Fire History and Age Structure Analysis
in the Sherburne National Wildlife Refuge:
Establishing Reference Conditions in a
­Remnant Oak Savanna Woodland
Kurt F. Kipfmueller1 and Tim Hepola2
Abstract—Oak savanna woodlands were once a dominant ecotone throughout the
upper Midwest. These ecosystems represented a transitional zone between prairie
communities to the west that eventually graded into Big Woods forest. Most of the
oak savanna landscapes of most of the Midwest were extensively homesteaded and
farmed during the middle 1800s and few intact savanna landscapes remain today.
Given the current interest in preserving, maintaining, and restoring these systems, it is
imperative that the natural factors that have shaped these areas are investigated. This
research investigates the potential of developing reference conditions in a relatively
intact oak savanna in the Sherburne National Wildlife Refuge, Minnesota. This research
provides a context for current management activities centered on maintaining and
restoring oak savanna ecosystems.
Introduction
Oak savanna woodlands were once a dominant ecotone in southwestern
Minnesota and throughout the upper Midwest. These ecosystems represented
a transitional zone between prairie communities to the west that eventually
graded into Big Woods forest. Most of the oak savanna landscape of southern
Minnesota (and indeed most of the Midwest) were extensively homesteaded
and farmed during the middle 1800s and few intact savanna landscapes remain today.
The structure, origin, and factors that have maintained presettlement oak
savanna have not been well documented or established. Fires are thought
to have maintained these landscapes, perhaps ignited by Native Americans,
but the disturbance regime of these landscapes prior to widespread EuroAmerican impact remains elusive. Given the current interest in preserving,
maintaining, and restoring these systems, it is imperative that the natural
factors that have shaped these areas are investigated. Prescribed fires have
become an important tool in this regard, but the fire management activities
can be better implemented if the natural role of fire is known.
Reconstructing the historic fire regime in oak savanna woodlands is
challenging because few are of sufficient age to extend to periods before appreciable impacts by human activities. However, this information is critical
for establishing reference conditions that can be utilized to guide restoration
activities and initiate management activities.
USDA Forest Service Proceedings RMRS-P-46CD. 2007. In: Butler, Bret W.; Cook, Wayne,
comps. 2007. The fire ­environment—
innovations, management, and ­policy;
conference proceedings. 26-30 March
2 0 0 7; D e s t i n , F L . P ro cee d i ng s
R MRS-P-46CD. Fort Collins, CO:
U. S. Department of ­ Agriculture,
Forest Ser v ice, Rock y Mou nta i n
Research Station. 662 p. CD-ROM.
1 Assistant Professor, Department of
Geography, University of Minnesota,
Minneapolis, MN. kurt@umn.edu
2 Federal Cooperator and Regional
Fire Ecologist, Region 3, U.S. Fish and
Wildlife Service, Fort Snelling, MN.
507
Fire History and Age Structure Analysis in the Sherburne National Wildlife Refuge:…
Kipfmueller and Hepola
This research will investigate the potential of developing reference conditions in a relatively intact oak savanna in the Sherburne National Wildlife
Refuge, Minnesota. Vegetation plots will be established throughout the
remnant to determine the variability in savanna structure and composition.
Fire scars will be collected throughout the remnant to determine the variability in fire frequency and associated changes in savanna age-structure and
composition related to fire activity.
This research provides a context for current management activities centered
on maintaining and restoring oak savanna ecosystems. The principal questions to be answered include:
• What is the fire frequency of the oak savanna remnant and how much
variation exists over space?
• How has fire frequency changed over time?
• Are there particular time periods where fire is more/less prominent?
• Are forest compositional and structural characteristics related to fire
frequency?
• Has the seasonality of fire changed over time?
Identifying the historic fire regime of oak savanna landscapes, the associated forest structures, and spatial variations is critical to developing sound
management guidelines for the restoration of these landscapes. Since this study
area is one of the few old-growth remnants of this ecosystem in the refuge,
information developed in this study will be useful to management of other
parts of the refuge. Few detailed investigations of oak savanna fire history
and age structure have been completed. This research will add substantially
to our understanding of the vegetation dynamics and disturbance regime
that characterize this landscape.
The information in this paper is taken from the study’s “Progress Report,
Spring 2007.”
Field Sampling
Field sampling for age structure and compositional patterns was completed
during summer 2006 at 55 points within the study area. Our sampling used
an alternating pattern of grid points. At every other grid point we collected
increment cores for age determination. Each grid point was inventoried for
compositional attributes. Age structure data were collected from 28 points.
Our original design included sampling of 65 grid points. However, some
points fell in wetland areas with little or no tree cover and some points were
omitted due to their proximity to a cemetery, road, or other impediment to
sampling.
Age and composition data collection is complete for the study area; however,
we have not collected fire-scar samples as of March 2007. Fire scar samples
will be collected during spring and summer 2007 outside the time period
where oaks are susceptible to oak wilt.
USDA Forest Service Proceedings RMRS-P-46CD. 2007. 508
Fire History and Age Structure Analysis in the Sherburne National Wildlife Refuge:…
Kipfmueller and Hepola
Stand Composition
A great deal of variation in stand composition exists across the study area
(table 1). This is mostly manifested as differences in relative amounts of pin
and bur oak as stem density and basal area are quite variable. The study area
is largely composed of northern pin oak that dominates both basal area and
stem density (and therefore species importance) with lesser amounts of bur
oak. In fact, only a limited number of sites are dominated by bur oak. Of
the grid points sampled, 92.7 percent contained pin oak and/or bur oak.
Aspen and sugar maple were also present at a few sites, sometimes in large
numbers.
Table 1—Species identification code for tree species located in plots. Number inventoried indicates the
number of each species present in all plots combined. Codes are used in figures in this paper.
Four-letter
code
QUEL
QUMA
POTR
POGR
ACSA
ACRU
FRPE
FRNI
ULAM
OSVI
PRPE
PRVI
PRSE
SASP
Scientific name
Common name
Quercus ellipsoidalis
Quercus macrocarpa
Populus tremuloides
Populus grandidentata
Acer saccharum
Acer rubrum
Fraxinus pennsylvanica
Fraxinus nigra
Ulmus Americana
Ostrya virginiana
Prunus pensylvanica
Prunus virginiana
Prunus serotina
Salix sp.
Northern pin oak
Bur oak
Quaking aspen
Big-tooth aspen
Sugar maple
Red maple
Green ash
Black ash
American elm
Eastern hophornbeam
Pin cherry
Choke cherry
Black cherry
Willow
Number inventoried
1,386
515
241
15
75
18
31
43
9
5
7
7
14
1
Stem density (fig. 1, 2) and basal area (fig. 3, 4) appear to have no spatial
pattern at a landscape level. That is, there does not initially appear to be
any clustering of large/small values (fig. 5). However, basal area and stem
density appear greatest south of the road running east-west through the
study area. This has not been investigated using a spatial autocorrelation or
variogram approach as of yet to determine if there are spatial patterns that
can be quantified.
Tree diversity is generally higher south of the road and includes sugar maple,
ash, and cherry in greater abundance than are found elsewhere within the
study area (fig. 6). Corylus sp. is present in high abundance at nearly every
sampled grid point (note we have not yet summarized sapling/shrub data).
There are places that have less Corylus than others. This may be related to
some aspect of the environmental conditions within a stand but more likely
reflects fire history to some degree. We suspect that those areas that are
burned most frequently might be enabling hazel to germinate readily from
root stock. This has led to enormously high stem densities in some areas of
the savanna at the sapling/shrub level.
USDA Forest Service Proceedings RMRS-P-46CD. 2007.
509
Fire History and Age Structure Analysis in the Sherburne National Wildlife Refuge:…
Kipfmueller and Hepola
Stem Density Information
Stem Density (stems/ha)
(A)
1000
n=55
Stem Density (stems/ha)
800
600
400
200
PR
VI
PR
SE
O
SV
I
PR
PE
FR
N
I
U
LA
M
PO
G
R
AC
SA
AC
R
U
FR
PE
M
A
PO
TR
Q
U
Q
UE
L
0
Species
12
Total Stem Density
(B)
Number of Plots
10
8
6
4
2
0
80
0+
70
180
0
60
170
0
0
0
50
160
40
150
30
140
0
20
130
0
10
120
010
0
0
Stem Density Class(stems/ha)
Figure 1—Stem density of sampled forest structure plots in the Sherburne National
Wildlife Refuge. (A) Box and whisker plots of stem density of individual species where
present in a plot. Horizontal line in each box represents the median value. Each outliner
is depicted as an open circle. (B) Total stem density determined in individual plots.
Figure 2—Stem density by species across the landscape. The relative size of each circle
reflects the total stem density of each plot.
USDA Forest Service Proceedings RMRS-P-46CD. 2007. 510
Fire History and Age Structure Analysis in the Sherburne National Wildlife Refuge:…
Kipfmueller and Hepola
Basal Area Information
30
Basal Area (m2/ha)
(A)
n=55
20
2
Basal Area (m /ha)
25
15
10
5
PR
VI
PR
SE
SA
SP
O
SV
I
PR
PE
FR
N
I
U
LA
M
AC
SA
AC
R
U
FR
PE
A
TR
U
M
PO
Q
Q
U
EL
0
Species
18
Basal Area (m2/ha)
(B)
16
Number of Plots
14
12
10
8
6
4
2
45
40
-
40
35
-
35
30
-
30
25
-
25
20
-
20
15
-
15
10
-
0
61
0-
6
0
Basal Area Class (m2/ha)
Figure 3—Basal area of sampled forest structure plots in the Sherburne National Wildlife
Refuge. (A) Box and whisker plots of basal for individual species where present in a
plot. Horizontal line in each box represents the median value. Each outlier is depicted
as an open circle. (B) Total basal area determined in individual plots.
Figure 4—Basal area by species across the landscape. The relative size of each circle
reflects the total stem density of each plot.
USDA Forest Service Proceedings RMRS-P-46CD. 2007.
511
Fire History and Age Structure Analysis in the Sherburne National Wildlife Refuge:…
Kipfmueller and Hepola
Species Importance Values
Importance Values
(A)
n=55
200
Importance Value
175
150
125
100
75
50
25
PR
VI
PR
SE
SV
I
PR
PE
O
N
I
U
LA
M
FR
G
R
AC
SA
AC
R
U
FR
PE
PO
U
M
A
PO
TR
Q
Q
U
EL
0
Species
100
(B) Percentage of Stands Containing Each Species
Percent Occurence
90
80
70
60
50
40
30
20
10
PR
VI
PR
SE
O
SV
I
PR
PE
FR
N
I
U
LA
M
PO
G
R
AC
SA
AC
R
U
FR
PE
PO
TR
A
Q
U
M
Q
U
EL
0
Species (four letter code)
Figure 5— (A) Box and whisker plots of importance values for individual species where
present in a plot. Importance value is calculated by adding together the relative density
and basal area of a given species (B) after first multiplying each by 100. Importance
values range from 0 to 200, with 200 indicative of a stand that is composed of only one
tree species. Horizontal line in each box represents the median value.
Figure 6—Basal area by species across the landscape.
USDA Forest Service Proceedings RMRS-P-46CD. 2007. 512
Fire History and Age Structure Analysis in the Sherburne National Wildlife Refuge:…
Kipfmueller and Hepola
Age Structure Data
We are processing increment cores for determination of age-class patterns.
We hope to use the age-structure information to refine our search for firescarred trees. We will concentrate our search effort for fire scars near areas that
contain older trees to help extend the fire history further back in time.
Preliminary analyses indicate that most of the sampled trees germinated in
the mid 1900s. There appears to be a dramatic pulse in regeneration during
the early 1950s. The oldest trees dated so far have inner-ring dates in the
middle 1800s with a few that extend further into the past.
The low density area near the center of figure 7 contains trees dating to
the early 1800s (1801 is the oldest so far). Increment cores from these trees
were collected separately from the age-structure grid points for a dendrochronology class during fall 2006. Older trees also appear to be near the
outer edges of the study area.
We have not assessed the relative ages of bur oak versus pin oak in a formal
sense, but it appears the oldest trees are mostly bur oak.
Figure 7—Oldest sampled trees by species.
USDA Forest Service Proceedings RMRS-P-46CD. 2007.
513
Fire History and Age Structure Analysis in the Sherburne National Wildlife Refuge:…
Kipfmueller and Hepola
Tree Size Observations
In nearly every plot where both Quercus ellipsoidalis and Q. macrocarpa
were present, ellipsoidalis was larger in diameter on average (fig. 8). However,
initial examination of the age data suggests that Q. macrocarpa were nearly
always older.
DBH Distribution: QUEL
(A)
300
Frequency
250
200
150
100
200
150
100
50
50
0
0
0
5 -5
10 - 10
15-15
20-20
25-25
30-30
35-35
40-40
45-45
50-50
55-55
60-60
65-65
70-70
75-15
80-80
85-85
9 -90
950-9
-1 5
1000
0+
Frequency
250
DBH Class (5 cm increments)
0
5 -5
10 - 10
15-15
20-20
25-25
30-30
35-35
40-40
45-45
50-50
55-55
60-60
65-65
70-70
75-15
80-80
85-85
9 -90
950-9
-1 5
1000
0+
300
DBH Distribution: QUMA
(B)
DBH Class (5 cm increments)
Figure 8—Diameter distribution of Quercus ellipsoidalis (A) and Q. macrocarpa (B) within the study area. Diameter
at breast height is reported in 5-cm classes. Trees less than 5 cm in diameter were considered saplings and were
recorded elsewhere.
USDA Forest Service Proceedings RMRS-P-46CD. 2007. 514