, oak borer attack and mortality rate for different species (groups).
Oak decline is an age-related episode in the Ozark Highlands of Missouri and Arkansas. Crown class, basal area in
larger trees (an index to reflect the competition from above) and dbh are the major risk factors of red oak decline
and mortality (Figures 1 and 2). For the red oak group species, CART identifies 5 and 7 risk groups for the
managed and unmanaged stands, respectively, and with annual mortality ranging from less than 1 to 7% (Shifley et
al. 2006). These models provide useful information for silvicultural treatments designed to alleviate oak decline and
mortality.
2.2
2.3
III
III,IV
I, II
crown
class
I, II
crown class
1.4
4.3
4.7
1.2
IV
>13.5
< 13.5
bal
6.3
3.6
Grp 1
Grp 2
II
1.7
0.8
Grp 5
>13.1
III
crown
class
I
crown class
7.3
>15.7
3.8
2.5
>24
>14.1
2.9
1.3
Grp 3
Grp 4
Figure1. Annual mortality (%) of red oak group
species by risk factors in managed stands (the
Pioneer Forest). [I = dominant;
II = codominant; III = intermediate; IV =
suppressed; bal = basal area in larger trees.
Labels in Figure 1 are also applied to Figure 2]
1.3
Grp 1
< 13.1
bal
< 15.7
bal
< 24
dbh
<14.1
bal
>30
< 30
dbh
4.3
2.0
3.8
1.9
1.5
0.9
Grp 2
Grp 3
Grp 4
Grp 5
Grp 6
Grp 7
Figure2. Annual mortality (%) of red oak group
species by risk factors in unmanaged stands (the
MOFEP Forest).
2
SO-EM-06-04
We also explored the distribution of crown condition (dieback) and oak borer attack. In MOFEP sites, the proportion
of trees suffering moderate or severe crown dieback is 9, 11, 6 and 4%, and the proportion of trees suffering
moderate and severe borer attack is 28, 24, 2 and 7% among scarlet oak, black oak, white oaks (white oak and
post oak) and other species, respectively. There is a strong association between crown dieback condition and oak
borer abundance.
We will use the FIA data (Cycle 5) to test CART models and map the oak decline/mortality across the Ozark
Highlands of Arkansas and Missouri in the coming year.
COSTS:
Item
Requested
Funding
Other-Source Funding
Source
YEAR 2006
Administration
Procurements
TOTAL YEAR 2006
YEAR 2007
Administration
Procurements
TOTAL YEAR 2007
TOTAL PROJECT
Salary and Benefits
Research Associate (5
mo.)
Southern Res, Stn
Travel
Supplies
Salary and Benefits
Research Associate (7
mo.)
Graduate Helper (6 mo.)
Overhead
Travel
Supplies
38,376
0*
3,000
500
$ 41,876*
39,736
13,162
0
1,500
500
$ 54,898
$ 96,774
Contributed salary (Spetich,
Kabrick, Shifley, Moser;
Moltzan, Lawrence) $25,000
$3,000
SRS, NCRS,
and MO
Dept. Cons.
SRS, NCRS
$28,000 (in-kind)
Contributed salary (Spetich,
Kabrick, Shifley, Moser;
Moltzan, Lawrence) $25,000
$3,000
SRS, NCRS,
and MO
Dept. Cons.
SRS, NCRS
$28,000 (in-kind)
$56,000 (in-kind)
* Note: The originally budgeted $6,592 “Southern Res. Stn” dollars were not needed for SRS overhead. These monies have been
moved to the 2007 budget request for 3 months of the graduate student help.
Literature Cited:
Breiman, L., Friedman, J. H., Olshen, R. A., and Stone, C. J. 1984. Classification and regression trees. Monterey, CA: Wadsworth and Brooks.
368p.
Brookshire, B.L.; Shifley, S.R. (eds.). 1997. Proceedings of the Missouri Ozark Forest Ecosystem Project Symposium: an experimental
approach to landscape research; Gen. Tech. Rep. NC-193. 378 p.
Buchman, R.G.; Pederson, S.P.; Walters, N.R. 1983. A tree survival model with application to species of the Great Lakes. Canadian Journal of
Forest Research 13:601-608.
Johnson, P.S.; Shifley, S.R.; Rogers, R. 2002. The ecology and silviculture of oaks. Wallingford, Oxon, UK: CABI Publishing, CAB
International. 503p.
Kabrick, J.M ; Shifley, S.R.;.Jensen, R.G.; Fan, Z.; Larsen, D.R. 2004. Factors associated with oak mortality in Missouri Ozark forests. Gen.
Tech. Rep. NE-316:27-35.
Keys, Jr., J. ; Carpenter, C.; Hooks, S.; Koenig, F; McNab, W.H.; Russell, W.E.; Smith, M-L. 1995. Ecological units of the eastern United
States—first approximation (map and booklet of map tables). Atlanta, GA: US Department of Agriculture, Forest Service.
Law, J.R.; Gott, J.D. 1987. Oak mortality in the Missouri Ozarks. In: Hay, R.L.; Woods, F.W.; DeSelm, H.; Proceedings of the Central Hardwood
Forest Conference VI; 1987 February 24-26; Knoxville, TN. 427-436.
Lawrence, R.; Moltzan, B.; Moser, W.K. 2002. Oak Decline and the Future of Missouri’s Forests. Missouri conservationsit 63(7):11-18.
Oak, S.W., Steinman, J.R., Starkey, D.A.,, Yockey, D.K. 2004. Assessing oak decline incidence and distribution in the southern U.S. using
forest inventory and analysis data. Gen. Tech. Rep. SRS-73:236-242.
Shifley, S.R.; Brookshire, B.L. (eds.). 2000. Missouri Ozark Forest Ecosystem Project: site history, soils, landforms, woody and herbaceous
vegetation, down wood, and inventory methods for the landscape experiment. Gen. Tech. Rep. NC-208. 314 p.
Shifley, S.R.; Kabrick, J.M. (eds.). 2002. Proceedings of the second Missouri Ozark Forest Ecosystem Project symposium: post treatment
results of the landscape experiment. Gen. Tech. Rep. NC-227. 228 p.
Shifley, S. R., Fan, Z., Kabrick, J. M., and Jensen, R. G. 2006 Oak mortality risk factors and mortality estimation. For. Eco. Manage. 229:16-26)
Spetich, M.A. 2004. Upland oak ecology symposium: a synthesis. Gen. Tech. Rep. SRS-73. 311p.
Spetich, M.A. 2006. Early changes in physical tree characteristics during an oak decline event in the Ozark Highlands. SRS-92. 640p
Starkey, D.A.; Oak, S.W. 1989. Site factors and stand conditions associated with oak decline in southern upland hardwood forests. In: Gen.
Tech. Rep. NC-132.
Starkey, D.A., Oliveria, F., Mangini, A. and Mielke, M. 2004. Oak decline and red oak borer in the Interior Highlands of Arkansas and Missouri:
natural phenomena, severe Occurrences. Gen. Tech. Rep. SRS-73: 217-222.
Voelker, S.L.; Muzika, R-M. 2004. Decline of red oaks in the Missouri Ozarks: the story continues. In: Gen. Tech. Rep. NE-316.
3