Journal of
Ecology 2003
91, 822– 836
Late-glacial and Holocene climatic effects on fire and
vegetation dynamics at the prairie–forest ecotone in
south-central Minnesota
Blackwell Publishing Ltd.
PHILIP CAMILL, CHARLES E. UMBANHOWAR JR*, REBECCA TEED†,
CHRISTOPH E. GEISS‡, JESSICA ALDINGER*, LEAH DVORAK*,
JON KENNING, JACOB LIMMER and KRISTINA WALKUP*
Department of Biology, Carleton College, One North College Street, Northfield, MN 55057, USA,
*Department of Biology, St Olaf College, 1520 St Olaf Avenue, Northfield, MN 55057, USA,
†Limnological Research Center, University of Minnesota, 310 Pillsbury Dr SE, Minneapolis, MN 55455, USA, and
‡Department of Physics, Trinity College, 300 Summit Street, Hartford, CT 06106, USA
Summary
1 Treeline ecotones, such as the prairie–forest boundary, represent climatically sensitive regions where the relative abundance of vegetation types is controlled by complex
interactions between climate and local factors. Responses of vegetation and fire to
climate change may be tightly linked as a result of strong feedbacks among fuel production, vegetation structure and fire frequency/severity, but the importance of these
feedbacks for controlling the stability of this ecotone is unclear.
2 In this study, we examined the prairie–forest ecotone in south-central Minnesota
using two lake sediment cores to reconstruct independent records of climate, vegetation
and fire over the past 12 500 years. Using pollen, charcoal, sediment magnetic analyses
and LOI properties, we investigated whether fires were controlled directly by climate or
indirectly by fuel production.
3 Sediment magnetic and LOI data suggest four broad climatic periods occurring
c. 11 350–8250 BP (cool/humid), c. 8250–4250 BP (warm/dry), c. 4250–2450 BP (warm/
humid), and c. 2450–0 BP (cool/humid), indicating that, since the mid-Holocene, climate has shifted towards wetter conditions favouring greater in-lake production and
fuel production on the landscape.
4 The area surrounding both lakes was characterized by boreal forest c. 12 500–
10 000 BP, changing to an Ulmus-Ostrya forest c. 10 000–9000 BP, changing to a
community dominated by prairie (Poaceae-Ambrosia-Artemisia) and deciduous forest
taxa c. 8000–4250 BP, and finally shifting to a Quercus-dominated woodland/savanna
beginning c. 4250–3000 BP.
5 Charcoal influx increased from an average of 0.11–0.62 mm2 cm−2 year−1 during the
early Holocene forest period (c. 11 350–8250 BP) to 1.71–3.36 mm2 cm−2 year−1 during
the period of prairie expansion (c. 8250–4250 BP) and again increased to 4.18–4.90
mm2 cm−2 year−1 at the start of the woodland/savanna period (c. 4250 BP).
6 As a result of the influence of climate on community composition and fuel productivity, changes in fire severity may be the result and not the cause of shifts in
vegetation.
Key-words: climate, ecotone, fire, Holocene, prairie–forest border
Journal of Ecology (2003) 91, 822–836
© 2003 British
Ecological Society
Correspondence: Philip Camill, Department of Biology, Carleton College, One North College Street, Northfield, MN 55057,
USA (tel. 507 646–5643; fax 507 646–5757; e-mail [email protected]).
823
Climatic effects on
fire and vegetation
© 2003 British
Ecological Society,
Journal of Ecology,
91, 822–836
Introduction
Grassland /woodland boundaries represent climatically sensitive regions where the relative abundance of
vegetation types is controlled by complex interactions
among regional factors (climate) and local factors (fire,
vegetation structure, grazing, seed dispersal and land
use change) (Gleason 1913; Grimm 1983, 1984; Archer
1990; Baker et al. 1992; Peterson & Reich 2001). In
North America, this ecotone ranges in width from 50
to 100 km and consists of an often-varied mosaic of
prairie and forest stretching from Alberta to Texas
(Anderson 1982).
Climate warming of 1.4 –5.8 °C over the next century
(IPCC 2001a; Knutti et al. 2002; Stott & Kettleborough
2002) has the potential to initiate large-scale changes
in vegetation and fire regimes in semiarid regions
(Overpeck et al. 1990; Pitelka et al. 1997; Flannigan et al.
2000; IPCC 2001b; Walther et al. 2002). For instance,
moisture deficits promote strongly negative xylem
pressure potentials, which induce hydraulic cavitation
and constrain treelines (Sperry & Sullivan 1992; Bragg
et al. 1993; Sperry et al. 1994; Pockman & Sperry 2000).
Climate also controls fire frequency (Clark 1988a, 1990),
which can determine the relative abundance of prairie
and forest species (White 1983; Faber-Langendoen &
Davis 1995; Peterson & Reich 2001; Briggs et al. 2002)
and may have mediated expansion of deciduous forest
taxa into prairie in Minnesota following Little Ice Age
(LIA) cooling (Grimm 1983).
Local factors and indirect responses to climate may
make it difficult to predict the timing and direction of
responses to future climate change at grassland/woodland boundaries (Camill & Clark 2001). Fire frequency
and severity may be controlled by fuel load and type
(Johnson & Risser 1975; Streng & Harcombe 1982;
Fonteyn et al. 1984; Umbanhowar 1996; Archer 1990;
Clark et al. 2001), suggesting that climate might have
a significant indirect effect on fire by controlling community composition and ecosystem productivity. In
this case, vegetation change may directly control fire
dynamics, which, in turn, can feed back on vegetation
change (Dale et al. 2000). The invasion of tallgrass
prairie in Kansas by Quercus macrocarpa and Q. muhlenbergii seedlings is not limited by soil moisture in
most years, but rather by fire frequency, herbivory and
seed predation (Danner & Knapp 2001). Landscape
heterogeneity, such as fire breaks associated with water
bodies or slopes, may provide refuges and slow the
transition of one vegetation type to another (Grimm
1984; Clark 1990).
Predicting future dynamics at grassland/woodland
boundaries is difficult in modern systems because the
time scales of variation in climate, fire and vegetation
change often exceed the duration of instrumented records and field experiments, and most areas are highly
fragmented by agriculture. Multi-proxy palaeoecological records from lake sediments offer sufficiently long,
high-resolution records for understanding dynamics
of this ecotone (Almendinger 1990, 1992; AlmquistJacobson et al. 1992; Clark et al. 2001; Geiss et al. in
press). The glacial landscape of the upper mid-western
USA has received considerable attention from palaeoecologists studying climate, fire and vegetation at the
prairie–forest ecotone (Jelgersma 1962; Wright et al. 1963;
McAndrews 1968; Wright & Watts 1969; Amundson &
Wright 1979; Swain 1979; Grimm 1983; Webb et al.
1983; Bartlein et al. 1984; Grimm 1984; Winkler et al.
1986; Almendinger 1992; Baker et al. 1992; Wright
1992; Bartlein & Whitlock 1993; Camill & Clark 2001;
Clark et al. 2001; Baker et al. 2002).
Results from these studies indicate substantial climatic and vegetation changes in the upper mid-western
USA over the past 12 000 years. Reconstructions of climate and lake-levels indicate cool and moist conditions
from the late glacial period until approximately 8500–
7800 BP, when average summer and winter temperatures increased by 0.5–2 °C and 5–10 °C, respectively,
accompanied by a drop (c. 50–100 mm) in annual
precipitation (Bartlein et al. 1984; Winkler et al. 1986;
Bartlein & Whitlock 1993; Webb et al. 1993; Schwalb
& Dean 2002). Maximum aridity (warm and dry)
occurred between 7000 and 4500 BP, although brief
warm, wet periods were recorded for Elk Lake,
Minnesota, at 6750 – 6250 BP and again 5400–4800 BP.
Beginning at c. 4500 BP, January temperatures and
annual precipitation rose again (warm and moist)
initially to exceed modern values until 3500 BP, when
temperature and precipitation dropped essentially to
match modern values. Changes in climate had a large
impact on the geographical location of the prairie–
forest border. Baker et al. (1992, 2002) proposed two
modes of prairie expansion resulting from changes in
air mass boundaries. Sites in north-central Iowa (Clear
Lake) and central Minnesota (Kirchner Marsh) showed
maximum prairie expansion between 8000 and 5500 BP
as a result of warm, dry Pacific air, which was blocked
by tropical Caribbean air bordering north-eastern Iowa,
south-eastern Minnesota and southern Wisconsin
(Fig. 1a). Sites dominated by Caribbean air, however,
supported hardwood forest taxa in this south-eastern
region until c. 5500 BP, when the Caribbean air boundary relaxed, possibly as a result of a diminishing Laurentide ice sheet, expanding continental summer low
pressure, and shifting of westerly flow (dry Pacific air)
south-eastwards (Wright 1992), causing an attendant
rise in prairie taxa between 5500 and 3000 BP at
Roberts Creek (Iowa) and Devils Lake and Lima Bog
(Wisconsin) (Fig. 1a, Baker et al. 1992).
Despite large changes in climate and vegetation,
however, the causal links between fire frequency/severity and vegetation dynamics in response to changes in
climate are not entirely clear. Using fossil pollen from
two lake sediment cores, Grimm (1983) reconstructed
the past 5600 years of vegetation dynamics in central
Minnesota (Wolsfeld and French lakes, Fig. 1b). The
mid-19th century, pre-settlement prairie–forest border
in this region was characterized as a landscape mosaic
824
P. Camill et al.
Fig. 1 (a) Location of study region (outlined square) along the modern prairie–forest ecotone (dark line) in Minnesota in relation
to previously described sites discussed in this paper from the upper mid-western USA (see text for references): (1) Elk Lake, (2)
Williams and Shingobee Lakes, (3) Parkers Prairie sandplain, (4) Clear Lake, (5) Roberts Creek, (6) Devils Lake, and (7) Lima
Bog. Dashed line separates regions where maximum prairie expansion occurred between 8000 and 5500 BP in the north and west
from those where it occurred from 5500 to 3000 BP in the south and east (modified from Baker et al. 2002). (b) Inset shows the
location of Kimble Pond and Sharkey Lake with respect to the pre-settlement geographical ranges of the Big Woods region
(black), oak woodland (grey), and prairie (white) in south-central Minnesota (Marschner 1974; Grimm 1984) and previously
described sites discussed in this paper (see text for references): (8) French Lake, (9) Rutz Lake, (10) Stone Lake Tamarack Swamp,
(11) Pogonia Bog Pond, (12) Wolsfeld Lake, (13) Lake Carlson, and (14) Kirchner Marsh.
© 2003 British
Ecological Society,
Journal of Ecology,
91, 822–836
of tallgrass prairie, oak savannas and bigwoods, a
northern deciduous forest assemblage dominated by
Quercus, Ulmus, Acer saccharum, Fraxinus and Tilia
(Marschner 1974) (Fig. 1b). Throughout this paper,
our nomenclature is consistent with that of Grimm
(1983, 1984), who differentiates between the ‘Big Woods’
region (Fig. 1b) and ‘bigwoods’ forest taxa. His analyses showed expansion of bigwoods taxa at approximately 1600 AD, but more recent work (Umbanhowar
2003) suggests that the shift to a more forested landscape began c. 1200 AD and that the timing of bigwoods forest expansion was variable among sites.
Grimm (1983) attributed the expansion of bigwoods
taxa to a long-term decrease in fire frequency, although
he did not quantify fire in his cores to test this hypothesis. He speculated that fire frequency at this ecotone
was probably controlled by climate, specifically by
decreased precipitation : evaporation ratios, that may
have been associated with LIA cooling. Clark et al.
(2003) used pollen and charcoal records from three lakes
spanning the modern prairie–forest border in North
Dakota, Minnesota and Wisconsin, to suggest that
fire severity has been controlled by fuel production
and moisture.
Determining the relative importance of climate vs.
fuel production vs. community composition on fire
has important consequences for understanding vegetation dynamics. For example, if fire severity (biomass
consumed) and frequency correlate with aridity, prairies should expand at the expense of forest because of
recruitment limitations imposed by fire and aridity.
Alternatively, because of the flammability of fine fuels,
fire severity might rise with the shift from forests to
prairie, which is an indirect effect of climate on fire
mediated by changes in community composition. Furthermore, if humid conditions lead to higher productivity and greater fire severity, this might sharpen the
grassland/woodland boundary. The fuel production
hypothesis assumes a positive correlation between net
primary production (NPP) and precipitation, which
has been well documented for prairie regions in North
America (Briggs & Knapp 1995; Burke et al. 1998;
Paruelo et al. 1999; Fay et al. 2000; Lane et al. 2000;
Knapp et al. 2001). However, humid periods could
also favour increased recruitment and expansion of
forest taxa, which, because of lower litter flammability
(Streng & Harcombe 1982; Fonteyn et al. 1984), could
decrease fire frequency/severity and lead to greater
forest recruitment (Grimm 1983). These hypotheses
can be tested directly by examining relative changes in
the timing and magnitude of climate change, community composition and fire severity. Unfortunately, most
palaeoecological studies to date have been limited by
the lack of climate or fire records required for fully
characterizing these kinds of climate–fire–vegetation
interactions.
825
Climatic effects on
fire and vegetation
In this study, we examined the prairie–forest ecotone
in south-central Minnesota using two lake sediment
cores to reconstruct independent records of climate,
vegetation and fire over the past 12 500 years. We took
advantage of a late glacial and Holocene climate
reconstruction based on multiple sediment magnetic
properties (Geiss et al. 2003). We used pollen analysis
to describe landscape-scale vegetation change during
periods of climate change. Macroscopic charcoal analysis enabled us to test if fire severity changed significantly across broad climatic changes and whether fire
severity was controlled mostly by aridity during xeric
periods, fuel production and moisture during humid
periods (sensu Clark et al. 1996, 2001), or changes in
community composition. Finally, we examined the
stability of this ecotone, in terms of community composition, throughout the Holocene and asked whether
stability is affected by fire severity.
Study region
© 2003 British
Ecological Society,
Journal of Ecology,
91, 822–836
We analysed sediment from two lakes located on the
west and east sides of the prairie–forest border in
south-central Minnesota near the towns of Mankato
and New Market (Fig. 1b). Kimble Pond (44°13′15″ N,
93°50′24″ W) and Sharkey Lake (44°35′39″ N, 93°24′49″ W)
are kettle lakes that formed in the Altamont stagnation
moraine of the Des Moines lobe during the last phase
of the Wisconsin glaciation (Hobbs & Goebel 1982).
Kimble Pond has an area of 4.9 ha with a maximum
depth of 16 m. Vegetation surrounding the lake consists
of a mosaic of oak woodland (Quercus macrophylla)
and bigwoods taxa (Tilia americana, Ulmus americana,
Acer saccharum and Fraxinus pennsylvanica) within an
agricultural landscape (Zea mays and Glycine max).
Sharkey Lake has an area of 28 ha and consists of two
basins with a maximum depth of 15 m in the south-east
basin and a maximum depth of c. 1–2 m in the northern
basin. This lake also lies within an agricultural landscape, but oak woodland is more prominent relative to
bigwoods vegetation at this site (Fig. 1B).
Lithology for these sites has been described previously (Geiss et al. 2003), and only a brief description is
provided here. Sediment at Kimble Pond consisted
of late-Pleistocene to early Holocene (> 9500 BP)
black-laminated diatomaceous ooze, followed by
black-laminated to partially laminated diatomaceous silt deposited during the early to mid-Holocene
(9500 – 4000 BP). Late Holocene (< 4000 BP) sediment
consisted of fibrous gyttja and black-banded diatomaceous silt and clay. At Sharkey Lake, early sediment
(> 8500 BP) consisted of grey, partially laminated silt
with diatoms and very dark grey, massive diatomaceous ooze. Early Holocene (11 000–7500 BP) sediment
consisted of very dark grey, finely laminated diatomaceous ooze, changing to partially laminated diatomaceous ooze and silt (7500 ka−3500 BP). Late Holocene
(< 3500 BP) sediment consisted of massive black silt
with diatoms.
Methods
   
Multiple overlapping sediment cores (18.53 m from
Kimble Pond and 15.66 m from Sharkey Lake) were
obtained from the deepest point of each basin in
January of 2000, 2001 and 2002 using a clear polycarbonate piston corer to sample flocculent sediments at
the surface and a modified Livingston corer to collect
sediment at depth (Wright et al. 1984). Cores were split
and stored at 4 °C until analysed. Age control for Kimble and Sharkey is based on nine accelerator mass spectrometry (AMS) 14C dates for each lake obtained from
wood macrofossils, seeds, leaves and charcoal (Table 1).
The post-settlement rise in Ambrosia pollen at c. 90 BP
(AD 1860) was also used as a chronological marker.
Radiocarbon dates were calibrated using CALIB vs.
4.3 (Stuiver et al. 1998; Stuiver et al. 1999), and all dates
reported in this paper are expressed in calendar years
BP. To assemble independent records of climate, fire
and vegetation, we analysed magnetic properties, loss
on ignition, fossil pollen and macroscopic charcoal.
Unless otherwise noted, samples for these analyses were
taken roughly every 10 cm, which resulted in a mean
resolution between successive samples of 75 ± 5 years
in Kimble and 85 ± 9 years in Sharkey sediments.
 
We measured three magnetic properties useful for reconstructing Holocene climate change at these sites
(Banerjee et al. 1981; Thompson & Oldfield 1986; Geiss
& Banergee 1999; Geiss et al. 2003). Low-field magnetic susceptibility was measured using a Kappabridge,
model CLY-2 (Geofyzika, Brno, Czech Republic). Isothermal remanent magnetization (IRM) was acquired
in a magnetic field of 100 mT and used to estimate the
concentration of magnetic minerals. Anhysteretic
magnetic remanence (ARM) was acquired in a peak
alternating field of 100 mT and a bias field of 50 mT. It
was imprinted using a D-2000 alternating field demagnetizer (D-Tech Inc., Jamestow RI, USA), and all remanence parametres were measured with a model 760-R
cryogenic magnetometre (2G Enterprises, Mountain
View, CA, USA). Magnetic susceptibility and IRM
were used to estimate the abundance of magnetic minerals. The concentration-independent ratio of ARM/
IRM was used to characterize the relative abundance
of fine-grained (i.e. single domain, SD) magnetic particles with respect to coarser (pseudo-single-domain,
PSD, and multidomain, MD) magnetic grains. Low
ratios of ARM/IRM indicate a significant fraction
of coarse-grained particles, and vice versa (Özdemir &
Banerjee 1982; Geiss & Banergee 1999; Geiss et al.
2003). For these analyses, cores were sampled contiguously at 1–2 cm spacing (c. 5- to 15-year resolution).
We used loss on ignition (LOI) to measure organic
and inorganic sediment fractions (Dean 1974). The
826
P. Camill et al.
Table 1
14
C and calibrated dates for Kimble Pond and Sharkey Lake
Laboratory sample*
Depth below ice (cm)
Kimble Pond
Pollen horizon
CAMS-69887
CAMS-69888
CAMS-90104
CAMS-79286
CAMS-79285
CAMS-65374
CAMS-65373
CAMS-65365
CAMS-65364
1905.5
1974.5
2154.5
2174.0
2334.4
2600.5
2856.0
2966.5
3139.0
3405.0
Sharkey Lake
Pollen horizon
CAMS-83724
CAMS-77858
CAMS-77860
CAMS-90103
CAMS-77859
CAMS-90100
CAMS-90101
CAMS-90102
CAMS-83725
1569.0
1631.0
2001.3
2365.0
2730.5
2865.5
2935.5
2971.0
2989.5
2994.0
14
C year BP
Calibrated year BP†
Material dated
1040 ± 50
2150 ± 80
1995 ± 35
2270 ± 60
3890 ± 50
5590 ± 50
6620 ± 50
7910 ± 50
10570 ± 50
955 (828 –1060) P = 0.97
2142 (1986 –2335) P = 0.97
1942 (1868 –2007) P = 0.97
2245 (2117–2360) P = 0.99
4318 (4151–4422) P = 0.99
6366 (6289 – 6451) P = 0.98
7503 (7429 –7572) P = 1.00
8753 (8595 – 8906) P = 1.00
12622 (12319 –12924) P = 0.98
Ambrosia rise
Charcoal
Charcoal
Charcoal
Charcoal
Charcoal
Wood
Wood
Wood
Picea wood
1530 ± 50
3630 ± 40
4730 ± 60
6528 ± 35
6840 ± 50
7995 ± 40
9430 ± 45
9510 ± 130
10255 ± 45
1420 (1327–1524) P = 0.98
3940 (3833 – 4084) P = 1.00
5470 (5320 –5589) P = 1.00
7434 (7330 –7506) P = 0.96
7668 (7586 –7755) P = 0.98
8870 (8715 –9009) P = 0.97
10655 (10546 –10999) P = 0.95
10831 (10483 –11175) P = 0.99
12040 (11734 –12358) P = 0.98
Ambrosia rise
Charcoal
Grass spikelets
Charcoal
Charcoal
Charcoal
Deciduous leaf
Birch seed
Picea needle
Charcoal,
Picea needle
*Center for Accelerator Mass Spectrometry, Lawrence Livermore National Laboratory, Livermore, California, USA.
†Calibrated age, age range (in parenthesis) and probability that calibrated date falls within age range (Stuiver et al. 1999).
organic fraction was determined by the fraction of dry
mass loss at 550 °C, CaCO3 was determined based on
mass loss at 1000 °C, and the residual inorganic fraction was estimated as the fraction of dry mass remaining following combustion. Percentage of biogenic silica
from diatoms was measured to further describe the
residual inorganic sediment fraction. Silica was extracted
from 30-mg samples of freeze-dried sediment using a
1% solution of Na3O2 (Conley & Schelske 1993), and
the concentrations were measured colorimetrically
(McKnight 2000) using autoanalysis (Lachat Instruments, Milwaukee, WI, USA). Estimates of percentage
of silica were based on an average of 3, 4 and 5-h extracts.
 
Pollen analysis was conducted on 0.5-cm3 samples
using standard palynological methods (Fægri et al.
1989; Shane 1992) modified to include an additional
sieving step with 8-µm Nitex to remove detritus. Between
302 and 1120 endogenous grains were counted per slide
(mean = 471), and an exogenous spike (Eucalyptus
globula) was used to calculate pollen concentration and
influx. We report changes in pollen percentages for main
taxa and summary classes only; full pollen diagrams
for Kimble Pond and Sharkey Lake can be retrieved
from the North American Pollen Database (2003).
© 2003 British
Ecological Society,
Journal of Ecology,
91, 822–836
 
Macroscopic charcoal analysis was conducted on 1.0cm3 samples following the methods of Clark & Hussey
(1996). Sediment samples were soaked in a 10% KOH
for 24 h and sieved with a 180-µm screen. Sieved material was spread over the bottom of a gridded Petri plate,
and charred particles were identified at × 20 magnification using a stereoscope. Digital images of the charcoal
were made, and charcoal area and shape were measured from these images using SCION image analysis
software. Area measurements are reported as concentrations (mm2 cm−3) based on the volume of sediment
analysed. Charcoal influx rate (mm2 cm−2 year−1) was
calculated by multiplying concentration and sedimentation rate. Charcoal concentration and influx values
were also smoothed with a 200-year averaging filter to
estimate background levels that represent regional
source areas (Clark 1988b; Clark & Royall 1994, 1995,
1996; Clark & Hussey 1996; Clark et al. 1998). Throughout this paper we use charcoal concentration and influx
as measures of landscape-level fire severity, a proxy for
biomass consumed.
Theory of charcoal particle transport (Clark 1988b;
Clark et al. 1996; Clark & Patterson 1997) combined
with measurements of charcoal transport in experimental burns (Clark et al. 1998) indicate that 180-µm
sieved charcoal records local fires within tens to hundreds of metres from lake edges. For small lakes, charcoal may provide a more local record of fire (Whitlock
et al. 1997) compared with vegetation with pollen
source areas of tens to hundreds of metres (Prentice
1985; Jackson & Lyford 1999). We are confident that
our records of charcoal are capable of detecting the
influences of changing climate, however, because local
vegetation is likely to be affected by changes in climate,
827
Climatic effects on
fire and vegetation
and multiple lakes allow us to identify regional trends
in fire.
We could not estimate fire frequency or return interval with data analysed in this study. Identifying unique
fire events is difficult in prairie where background charcoal levels are high. Moreover, sample resolution in this
study (74 – 85 years) was too coarse to assess changes in
fire frequency characteristic of this ecotone. Finally,
fire return interval is sensitive to sedimentation rates,
which varied with depth. Therefore, our analysis of fire
is limited to severity and not frequency.
 
To determine if charcoal concentration and influx
changed with climate, we developed a robust, nonparametric procedure using S-Plus (Insightful Corp.,
Seattle, WA, USA) based on bootstrap resampling
(Efron & Tibshirani 1993) to test whether mean charcoal concentrations and influx from different climatic
periods deviated from a null hypothesized mean difference of zero (Ho: µ2 − µ1 = 0; Ha: µ2 − µ1 ≠ 0). Classical means comparisons, such as  contrasts,
cannot be performed on time-series data because
serial autocorrelation violates parametric assumptions of independence. For each lake, we calculated the
mean charcoal concentration and influx for each climatic period identified with the sediment magnetic
analysis, and we determined the difference in mean
values using pairwise comparisons among all climatic
periods. Charcoal data in each climatic period were
resampled with replacement, and the above procedure
was repeated 1000 times to bootstrap 95% confidence
intervals on µ2 − µ1. Climatic periods with significant
mean differences in charcoal concentration and influx
were identified by confidence intervals that did not
bound zero.
Results
 
© 2003 British
Ecological Society,
Journal of Ecology,
91, 822–836
The age-depth relationship was best fit with a third
order polynomial for Kimble Pond (r2 = 0.995, Fig. 2a).
A minor reversal (< 155 years) occurred between 21.54
and 21.74 m, suggesting that the charcoal in sample
CAMS 69888 might have contained trace amounts of
older, reworked charcoal or charcoal transported to the
lake later than the fire event (Whitlock & Millspaugh
1995). Sedimentation rates at the bottom 3 m of
Sharkey Lake were slow compared with the upper 12 m
(Fig. 2b). Instead of linearly interpolating sedimentation rates, which resulted in sharp, local changes in
sedimentation rate throughout this core, we fitted two
separate second-order polynomial functions to the data
(Fig. 2b, r2 = 0.99, r2 = 0.95). The 18.53-m sediment
record from Kimble Pond represented continuous
deposition over the past c. 12 500 years BP (Fig. 2), with
sedimentation rates ranging from 0.067 to 2.37 cm year−1
Fig. 2 Calibrated age-depth models for (a) Kimble Pond and
(b) Sharkey Lake.
(0.42–15.0 year cm−1). The 15.66 m of sediment in
Sharkey Lake also showed continuous deposition over
the past c. 13 000 years BP but with lower sedimentation rates ranging from 0.015 to 0.37 cm year−1 (2.3–
68 year cm−1).
  ,  
,  
Four distinct climatic periods were identified based on
changes in the concentration and particle-size of magnetic minerals present in the sediment, which can be
used as a proxy for biogenic productivity and the input
of terrigenous minerals (derived from surrounding
landscapes) (Geiss & Banerjee 1999; Geiss et al. 2003).
During warm and humid periods the watershed of
the studied lakes was covered by forest vegetation, erosion of terrigenous material into the lake was low and
the lake sediment consisted mainly of organic matter
and authigenic magnetic minerals, such as magnetite.
These magnetite particles are very small (SD) grains
(d < 0.1 µm) and occur in low concentrations. The
resulting magnetic signature is defined by low values of
concentration-dependent magnetic parameters (magnetic susceptibility, ARM and IRM) and high ratios of
ARM/IRM, a concentration-independent parameter,
highly sensitive to small (SD) grains. During dry periods,
erosion and dust deposition increased and lake sediment contained a high fraction of large (MD) grains (d >
10 µm). This increased input of MD grains is reflected
in high values of all concentration-dependent parameters and a drop in ARM/IRM. The model is discussed
in detail by Geiss et al. (2003).
At Kimble Pond low values of IRM and high ratios
of ARM/IRM between 11 700 and 8500 BP and 4500
and 2800 BP indicate that the sediment is characterized
828
P. Camill et al.
Fig. 3 Profiles of sediment magnetics, sediment properties and charcoal for Kimble Pond and Sharkey Lake. Isothermal remanent
magnetization (IRM) and anhysteretic magnetic remanence (ARM) (a, b) are measures of the concentration of magnetic particles
(increases with IRM) and particle size (decreases as ARM/IRM ratio increases). Sediment properties include organic fraction (c,
d), CaCO3 fraction (e, f ), residual inorganic fraction (g, h), and fraction of biogenic silica in residual inorganic material (i, j). Fire
severity measures include charcoal concentration (mm 2 cm−3) with 200-year averaging filter (k, l), and charcoal influx (mm 2
cm−2 year−1) with 200-year averaging filter (m, n). Climatic zones derived from magnetics analyses are delimited by vertical lines.
© 2003 British
Ecological Society,
Journal of Ecology,
91, 822–836
by fine-grained, single-domain, authigenic magnetic
minerals (Fig. 3a). These sediments resulted from low
rates of erosion and imply a relatively humid climate.
Mid-Holocene aridity and the associated increase in
erosion rates and /or eolian dust deposition result in an
up to fivefold increase in IRM and a drop in the ARM/
IRM ratio between 8500 and 4500 BP. During this
period sediment magnetic properties reflect the input
of coarse-grained, multidomain magnetic minerals
typical of the surrounding soils and glacial deposits.
Values for ARM/IRM and IRM from 2800 to 0 BP
suggest intermediate climatic conditions.
At Sharkey Lake, shifts in magnetic properties
occurred c. 500–700 years later than in Kimble Pond
(Fig. 3b), possibly reflecting later changes in climate at
the eastern edge of the prairie–forest border, uncertainties in the age-depth models (Fig. 2), or differences in
timing between changing vegetation and magnetic
properties. The early Holocene climate signal is less
resolved due to low sedimentation rates, but ARM/
IRM ratios were high during 11 000–8000 BP, which
implies a relatively humid climate. The mid-Holocene
period of maximum aridity is characterized by a pronounced increase in IRM and lower ratios of ARM/
Table
829 2 Results of non-parametric statistical test for differences in mean charcoal concentration for climatic periods identified in each core
Climatic effects on
fire and vegetation Mean charcoal
Comparison
Mean charcoal
concentration difference
(mm2 cm−3)
95% confidence
interval†
Significance
65
28
48
22
0 –2800 vs. 2800 – 4500
0 –2800 vs. 4500 – 8500
0 –2800 vs. 8500 –11700
2800 – 4500 vs. 4500 –8500
2800 – 4500 vs. 8500 –11500
4500 – 8500 vs. 8500 –11700
23.27
7.16
1.37
16.22
21.90
5.69
(18.24 –29.40)
(4.18 –10.01)
(−1.61– 4.88)
(10.95 –22.56)
(16.50 –28.05)
(1.78 –9.06)
Yes
Yes
No
Yes
Yes
Yes
26
24
85
10
0 –2100 vs. 2500 – 4000
0 –2100 vs. 4000 – 8000
0 –2100 vs. 8000 –11000
2100 – 4000 vs. 4000 –8000
2100 – 4000 vs. 8000 –11000
4000 – 8000 vs. 8000 –11000
16.31
0.78
12.23
17.09
28.54
11.44
(9.95 –23.74)
(−2.64 –3.82)
(9.01–15.46)
(10.90 –23.74)
(22.49 –35.06)
(8.97–13.79)
Yes
No
Yes
Yes
Yes
Yes
Climatic period*
concentration
(mm2 cm−3)
n
Kimble Pond
0 –2800 BP
2800–4500 BP
4500–8500 BP
8500–11700 BP
6.48
29.76
13.54
7.85
Sharkey Lake
0 –2100 BP
2100–4000 BP
4000–8000 BP
8000–11000 BP
15.08
31.39
14.30
2.85
*Derived from ARM/IRM ratios.
†Estimated using 1000 bootstrap simulations. Mean differences in charcoal concentration are tested against the null model Ho: µ2 − µ1 = 0.
© 2003 British
Ecological Society,
Journal of Ecology,
91, 822–836
IRM between 8000 and 4000 BP. The most recent 4000
years are characterized by two periods of increased
ARM/IRM ratios (4000 –2100 BP and 2100–0 BP),
suggesting possible humid conditions beginning around
4000 BP (Fig. 3b).
The probable climatic progression from cool/humid
(c. 11 350 – 8250 BP), warm /dry (c. 8250–4250 BP),
warm/ humid (c. 4250 –2450 BP), and cool/humid (c.
2450 – 0 BP) has also been documented for other lakes
in Minnesota (Elk, Williams and Shingobee, Parkers
Prairie sandplain, Fig. 1a) based on pollen-derived
reconstructions of temperature and precipitation
(Bartlein et al. 1984; Bartlein & Whitlock 1993; Webb
et al. 1993), lake levels (Digerfeldt et al. 1992; Webb
et al. 1993), and δ18O and δ13C isotopic fractionation in
ostracode valves (Schwalb et al. 1995; Schwalb & Dean
2002).
Loss-on-ignition data support our climatic interpretation. At both lakes, the organic fraction correlated with the ARM / IRM ratios (r2 = 0.43 for Kimble,
r 2 = 0.50 for Sharkey, P < 0.0001 for both lakes), suggesting greater autochthonous carbon production
during humid conditions (Fig. 3c,d) when lake levels
were higher (Digerfeldt et al. 1992; Webb et al. 1993).
The organic fraction was greater in Sharkey Lake
compared with Kimble Pond over the most recent
3000 years. Precipitation of CaCO3 tended to be highest during arid conditions in the mid-Holocene when
ARM/IRM ratios were low, but this pattern was weak
(Fig. 3e,f ). The remaining inorganic fraction fluctuated inversely with ARM/IRM, suggesting the bulk
of mineral deposition occurred during the relatively
xeric period c. 8250 – 4250 BP (Fig. 3g,h). The inorganic
fraction was also high over the past 3000 years. Relatively small changes in biogenic silica fraction from
about 8500 – 8000 BP to the present imply that inorganic
mineral fluctuations were largely terrestrial in origin
(Fig. 3i,j).
  
Charcoal concentration in both lakes showed a statistically significant peak between approximately 4000
and 3000 BP (Fig. 3k,l, P < 0.05), and it changed significantly across climatic periods (Table 2). For Kimble
Pond, tests for differences in mean charcoal concentration were significant in all but one comparison (Table 2,
P < 0.05), which represented two humid periods as
inferred from large values of ARM/ IRM ratio (Fig. 3a).
Sharkey Lake showed a similar trend in mean concentrations, but the single lack of significant difference
occurred for a different comparison (Table 2, Fig. 3l,
P < 0.05), and, unlike Kimble Pond, there was a significant difference in mean charcoal concentration
between recent (2100–0 BP) and early Holocene sediments (11 000–8000 BP) (P < 0.05). Similar trends in
charcoal at both lakes indicate that source areas were
large enough to detect regional changes in fire associated with climatic changes.
Charcoal influx indicated that fire severity was
greatest during the warm/wet period c. 4250–2450 BP
and lowest in the cool/wet early Holocene (c. 11 350–
8250 BP) for both lakes (Table 3). Influx was statistically greatest c. 4500–2800 BP at Kimble Pond (P <
0.05) compared with the other three climatic periods
(Table 3). Sharkey Lake charcoal influx rose rapidly at
8000 BP and remained high during the warm/dry mid
Holocene (8000–4000 BP, Fig. 3n). Thus, the period
of maximum mean charcoal influx at Sharkey Lake
(4000–2100 BP) was marginally significantly greater
(P = 0.06) compared with mean influx during the midHolocene (Fig. 3n, Table 3). In contrast, mean influx
830
P. Camill et al.
Table 3 Results of non-parametric statistical test for differences in mean charcoal influx for climatic periods identified in
each core
Mean charcoal
influx
Climatic period* (mm2 cm−3 year−1) n
Comparison
Mean charcoal
influx difference
95% confidence
(mm2 cm−3 year−1) interval†
Significance
Kimble Pond
0 –2800 BP
2800 – 4500 BP
4500 – 8500 BP
8500 –11700 BP
1.80
4.90
1.71
0.62
65
0 –2800 vs. 2800 – 4500
28
0 –2800 vs. 4500 –8500
48
0 –2800 vs. 8500 –11700
22 2800 – 4500 vs. 4500 – 8500
2800 – 4500 vs. 8500 –11700
4500 – 8500 vs. 8500 –11700
3.10
0.09
1.18
3.19
4.28
1.09
(2.23 – 4.10)
(−0.49 – 0.67)
(0.69 –1.72)
(2.43 – 4.18)
(3.56 –5.21)
(0.66 –1.50)
Yes
No
Yes
Yes
Yes
Yes
Sharkey Lake
0 –2100 BP
2100 – 4000 BP
4000 – 8000 BP
8000 –11000 BP
1.64
4.18
3.36
0.11
26
0 –2100 vs. 2500 – 4000
24
0 –2100 vs. 4000 – 8000
85
0 –2100 vs. 8000 –11000
10 2100 – 4000 vs. 4000–8000
2.54
1.53
1.53
0.82
(1.64 –3.55)
(1.02 –2.11)
(1.22–1.83)
(−0.01–1.72)
2100 – 4000 vs. 8000 –11000 4.07
4000 – 8000 vs. 8000 –11000 3.25
(3.24 –5.06)
(2.80 –3.70)
Yes
Yes
Yes
No (α = 0.05),
Yes (α = 0.06)
Yes
Yes
*Derived from ARM/IRM ratios.
†Estimated using 1000 bootstrap simulations. Mean differences in charcoal concentration are tested against the null model Ho:
µ2 − µ1 = 0.
was statistically lowest in the early Holocene for both
lakes (P < 0.05, Table 3). Influx at Kimble Pond was
not significantly different between the 2800–0 BP and
8500 – 4500 BP comparison (Tables 3, P > 0.05).
 
© 2003 British
Ecological Society,
Journal of Ecology,
91, 822–836
Pollen percentages for major taxa and summary classes
showed large changes corresponding to climatic changes
derived from magnetics analyses (Fig. 4). At Kimble
Pond, the shift from boreal taxa following deglaciation
to a deciduous forest dominated by Ulmus, Fraxinus,
Ostrya/Carpinus, Carya and Quercus occurred between
c. 11 000 and 9000 BP. During this period, both charcoal concentration and influx were low (Fig. 3k, m).
It was followed by a rise in herbaceous prairie taxa
beginning c. 9000 – 8000 BP that peaked for most
taxa c. 7000–6000 BP during the mid-Holocene climate
warming (Fig. 4). Pollen percentages ranging from 1 to
5% for taxa such as Acer, Fraxinus, Carya, Tilia and
Ulmus indicated the continued presence of these taxa
within the landscape. Shrubs, seedless vascular, and
aquatic taxa constitute relatively minor fractions of
total pollen sum. Variance in pollen percentages for
each taxon was, in general, highest for prairie taxa,
shrubs and aquatic taxa during the warm/dry period
between 8500 and 4500 BP, which could have been
associated with increased seasonality and possibly
greater variability in precipitation during the midHolocene (Fig. 4, Kutzbach & Webb 1993). Herbaceous taxa pollen percentages decreased between
4500 and 3000 BP, with deciduous tree types, including
Tilia, Ostrya/Carpinus, Carya, and especially Quercus,
increasing in abundance during the same period. Fire
severity was at its highest levels in the Holocene during
this transition (Fig. 3k,m). The most recent 3000 years
were characterized by expansion of Quercus and Pinus
and declines in most deciduous tree taxa. Bigwoods
taxa, including Ulmus, Tilia and Ostrya increased
prior to settlement, while Quercus declined. The recent
expansion of herbaceous taxa coincided with settlement, land clearing and agriculture.
Vegetation changes at Sharkey Lake showed similar
correlations with climate, but some important differences with Kimble Pond existed. The early Holocene
shift from boreal to deciduous tree taxa occurred
between 11 000 and 8000 BP (Fig. 4). Fire severity was
very low during this transition (Fig. 3l,n). Unlike
Kimble Pond, the peak in herbaceous prairie taxa occurred earlier and more rapidly, beginning at c. 8000 BP,
and these taxa showed less overall variability in pollen
percentages than in Kimble sediments. Grasses also
achieved greater and more consistent dominance at
Sharkey Lake: 20–30% of the total pollen sum compared with 10–20% at Kimble Pond (Fig. 4). The timing of the shift from forest to prairie corresponded
with the rapid influx of mineral material (Fig. 3b) and
increased charcoal influx (Fig. 3n). Like Kimble Pond,
pollen for bigwoods forest taxa persisted at 1–10%,
indicating their continued presence within the landscape. Herbaceous taxa were dominant throughout
the mid-Holocene, and of particular note is the rise
in Typha pollen percentages, indicating its increased
dominance in the shallow northern basin of the lake.
Typha and herbaceous taxa declined significantly after
c. 4000 BP, coincident with declining mineral input,
a further increase in fire, and greater organic influx
(Fig. 3b,d,l,n, Table 3). Some deciduous tree species,
including Tilia, Ostrya/Carpinus and Quercus, expanded
in abundance around this time (Fig. 4). The most
831
Climatic effects on
fire and vegetation
Fig. 4 Pollen profiles for major taxa and summary classes counted in Kimble Pond and Sharkey Lake sediment. Data are expressed as percentage of total
pollen sum and are shown with magnetics and charcoal influx data. Climatic zones derived from magnetics analyses are delimited by horizontal lines.
© 2003 British
Ecological Society,
Journal of Ecology,
91, 822–836
recent 2000 years were characterized by a shift to an
oak-dominated woodland with minor presence of
bigwoods and herbaceous species (Fig. 4). This shift
coincided with declining mineral influx, increased
relative abundance of fine particles (ARM/ IRM),
decreased fire severity, and high organic influx.
Pollen concentrations in both lakes were highest
before 8000 BP because of slower sedimentation rates
(Fig. 5a,b). Pollen influx was more variable at Kimble
Pond, with increased rates between 4000 and 2000 BP,
and in post-settlement sediment (Fig. 5c). Pollen influx
peaked around 7600 BP at Sharkey Lake (Fig. 5d).
Pollen records from Kimble Pond and Sharkey Lake,
compared with other lakes from the upper mid-west,
further clarify the timing and geographical location of
the prairie–forest border in this region. Six pollen records
from south-central Minnesota (Fig. 1b), including
Kirchner Marsh and Lake Carlson (Wright et al. 1963),
Rutz Lake (Waddington 1969), and Pogonia Bog Pond
and Stone Lake Tamarack Swamp (Swain 1979;
Brugam & Swain 2000), show similar patterns in vegeta-
tion during the Holocene. Tree pollen types declined
and prairie taxa increased about 8000 BP, peaked
about 6000 BP, and declined after 4000 BP (as late as
3000 BP at Rutz Lake). Prairie expansion has been
described as far south as Clear Lake in north-central
Iowa between 8000 and 5500 BP (Fig. 1a, Baker et al.
1992), indicating broadly coherent vegetation changes
at the prairie–forest border in this region. There were,
however, some interesting geographical differences in
the timing and magnitude of pollen percentage changes
among southern Minnesota and northern Iowa lakes
during the late glacial and early Holocene. Figures 1
and 4 show that Picea pollen percentages declined
earliest (c. 11 700–11 000 BP) in the southernmost
lakes (Kimble and Clear) compared with more northern lakes (Carlson, Sharkey, Kirchner, Pogonia, Rutz,
Stone and Wolsfeld), where Picea pollen declined
between 10 500 and 9500 BP (Wright et al. 1963;
Waddington 1969; Swain 1979; Grimm 1983; Baker
et al. 1992; Brugam & Swain 2000). The oldest
deciduous-forest assemblages at the southernmost
832
P. Camill et al.
Fig. 5 Pollen concentration (grains 103 cm−3) (a, b) and pollen influx (grains 103 cm−2 year−1) (c, d) for Kimble Pond and Sharkey
Lake. Climatic zones derived from magnetics analyses are delimited by vertical lines.
lakes (Sharkey, Kimble, Clear) contained much higher
percentages of Ulmus pollen (40 – 60%) than did those
of the other records, which generally contained 20 – 25%
Ulmus, or as low as 10 –15% in the case of Rutz and
Pogonia (Figs 1 and 4).
Discussion
© 2003 British
Ecological Society,
Journal of Ecology,
91, 822–836
Our results support a growing body of evidence documenting regional differences in the dynamics of the
prairie–forest border in the upper mid-western USA,
during the Holocene (Webb et al. 1983; Baker et al.
1992, 2002), and the complexity of long-term climate–
fire–vegetation interactions at this ecotone (Camill &
Clark 2001; Clark et al. 2001). The Kimble Pond and
Sharkey Lake records clearly support Baker et al.’s
(1992, 2002) model, which indicates that most of south
and central Minnesota and north-central Iowa were
characterized by prairie expansion between 8000 and
5500 BP. As described previously (Webb et al. 1983;
Baker et al. 1992; Wright 1992), prairie expansion
occurred much later to the south-east – in northeastern Iowa (Roberts Creek) and southern Wisconsin
(Devil’s Lake, Lima Bog) – which recorded the existence of deciduous forest until c. 5500 BP and subsequent expansion of the prairie–forest ecotone between
5500 and 3000 BP, possibly as a result of increasing
influence of warm, dry Pacific air and decreasing Caribbean air at that time (Baker et al. 1992; Wright 1992).
However, the overall floristic composition of the
prairie–forest border in south-central Minnesota has
been relatively stable throughout the Holocene despite
significant changes in climate and fire. Taxa comprising
the modern vegetation assemblage were established in
this region by 9000 BP (Fig. 4), and while the relative
abundance of woody and herbaceous taxa changed
substantially with climate, reflecting shifts in the relative importance of grassland and forest within the Big
Woods landscape (Jacobson & Grimm 1986), at no
time during the Holocene did any of the modern taxa
become regionally displaced. These observations support Grimm’s (1983; 1984) hypothesis that landscape
heterogeneity – in the form of water bodies and topographic relief – provided local refuges for bigwoods
taxa, especially during humid periods when fire severity was greatest (Fig. 3k–n). In this region, c. 20% of
the core area of the Big Woods landscape (including
Kimble and Sharkey) is wetland or lake compared
with 5% further west in the adjacent prairie areas of
the Big Woods region (C.E. Umbanhowar Jr, unpublished data), where vegetation composition changes
in response to climate are more pronounced (Clark
et al. 2001). Because landscape structure may impart
significant inertia to the spatial movement of this
ecotone, large changes in vegetation and fire may not
be a universal response across grassland–woodland
boundaries.
Our results are consistent with the hypothesis that
climate controls vegetation composition and productivity directly, and both of these factors, in turn,
control fire severity (Clark et al. 1996; Clark et al. 2001).
Shifts in community composition were important
for changes in fire severity. In both lakes, the rise in
charcoal influx followed the shift from deciduous forest
to peak dominance of grasses (Poaceae) and prairie forbs,
which happened around 7000– 6000 BP at Kimble
Pond and 8000 BP at Sharkey Lake (Figs 3m,n and 4).
If aridity were the primary factor controlling fire severity, charcoal influx would have risen simultaneously at
both lakes c. 8000 BP, coincident with increasing aridity (Fig. 3a,b). Furthermore, the period of maximum
fire severity (Fig. 3k,l, Table 2) did not occur during
the mid-Holocene (8250–4250 BP) increase in warming and aridity (Fig. 3a,b, Tables 2 and 3). Instead, fire
severity increased significantly after 4000  , with a
shift to more humid (and presumably more productive)
conditions, decline in the relative abundance of herbaceous taxa, and expansion of bigwoods taxa (Table 2,
Figs 3k–m and 4). The effect of community composition on fire is further illustrated by significantly lower
fire severity during the early Holocene when the
landscape was dominated by deciduous forest taxa
(Tables 2 and 3, Fig. 4). In this case, possible increased
productivity during this humid period (Fig. 3a,b)
833
Climatic effects on
fire and vegetation
© 2003 British
Ecological Society,
Journal of Ecology,
91, 822–836
would have had little consequence on fire severity if the
landscape were predominantly deciduous forest with
lower litter flammability (Streng & Harcombe 1982;
Fonteyn et al. 1984). Collectively, these results imply
that the shift from forest to prairie, and not aridity,
increased fire severity initially in the mid-Holocene,
followed by a further increase in fire severity caused by
humid conditions and greater fuel production after
4000 BP when prairie taxa were still major components
of the landscape, though not as dominant as from
c. 8000 – 4000 BP. It appears that fire severity is not a
mechanism by which the prairie–woodland boundary
is stabilized because deciduous forest taxa, especially
Quercus, expanded during the period of greatest landscape burning over the last 12 500 years (Figs 3k,l
and 4). Moreover, increased fire severity during humid
periods did not feed back on vegetation composition
because deciduous forest taxa abundance increased
rather than the expected increase in herbaceous prairie
taxa abundance.
The simultaneous increase in fire severity and expansion of forest species into prairie after 4000 BP has
not been described previously and, initially, appears
contradictory. However, both events can occur in a
heterogeneous landscape during the probable climatic progression from warm /dry (c. 8250 – 4250 BP)
to warm/ humid (c. 4250 – 2450 BP) to cool/humid
(c. 2450 – 0 BP). Such conditions are documented in this
study (Fig. 3a–h) and from other lakes in Minnesota
(Bartlein et al. 1984; Winkler et al. 1986; Bartlein &
Whitlock 1993; Webb et al. 1993; Schwalb et al. 1995;
Schwalb & Dean 2002). At 4000 BP, both woody and
herbaceous species were abundant in the landscape
(Fig. 4). Increased humidity most likely increased NPP
of herbaceous taxa, fuel production and biomass consumption by fire (Briggs & Knapp 1995; Umbanhowar
1996; Burke et al. 1998; Paruelo et al. 1999; Fay et al.
2000; Lane et al. 2000; Knapp et al. 2001) (Fig. 3k,l).
It also increased the expansion of bigwoods taxa
(Fig. 4), possibly as a result of reduced moisture stress
and fire frequency. Lengthening the fire return interval
should promote an unstable assemblage by favouring
the establishment of deciduous forest species (Grimm
1983; Archer 1990; Peterson & Reich 2001), even in
more productive landscapes with greater total biomass
burning. Once established, the low flammability of
forest litter (Streng & Harcombe 1982; Fonteyn et al.
1984) would reduce landscape-level fire severity and
charcoal concentration recorded in lake sediment.
After c. 3000 BP, relatively humid conditions persisted
as indicated by high ARM/IRM ratios (Fig. 3a,b),
and the expansion of oak forests probably further
diminished fire (Figs 3k,l and 4). There is little evidence for direct climatic control of fire severity, because
during this period of relative climatic stability (Fig. 3a–
d), mean charcoal influx declined from 4.56–4.98 to
1.82 –1.95 mm2 cm2 year−1 (Table 3, Fig. 3k–n), coincident with the increased dominance of forest species and
the shift to relatively inflammable fuels.
By examining the relationship between combustion
temperature and charcoal production, Umbanhowar
& McGrath (1998) suggested an alternate hypothesis
for why charcoal might increase during humid periods.
They burned litter in a muffle furnace for 1 hour at temperatures of 300–500 °C and found that charcoal yield
(weight charcoal: weight unburned material) declined
with rising temperatures. It is possible that fires burned
cooler during humid periods if fuels were consistently
more moist, resulting in greater charcoal production
without increased fire severity. Although this effect
may occur to some extent, we believe that the primary
response we observe in our data is the productivity
effect on fire. The relationship between grassland NPP
and precipitation is strong (Briggs & Knapp 1995;
Burke et al. 1998; Paruelo et al. 1999; Fay et al. 2000;
Lane et al. 2000; Knapp et al. 2001), and the majority
(> 90%) of herbaceous fuel is consumed in fires (Johnson
& Risser 1975; Shea et al. 1996; Stocks & Kauffman
1997). We suspect that, even during humid periods,
there is sufficient opportunity for the drying of fine
fuels like grasses and forbs. Finally, fires burning at
cooler temperatures tend to produce charcoal that
is qualitatively more brown than black in colour
(Umbanhowar & McGrath 1998). Almost all charcoal
in Kimble Pond and Sharkey Lake was black, suggesting consistently high average burn temperatures at the
landscape scale throughout the Holocene.
Increased charcoal concentrations during humid
periods may also be a product of increased overland flow
and higher soil erosion rates (Whitlock & Millspaugh
1995). However, inorganic sediment and IRM declined
during humid conditions (Fig. 3a,b,g,h), indicating
that this is probably not the case. In addition, neither lake
has an inlet stream, and there was no indication from
AMS 14C dating (i.e. abnormal slope of the age-depth
model or large reversals) that significant quantities of
old charcoal from the landscape is being eroded and
transported.
The degree to which fire frequency controls changes
in community composition is not clear from our study.
Our data cannot be used to test directly Grimm’s (1983)
hypothesis that decreased fire frequency caused the
spread of deciduous bigwoods forest taxa because the
resolution of charcoal concentration in this study
(74–85 years) was too coarse to assess changes in fire
frequency at time-scales meaningful to recruitment of
fire-sensitive tree or shrub species (c. 3–5 years, Peterson
& Reich 2001; Heisler et al. 2003). However, the shift to
deciduous tree species during the humid period 3000–
4500 BP when fire severity was at its peak is prima facie
evidence that water stress limitations on tree recruitment and/or fire frequency declined during this period.
Changes in fire seasonality might also affect vegetation
(Grimm 1983; Clark et al. 2001) but would probably
only affect relative shifts in herbaceous taxa (e.g. C3
vs. C4 grasses) rather than shifts between grasses and
trees, which are controlled by fire frequency (Peterson
& Reich 2001). High-resolution fire records, such as
834
P. Camill et al.
annually laminated lake sediments, may allow assessments of how fire frequency, rather than severity,
changes across large climate transitions.
Acknowledgements
This work was supported by National Science Foundation DEB grant 0092704, NSF/ATM grant 9909523,
grants to Carleton College and St Olaf College from
the Howard Hughes Medical Institute, a grant to
St Olaf from Merck/AAAS and the LLNL Center for
Accelerator Mass Spectrometry, and Bush, Class of’49,
and FDE Fellowships from Carleton College. Sediment magnetic analyses were conducted at the Institute for Rock Magnetism, University of Minnesota.
Special thanks go to M. Sharkey and B. Reichel for
access to the lakes. We thank T. Brown at the Center
for Accelerator Mass Spectrometry, Lawrence Livermore National Laboratory, for assisting with the AMS
14
C dating and N. Nowinski for assistance with charcoal analysis. The following people were instrumental
in assisting with field work: T. Jeffers, A. Matney, M.
Swift, B. Welch and J. Lavine. P. Moore, L. Haddon and
two anonymous reviewers made helpful comments on
this manuscript.
References
© 2003 British
Ecological Society,
Journal of Ecology,
91, 822–836
Almendinger, J.E. (1990) The decline of soil organic matter,
total N, and available water capacity following the
late-Holocene establishment of jack pine on sandy
mollisols, north-central Minnesota. Soil Science, 150 (1),
680 – 694.
Almendinger, J.E. (1992) The late Holocene history of prairie,
brush-prairie, and jack pine (Pinus banskiana) forest on
outwash plains, north-central Minnesota, USA. Holocene,
2, 37–50.
Almquist-Jacobson, H., Almendinger, J.E. & Hobbie, S.
(1992) Influence of terrestrial vegetation on sedimentforming processes in kettle lakes of west-central Minnesota.
Quaternary Research, 38, 103 –116.
Amundson, D.C. & Wright, H.E. Jr (1979) Forest changes in
Minnesota at the end of the Pleistocene. Ecological Monographs, 49 (1), 1–16.
Anderson, R.C. (1982) The eastern prairie-forest transition –
an overview. Proceedings of the Eighth North American
Prairie Conference (eds J. Ballard, L.S. Brewer & C. Fox).
Western Michigan University, Kalamazoo, Michigan.
Archer, S. (1990) Development and stability of grass woody
mosaics in a subtropical savanna parkland, Texas, USA.
Journal of Biogeography, 17, 453 – 462.
Baker, R.G., Bettis, E.A., Denniston, R.F., Gonzalez, L.A.,
Strickland, L.E. & Krieg, J.R. (2002) Holocene paleoenvironments in southeastern Minnesota – chasing the prairie-forest
ecotone. Paleogeography, Palaeoclimatology, Palaeoecology,
177 (1–2), 103 –122.
Baker, R.G., Maher, L.J., Chumbley, C.A. & Vanzant, K.L.
(1992) Patterns of Holocene environmental-change in the
Midwestern united-states. Quaternary Research, 37 (3),
379 –389.
Banerjee, S.K., King, J. & Marvin, J. (1981) A rapid
method for magnetic granulometry with applications to
environmental studies. Geophysical Research Letters, 8,
333 –336.
Bartlein, P., Webb, T. & Fleri, E. (1984) Holocene climatic
change in the northern Midwest: pollen derived estimates.
Quaternary Research, 22, 361–374.
Bartlein, P.J. & Whitlock, C. (1993) Paleoclimatic interpretation of the Elk Lake pollen record. Geological Society of the
American Special Paper, 276, 275 –293.
Bragg, W.K., Knapp, A.K. & Briggs, J.M. (1993) Comparative water relations of seedling and adult quercus species
during gallery forest expansion in tallgrass prairie. Forest
Ecology and Management, 56 (1– 4), 29 – 41.
Briggs, J.M. & Knapp A.K. (1995) Interannual variability in
primary productivity in tallgrass prairie: climate, soil moisture,
topographic position, and fire as determinants of aboveground
biomass. American Journal of Botany, 82, 1024 –1030.
Briggs, J.M., Knapp, A.K. & Brock, B.L. (2002) Expansion of
woody plants in tallgrass prairie: a fifteen-year study of fire
and fire–grazing interactions. American Midland Naturalist,
147 (2), 287–294.
Brugam, R.B. & Swain, P. (2000) Diatom indicators of peatland development at Pogonia Bog Pond, Minnesota, USA.
Holocene, 10 (4), 453 – 464.
Burke, I.C., Lauenroth, W.K., Vinton, M.A., Hook, P.B.,
Kelly, R.H., Epstein, H.E. et al. (1998) Plant–soil interactions in temperate grasslands. Biogeochemistry, 42 (1–2),
121–143.
Camill, P. & Clark, J.S. (2001) Long-term perspectives on
lagged ecosystem responses to climate change: permafrost
in boreal peatlands and the grassland /woodland boundary.
Ecosystems, 3, 534 –544.
Clark, J.S. (1988a) Effect of climate changes on fire regimes in
northwestern Minnesota. Nature, 334, 233–235.
Clark, J.S. (1988b) Particle motion and the theory of charcoal
analysis: source area, transport, deposition and sampling.
Quaternary Research, 30, 67– 80.
Clark, J.S. (1990) Fire and climate change during the last 750yr in northwestern Minnesota. Ecological Monographs, 60,
135 –159.
Clark, J.S., Grimm, E.C., Lynch, J. & Mueller, P.G. (2001)
Effects of Holocene climate change on the C-4 grassland/
woodland boundary in the Northern Plains, USA. Ecology,
82, 620 – 636.
Clark, J.S. & Hussey, T.C. (1996) Estimating the mass flux of
charcoal from sedimentary records: effects of particle size,
morphology, and orientation. Holocene, 6, 129–144.
Clark, J.S., Lynch, J., Stocks, B.J. & Goldammer, J.G. (1998)
Relationships between charcoal particles in air and sediments in west-central Siberia. Holocene, 8, 19–29.
Clark, J.S. & Patterson, W.A. III (1997) Background and local
charcoal in sediments: scales of fire evidence in the paleorecord. Sediment Records of Biomass Burning and Global
Change (eds J.S. Clark, H. Cachier, J.G. Goldammer & B.
Stocks), pp. 23 – 48. Springer, New York.
Clark, J.S. & Royall, P.D. (1994) Preindustrial particulateemissions and carbon sequestration from biomass burning
in North America. Biogeochemistry, 24, 35–51.
Clark, J.S. & Royall, P.D. (1995) Particle-size evidence for
source areas of charcoal concentration in late Holocene
sediments of eastern North-American lakes. Quaternary
Research, 43, 80 – 89.
Clark, J.S. & Royall, P.D. (1996) Local and regional sediment
charcoal evidence for fire regimes in presettlement northeastern North America. Journal of Ecology, 84, 365–382.
Clark, J.S., Royall, P.D. & Chumbley, C. (1996) The role of fire
during climate change in an eastern deciduous forest at
Devil’s Bathtub, New York. Ecology, 77, 2148 –2166.
Conley, D.J. & Schelske. C.L. (1993) Potential role of sponge
spicules in influencing the silicon biogeochemistry of
Florida lakes. Canadian Journal of Fisheries and Aquatic
Science, 50, 296 – 302.
Dale, V.H., Joyce, L.A., McNulty, S. & Neilson, R.P. (2000)
The interplay between climate change, forests, and disturbances. Science of the Total Environment, 262, 201–204.
835
Climatic effects on
fire and vegetation
© 2003 British
Ecological Society,
Journal of Ecology,
91, 822–836
Danner, B.T. & Knapp, A.K. (2001) Growth dynamics of
oak seedlings (Quercus macrocarpa Michx. and Quercus
muhlenbergii Engelm.) from gallery forests: implications
for forest expansion into grasslands. Trees – Structure and
Function, 15 (5), 271–277.
Dean,W.E. Jr (1974) Determination of carbonate and organic
matter in calcareous sediments and sedimentary rocks by
loss on ignition: comparison with other methods. Journal of
Sedimentary Petrology, 44 (1), 242–248.
Digerfeldt, G., Almendinger, J.E. & Bjorck, S. (1992) Reconstruction of past lake levels and their relation to groundwater hydrology in the Parkers Prairie Sandplain, west-central
Minnesota. Mammology Palaeoclimatology Palaeoecology,
94 (1– 4), 99–118.
Efron, B. & Tibshirani, R. (1993) An Introduction to the Bootstrap. Chapman & Hall, New York.
Faber-Langendoen, D. & Davis, M.A. (1995) Effects of fire
frequency on tree canopy cover at Allison Savanna, eastcentral Minnesota, USA. Natural Areas Journal 15, 319 –
328.
Fægri, K., Iversen, J., Kaland, P.E. & Krzywinski, K. (1989)
Textbook of Pollen Analysis, 4th edn. John Wiley and Sons,
Chichester.
Fay, P.A., Carlisle, J.D., Knapp, A.K., Blair, J.M. & Collins,
S.L. (2000) Altering rainfall timing and quantity in a mesic
grassland ecosystem: design and performance of rainfall
manipulation shelters. Ecosystems, 3 (3), 308 –319.
Flannigan, M.D., Stocks, B.J. & Wotton, B.M. (2000) Climate
change and forest fires. Science of the Total Environment,
262, 221–119.
Fonteyn, P.J., Stone, M.W., Yancy, M.A. & Baccus, J.T. (1984)
Interspecific and intraspecific microhabitat temperature
variations during a fire. American Midland Naturalist, 112,
246 –250.
Geiss, C.E. & Banergee, S.K. (1999) Comparison of two interglacial records from the Midwestern USA. Physics and
Chemistry of the Earth, 24, 793–798.
Geiss, C.E., Umbanhowar, C.E., Camill, P. & Banerjee, S.K.
(2003) Sediment magnetic properties reveal Holocene climate change along the Minnesota prairie-forest ecotone.
Journal of Paleolimnology, in press.
Gleason, H.A. (1913) The relation of forest distribution and
prairie fires in the middle west. Torreya, 13, 173 –181.
Grimm, E.C. (1983) Chronology and dynamics of vegetation
change in the prairie-woodland region of southern Minnesota, USA. New Phytologist, 93, 311–350.
Grimm, E.C. (1984) Fire and other factors controlling the Big
Woods vegetation of Minnesota in the mid-nineteenth
century. Ecological Monographs, 54, 291–311.
Heisler, J.L., Briggs, J.M. & Knapp, A.K. (2003) Long-term
patterns of shrub expansion in a C4–dominated grassland:
fire frequency and the dynamics of shrub cover. American
Journal of Botany, 90, 423 – 428.
Hobbs, H.C. & Goebel, J.E. (1982) Geologic Map of Minnesota, Quaternary Geology, Map S-1. Minnesota Geological
Survey, St Paul.
Intergovernmental Panel on Climate Change (IPCC) (2001a)
Climate Change 2001: the Scientific Basis. Cambridge University Press, Cambridge.
Intergovernmental Panel on Climate Change (IPCC) (2001b)
Climate Change 2001: Impacts, Adaptation and Vulnerability. Cambridge University Press, Cambridge.
Jackson, S.T. & Lyford, M.E. (1999) Pollen dispersal models
in Quaternary plant ecology: assumptions, parameters, and
prescriptions. Botanical Review, 65 (1), 39–75.
Jacobson, G.L. & Grimm, E.C. (1986) A numerical analysis of
Holocene forest and prairie vegetation in central Minnesota. Ecology, 67, 958 –966.
Jelgersma, S. (1962) A late-glacial pollen diagram from
Madella, south-central Minnesota. American Journal of
Science, 260, 522 – 529.
Johnson, F.L. & Risser, P. (1975) A quantitative comparison
between an oak forest and an oak savanna in Central Oklahoma. Southwest Naturalist, 20, 75–84.
Knapp, A.K., Briggs, J.M. & Koelliker, J.K. (2001) Frequency
and extent of water limitation to primary production in a
mesic temperate grassland. Ecosystems, 4 (1), 19–28.
Knutti, R., Stocker, T.F., Joos, F. & Plattner, G.K. (2002)
Constraints on radiative forcing and future climate change
from observations and climate model ensembles. Nature,
416, 719–723.
Kutzbach, J.E. & Webb, T. III (1993) Conceptual basis for
understanding late-Quaternary climates. Global Climates
Since the Last Glacial Maximum (eds H.E. Wright et al.),
pp. 5–11. University of Minnesota Press, Minneapolis.
Lane, D.R., Coffin, D.P. & Lauenroth, W.K. (2000) Changes
in grassland canopy structure across a precipitation gradient. Journal of Vegetation Science, 11 (3), 359–368.
Marschner, F.J. (1974) The Original Vegetation of Minnesota
(map). U.S. Department of Agriculture, Forest Service, North
Central Forest Experiment Station, St Paul, Minnesota.
McAndrews, J.H. (1968) Pollen evidence for the protohistoric
development of the ‘Bigwoods’ in Minnesota (USA).
Review of Paleobotany and Palynology, 9, 201–211.
McKnight, R. (2000) Determination of Silica in Waters by
Flow Injection Analysis (Quikchem Method 10–114–21–
1A). Lachat Instruments, Milwaukee.
North American Pollen Database (2003) http://
www.ngdc.noaa.gov/paleo/napd.html.
Overpeck, J.T., Rind, D. & Goldberg, R. (1990) Climateinduced changes in forest disturbance and vegetation.
Nature, 343, 51–53.
Özdemir, Ö. & Banerjee, S.K. (1982) A preliminary magnetic
study of soil samples from west-central Minnesota. Earth
and Planetary Science Letters, 59, 393 – 403.
Paruelo, J.M., Lauenroth, W.K., Burke, I.C. & Sala, O.E.
(1999) Grassland precipitation-use efficiency varies across
a resource gradient. Ecosystems, 2 (1), 64 – 68.
Peterson, D.W. & Reich, P.B. (2001) Prescribed fire in oak
savanna: fire frequency effects on stand structure and
dynamics. Ecological Applications, 11 (3), 914–927.
Pitelka, L.F., Gardner, R.H., Ash, J., Berry, S., Gitay, H.,
Noble, I.R. et al. (1997) Plant migration and climate change.
American Scientist, 85, 464 – 473.
Pockman, W.T. & Sperry, J.S. (2000) Vulnerability to
xylem cavitation and the distribution of Sonoran desert
vegetation. American Journal of Botany, 87 (9), 1287–
1299.
Prentice, I.C. (1985) Pollen representation, source area, and
basin size: toward a unified theory of pollen analysis.
Quaternary Research, 23, 76–86.
Schwalb, A. & Dean, W.E. (2002) Reconstruction of hydrological changes and response to effective moisture variations from North-Central USA lake sediments. Quaternary
Science Reviewss, 21, 1541–1554.
Schwalb, A., Locke, S.M. & Dean, W.E. (1995) Ostracode
δ18O and δ13C evidence of Holocene environmental changes
in the sediments of two Minnesota lakes. Journal of Paleolimnology, 14, 281–296.
Shane, L.C.K. (1992) Palynological Procedures (Draft Manuscript). University of Minnesota, Limnological Research
Center, Minneapolis.
Shea, R.W., Shea, B.W., Kauffman, J.B., Ward, D.E.,
Haskins, C.I. & Scholes, M.C. (1996) Fuel biomass and
combustion factors associated with fires in savanna
ecosystems of South Africa and Zambia. Journal of
Geophysical Research-Atmospheres, 101 (D19), 23551–
23568.
Sperry, J.S., Nichols, K.L. & Sullivan, J.E.M. (1994) Xylem
embolism in ring-porous, diffuse-porous, and coniferous
trees of northern Utah and interior Alaska. Ecology, 75 (6),
1736 –1752.
836
P. Camill et al.
© 2003 British
Ecological Society,
Journal of Ecology,
91, 822–836
Sperry, J.S. & Sullivan, J.E.M. (1992) Xylem embolism in
response to freeze-thaw cycles and water-stress in ring-porous,
diffuse-porous, and conifer species. Plant Physiology, 100
(2), 605 – 613.
Stocks, B.J. & Kauffman, J.B. (1997) Biomass consumption
and behavior of wildland fires in boreal, temperate, and
tropical ecosystems: parameters necessary to interpret historic fire regimes and future fire scenarios. Sediment Records
of Biomass Burning and Global Change (eds J.S. Clark,
H. Cachier, J.G. Goldammer & B. Stocks), pp. 169–188.
Springer, New York.
Stott, P.A. & Kettleborough, J.A. (2002) Origins and estimates of uncertainty in predictions of twenty-first century
temperature rise. Nature, 416, 723 –726.
Streng, D.R. & Harcombe, P.A. (1982) Why don’t east Texas
savannas grow up to forest? American Midland Naturalist,
108, 278 –294.
Stuiver, M., Reimer, P.J., Bard, E., Beck, J.W., Burr, G.S.,
Hughen, K.A. et al. (1998) INTCAL98 radiocarbon age
calibration 24 000–0 cal BP. Radiocarbon, 40, 1041–1083.
Stuiver, M., Reimer, P.J. & Reimer, R. (1999) CALIB
Radiocarbon Calibration (HTML Version 4.2). http://
depts.washington.edu /qil /calib/.
Swain, P.C. (1979) The development of some bogs in eastern Minnesota. Dissertation. University of Minnesota,
Minneapolis, Minnesota.
Thompson, R. & Oldfield, F. (1986) Environmental Magnetism. Unwin-Hyman, London.
Umbanhowar, C.E. Jr (1996) Recent fire history of the
northern Great Plains. American Midland Naturalist, 135,
115 –121.
Umbanhowar, C.E. Jr (2003) Interaction of fire, climate and
vegetation change at a large landscape scale in the Big
Woods of Minnesota, USA. Holocene, 14 (3), in press.
Umbanhowar, C.E. Jr & McGrath, M.J. (1998) Experimental
production and analysis of microscopic charcoal from
wood, leaves and grasses. Holocene, 8, 341–346.
Waddington, J.C.B. (1969) A stratigraphic record of the pollen
influx to a lake in the Big Woods of Minnesota. Geological
Society of America Special Paper, 123, 263 –282.
Walther, G.R., Post, E., Convey, P., Menzel, A., Parmesan,
C., Beebee, T.J.C. et al. (2002) Ecological responses to
recent climate change. Nature, 416, 389–395.
Webb, T., Bartlein, P.J., Harrison, S.P. & Anderson, K.H.
(1993) Vegetation, lake levels, and climate in eastern North
America for the past 18 000 years. Global Climates Since the
Last Glacial Maximum (eds H.E. Wright, et al.), pp. 415–
467. University of Minnesota Press, Minneapolis.
Webb, T. III, Cushing, E.J. & Wright, H.E. Jr (1983) Holocene
changes in the vegetation of the Midwest. Late Quaternary
Environments of the United States. Volume 2. The Holocene
(ed. H.E. Wright Jr ), pp. 142–165. University of Minnesota
Press, Minneapolis.
White, A.S. (1983) The effects of thirteen years of annual
prescribed burning on a Quercus ellipsoidalis community in
Minnesota. Ecology, 64 (5), 1081–1085.
Whitlock, C., Bradbury, J.P. & Millspaugh, S.H. (1997)
Controls on charcoal distribution in lake sediments: case
studies from Yellowstone National Park and northwestern
Minnesota. Sediment Records of Biomass Burning and
Global Change (eds J.S. Clark, H. Cachier, J.G. Goldammer
& B. Stocks), pp. 367–386. Springer, New York.
Whitlock, C. & Millspaugh, S.H. (1995) Testing the assumptions of fire history studies: an examination of modern
charcoal accumulation in Yellowstone National Park,
USA. The Holocene, 6, 7–15.
Winkler, M.G., Swain, A.M. & Kutzbach, J.E. (1986) Middle
Holocene dry period in the northern Midwestern United
States: lake levels and pollen stratigraphy. Quaternary
Research, 25, 235–250.
Wright, H.E. Jr (1992) Patterns of Holocene climate change
in the Midwestern United States. Quaternary Research, 38,
129 –134.
Wright, H.E. Jr & Watts, W.A. (1969) Glacial and vegetational history of northeastern Minnesota. Minnesota Geological Survey Special Publication SP-II.
Wright, H.E. Jr, Winter, T.C. & Patten, D.L. (1963) Two
pollen diagrams from southeastern Minnesota: problems
in the regional late-glacial and postglacial vegetational
history. Geological Society of America Bulletin, 74, 1371–
1396.
Wright, H.E. Jr, Mann, D.H. & Glaser, P.H. (1984) Piston
corers for peat and lake sediments. Ecology, 65 (2), 657–659.
Received 23 December 2003
revision accepted 13 May 2003