SOIL-DEPENDENT FIRE FREQUENCY: A NEW HYPOTHESIS FOR THE DISTRIBUTION OF PRAIRIES AND OAK WOODLANDS/SAVANNAS IN NORTH CENTRAL AND EAST TEXAS

15 Pages·2003·9 MB·English

by  George M Diggs

}
SOIL-DEPENDENT EREQUENCY:
FIRE
NEW
A HYPOTHESIS EOR THE DISTRIBUTION
AND
OAK WOODLANDS/SAVANNAS
OF
PRAIRIES
AND
NORTH CENTRAL EAST TEXAS
IN
George
M.Diggs, C.SchuIze
Peter
Jr.
Department & Department &
of Biology of Biology
Sherman, TX Sherman, TX
75090, U.S.A. 75090, U.S.A.
and
[email protected]
Botanical Research Institute of Texas
Fort Worth, TX 76 02-4060
1
[email protected]
RESUME^
El patron listoricode vegetacion en [que alternan pra
e
robledales /sabanaen suelosarenosos en centro-norte
el
deuns.glci.Muchosaiitoreshanatnb uido estepanona
enelsuelo disponible
idLU
humedad lopn
niveles ina e en Isue
e
embargo, .staexpl.ca,cion no es cons isten tc Lon IS o\
e
1
none
historica de las prade ras y bosques e entio
itesdelsueloen
frecuencia e intensidad del fuego.
INTRODUCTION
Conditions the time of European Settlement, Approximately 1800 A.D.
at
The vegetation of North Central and East Texas at the time of European settle-
ment was
(hereafter referred to as presettlement vegetation) characterized by
and oak woodlands/savannas
well-defined zones of tallgrass prairie (Fig.
1).

From west praine bands Grand (composed
to east, the are the Prairie of the
Fort Worth Prairie and the Lampasas Cut Plain), the main belt of the Blackland
San Antonio and whereas oak wood-
Prairie, the Prairie, the Fayette Prairie, the
land/savanna belts are the West Cross Timbers, the East Cross Timbers, and
Oak
three bands of Post Savanna. These were highly recognizable areas of veg-
maps
etation, with the Cross Timbers appearing on the earliest and the
numerous and
presettlement prairies described by explorers, early
settlers, re-
searchers as almost devoid of trees Gregg 1844; Brooke 1849; Parker 1856;
(e.g.
&
Thomas
Hatch
Dyksterhuis 1946; 1962; Correll Johnston 1970; 1990;
et
al.
we
Diggs et 1999; Telfair 1999; Francaviglia 2000). Hereafter refer to these
al.
two and Woodlands. The
vegetation types as Prairies current climate of the re-
presumably unchanged
gion, relatively since presettlement times, character-
is
summers
ized by wet springs, dry with occasional thunderstorms, moderate to
many
and As
strong winds, periodic multi-year droughts. in areas of central
common
North America, prairie fires were prior to settlement (Stewart 1951;
&
Komarek 1965, 1966; Wells 1970; Wright and Bailey 1982; Collins Wallace
The
1990; Bragg 1995). relative frequency of ignition by lightning versus Native
The Woodlands represent the westernmost extension of the eastern decidu-
ous In general, the vegetation was composed of an oak overstory with
forest. tall
The were dominated by Schizachyrium scoparium
grasses. grasses
(little
bluestem), with Andrapogongerardii (big bluestem) and 5orghastrumnutans(In-
While woody was
dian grass) as lesser dominants. varying locally, the vegetation
in general dominated by two trees, Quercus stellata (post oak) and Quercus
marilandica (blackjack oak). The tree density of the oak woodlands was variable,
ranging from quite open to dense thickets. Some early accounts described wood-
lands through which wagons could easily pass Marcy 1853, 1866), while oth-
(e.g.
ers described almost impenetrable thickets Kendall 1845). Gregg (1844) ob-
(e.g.
served a variety of tree densities, noting that, "Most of the timber appears to be
kept small by the continual inroads of the 'burning being killed
prairies;' al-
for,
most annually, constantly replaced by scions of undergrowth; so that be-
it is it
comes more and more some
dense every reproduction. In places, however, the oaks
are of considerable and able to withstand the conflagrations."
size,
The presettlement Prairies were vast grasslands dominated by
Schizachyrium scoparium bluestem), with Andropogon gerardii (big
(little
bluestem) and Sorghastrum nutans (Indian grass) as lesser dominants, and
woody
vegetation generally limited to areas along the larger watercourses, as
scattered mottes, or associated with locations that were protected from such
fire,
as mesas and buttes (Smythe 1852; Parker 1856; Hill 1901; Diggs et 1999). For
al.
Grand
example, Hill (1887) described the Prairie as prairie region, utterly
"a
destitute of timber" and Kendall (1845) wrote, also of the Grand Prairie, "As far
as the eye could reach nothing could be seen but a succession of smooth,
... ,
gently-undulatmg prairies." Early accounts of the Blackland Prairie were simi-
Smy
lar the (1852) described the eastern edge of the Blackland Prairie as having
a view of almost boundless Prairie stretching to the north, as far as the eye
'\.,
could reach. and further, as "nearfy destitute of trees."
..."
Soils of North Central and East Texas
Most soils of the Prairies of North Central and East Texas are derived from lime-
rich Upper Cretaceous rocks which weather to form soils with substantial lev-
Outlying segments Blackland and San Anto-
els of clay of the Prairie (Fayette
nio prairies) have soils developed from younger Tertiary age deposits. While
the majority of Tertiary deposits in East Texas are sandy in nature those
(i.e.,
Oak
supporting the Pineywoods and Post Savanna), those underlying Prairie
and Cook Mountain
Eleming, Oakville Sandstone, formations) general
in
(e.g.,
and some which
have a relatively high clay content in cases develop soils dis-
microtopography high
play the gilgai so typical of certain clay soils
& &
Diamond
(Launchbaugh Smeins Smeins
1955; 1983; Miller 1988). Eurther,
where clay lenses are found in other geologic strata outcropping in isolated pock-
Oak
ets of the Cross Timbers, Post Savanna, and Pineywoods, areas of prairie
vegetation can again be found (Dyksterhuis 1948; Hill 1991). thus clay that
It is
appears crucial in the development and maintenance of the grassland vegeta-
some abundant throughout
tion characteristic of the Prairies. In cases clay
is
loam
all soil horizons, while in others there a clay-loam or surface layer—all
is
amounts
the Prairie however, have significant of clay (Godfrey 1973;
soils, et al.
&
Smems
Diamond Woodlands
1985). Conversely, the are developed in general
from Cretaceous and Tertiary sandstone rocks of such geologic layers as the
Antlers/Trinity, Woodbine, Carrizo, and Wilcox, and can be generally described
&
Hartmann
as sandy (Sellards et ak 1932; Scranton 1992).
Prevailing Explanation for the Distribution of Prairies and Woodlands
As discussed above, the presettlement distribution of the Prairies corresponds
to the distribution of limestone parent material overlain by alkaline soils with
Woodlands
high clay content, while the presettlement occurred on either
a
m
westernmost on
sandy, slightly acidic soils the part of the region, gravelly
or,
and
rocky substrates (Sellards et 1932; Dyksterhuis 1948; Diggs et 1999).
al. al.
The correspondence and
striking of vegetation soil types led eariy writers to
propose that the distribution of Prairies and Woodlands was due to the differ-
ent water holding capacities of clays and sands. While the basic hypothesis has
many
been stated ways, essence that greater infiltration and water storage
its is
m
at depth sandy soils permits tree survival, while clay soils do not support
trees because of inadequate infiltration of water (and reduced root penetration)
due to the clay's relative impermeability.
Hill (1887) appears to have been the first to propose this hypothesis. He
argued
that.
DSCHULZE,SOIL-DEPENDENT FREQUENCY
FIRE
Tharp
Subsequent authors reiterated contention. For example, (1926) sug-
Hill's
gested that the sandy soils increased available soil moisture for tree growth.
Weaver and Clements (1938) argued that, the oaks have been able to main-
"...
...
tain themselves against the competition of the grasses by virtue of the favor-
much
summarized
able chresard of the sandy Dyksterhuis (1948) of this
soil."
information, and apparently agreed with the general idea of sandy soil-mois-
growth and
ture availability favoring the of trees. Allred Mitchell (1955) reiter-
"Sandy, gravelly and rocky areas also improve local soil-moisture conditions in the grassland
..^
formation. Rainfall intake is high on these soils, which also yield up moisture to plants more
readily and more completely than do heavier soils. In these areas, soil-moisture conditions
...
have been improved sufficiently over the climatic normal so that trees or shrubs, particularly
oaks, form savannah with the understory composed of grassland dominants of the True Prairie."
This reasonably intuitive explanation for the historic distribution of the Prai-
and Woodlands has been nearly universally accepted, including by recent
ries
authors Diggs Francaviglia 2000).
(e.g., et al. 1999;
Woodlands sandy
However, the hypothesis that the are restricted to soils
because of inadequate moisture availability on clay soils cannot be correct be-
cause, in fact, trees occur extensively on the clay soils. As noted above, eariy
woody and
accounts noted vegetation along the larger watercourses as isolated
clumps on
mottes or of trees in scattered locations the clay soils of the Prairies.
More impressive, however the current invasion of trees onto extensive up-
is
land areas of clay soil that can be widely observed throughout the Prairies (e.g.,
&
Launchbaugh Smeins Diamond few remaining
1955; 1986). Indeed, the prai-
woody
management
remnants prevent invading veg-
require active to loss to
rie
&
Diamond mowing, some
etation (Smeins 1986). In the absence of or other
fire,
suitable disturbance, trees such asjuniperus virginiana (eastern red cedar),
Madura
Gleditsia triacanthos (honey-locust), Celtis laevigata (hackberry),
Ulmus
pomi/era(bois-d'arc),ProsopisgIandu/osa(mesquite),and
crassifolia (ce-
and
dar elm) rapidly invade the native prairie cause the vegetation to convert
A
and then woodland/forest. glance from the road along almost
to a thicket a
any major highway on the present day Blackland Prairie provides abundant
woody
evidence of rapid invasion by species in the absence of disturbance (Fig.
In areas no longer cultivated or otherwise disturbed, the invasion by trees
2).
can be observed within a relatively few years— this is particularly obvious in
the numerous areas that were cultivated until relatively recently therefore
It
growth on and
seems clear that there sufficient moisture for tree the clay soils
is
m
on Puui and bandy Woodlands
llcicncc the \cgctatioii ilic clav icb
modern
should be noted that the uood\ specie-^ piesent on the day Prai-
It
les and Woodlands aie largely different, probably due to the different soil re-
I
quirements the vaiious species, and we do not view wood) plant encroach-
of
ment on the Prairies as simply an expansion of the Woodlands. Nonetheless,
woody now on many The woody
vegetation extensive formerly Prairie areas.
is
species currently encroaching on areas of Prairie vegetation were probably not
components Woodland
general major of the adjacent presettlement vegeta-
in
Rather, they are probably species that in presettlement times were present
tion.
in low numbers on the Prairies themselves near streams or where topography
On
made some
unlikely. the other hand, species (e.g.Juniperus asheij.
fire
vi rgi nia na) are at present highly invasive on both tlie Prairies and Woodlands.
HYPOTHESIS
dependent frequency
Soil- fire
We
and Woodlands
propose that the presettlement distribution of Prairies in
North Central and East Texas was not due to insufficient moisture for tree
growth on clay soils, but rather to differences in fire frequency on different soil
We
types. hypothesize that higher fuel quantity on clay soils increased the fre-
quency and intensity of fire, and that fire in turn suppressed the growth of trees.
woody
Prairie fires are fueled primarily by grasses, as opposed to forbs or veg-
etation, so an increase in grass biomass leads to an increase in the quantity of
We
predict that grass biomass typically higher on clay than on sandy
fuel.
is
due and
soils to better moisture nutrient availability at the shallower rooting
woody
depth
of grasses versus plants.
This situation would represent two alternative positive feedbacks. High fuel
quantity on clay would encourage which would suppress woody vegeta-
fire,
tion and under certain conditions depending on season of burn— Howe
(e.g.,
1995) stimulate subsequent grass growth (by removing dead biomass which
new Low
hinders growth), thereby maintaining high quan-
fuel quantity. fuel
tity on sand would reduce the chance of which would foster invasion by
fire,
would
trees that then further suppress grass biomass by shading or other
(e.g.,
competition) and the subsequent frequency and intensity of These alter-
fire.
would and Woodlands
native feedbacks lead to alternative stable states. Prairies
and Wood-
(Fig. This particular hypothesis for the distribution of Prairies
3).
lands consistent with the more general conclusions of Scholes and Archer's
is
(1997) review of tree-grass interactions around the world. They write that "Moist
environments our support vigorous grass growth
fertile Prairies] a that,
[e.g. if
not grazed, leads frequent intense Semi-arid environments on sandy,
to
fires....
low our Woodlands] seldom
fertility soils are treeless."
[e.g.
Soil type (and effects on grass biomass) not the only variable that af-
its is
how
Whether
fects the frequency or intensity of fire. a fire ignites, hot it burns,
number
and ability to spread also depend on a of other variables including
its
wind
frequency of ignition events, season of year, rainfall, humidity, speed, to-
pography, and grazing (some of which also affect grass biomass). These vari-
would combine manner
ables in a stochastic to increase or decrease the fre-
quency and/or intensity of fire in any given location at any particular time.
For example, frequency and intensity would be low during a wet summer,
fire
but the resulting grass growth could combine with a subsequent windy drought
to increase fire likehhood and intensity the following year. Thus, the system-
atic effect of soil type on fire would be increased or decreased at any given loca-
tion at any particular time by the net effect of these other variables, with the
result being substantial variation in the time since the last fire in different ar-
When
both and Woodlands.
eas of the Prairies the this stochastic variation
is
taken into account, the fire-frequency hypothesis can explain not only the
(1)
historical distribution of Prairies and Woodlands, but also: (2) the historically
dominant tree species of the Woodlands; the historical occurrence of iso-
(3)
lated groves of trees on clay soils; (4) the present invasion of Prairies by trees;
and between dominant
the difference the historically tree species of the
(5)
Woodlands and abundance on Woodlands
the species that are increasing in the
and
invading the Prairies today (e.g.Juniperus species).
The VVoixil.uids were historically dominated by oaks, species with thick
bark, stubb)' branches, and the abihty to resprout from roots, features that im-
parl hence to grass Further, occasional extended periods without
rcsi fires. fire
on the Prairies would have allowed the establishment of the isolated groves of
trees that were observed by Western Once established, such groves
settlers.
would have been unlikely to burn due to the suppression of grass growth by
and
the trees the resistance of large trees to
fire.
The
present invasion of the Prairies by trees can be explained by a lack of
that has resulted from intentional suppression plus numerous and ex-
fire fire
human
tensive fire breaks that have been created by activities (roads, cultivated
fields, overgrazed areas). Moreover, once this process begins, any Prairie area
becomes invaded by would biomass and come
that substantially trees lose grass
an
to serve as additional firebreak, thereby further reducing the likelihood of
many
on remaining
fire adjacent prairies. Finally, the present tree invasion of
Woodlands
Cbut certainly not all) areas of the Prairies (and of the understory)
dominated by Juniperus virginiana (eastern red cedar) and Junipcrusashei
IS
and
(Ashe's juniper), species that are sensitive to fire (easily scorched/ignited
from
unable resprout
to roots).
The
key assumption of our hypothesis that the difference in grass biom-
is
was on Woodlands
ass sufficiently higher the Prairies than the to raise the like-
and on compared
lihood, frequency, intensity of fire the clay soils of the Prairies
to the sandy or rocky soils of the Woodlands. In addition, our hypothesis leads
to the prediction that patches of open grassland on sandy soil are very rare
management woody
except in instances of active suppression of species)
(e.g.,
or immediate proximity to clay soils (which would result in more frequent fires
due
to closeness to the fire-prone prairie vegetation).
We
have not measured grass biomass on intact remnants of the Prairies
and Woodlands, and do
therefore not have the data necessary to estimate dif-
number
with
ferences in a of variables associated frequency, intensity,
fire
(e.g.,
extent, pattern, season, Further, while some biomass and fuel loading data
etc.).
& &
are available in the literature (e.g.,Johnson Risser 1974; Engle Stritzke 1995),
we
have been unable to find directly comparable data for the area of study Plow-
substantial indirect evidence consistent with the assumption of higher
ever,
is
grass biomass on clay soils. First, during dry periods clay soils generally hold
more water at grass rooting depths than do sandy This due to the rela-
soils. is
number
and
tively large surface areas of the individual clay particles the large
which
of very small pores act as billions of capillary tubes collectively holding
amounts The
large of water (Vankat 1979). result of this increased water hold-
may much
ing capacity that plants rooted in such continue active growth
is soils
later in the dry season than plants rooted in coarser soils (Daubenmire 1974;
Burgess 1995; McAuliffe 1995; Tucker 1999; Greeves et 2000; Ball 2001). Fur-
al.
thermore, undisturbed Blackland form microtopographical surface
soils gilgai,
features that function like shallow basins, increasing water retention during
& &
heavy rains (Hay ward Yelderman 1991; Diamond Smeins 1993). Early set-
tler accounts and observations of existing prairie remnants the Nature
(e.g.,
Conservancy's Clymer Meadow preserve in Hunt County, the Matthews-
Kaufman
Cartwright-Roberts Prairie in County, and Austin College's Garnett
Prairie in Grayson County) suggest that these "hog wallows," (as they were
known
to early settlers) were abundant on Vertisols of the presettlement Black-
Temporary
land Prairie. water storage in gilgai depressions of one-half acre foot
much
of water per acre of flat prairie have been estimated. As as six inches of
began
rain could be temporarily trapped in these structures before runoff (Hay-
&
ward Yelderman 1991). Meanwhile, the high surface area and negative sur-
face charges of clay particles give clay soils high cation exchange capacity. This
more
allows these soils to hold ionized minerals or nutrients, including those
essential for plant growth (Foth 1990; Whitehead 2000; Harpstead et 2001;
al.
O'Connell 2001). therefore not surprising that indirect evidence, such as
It is
among
agricultural productivity, suggests that the Blackland clay soils were
&
most (Haywood Yelderman
the fertile soils west of the Mississippi River 1991).
In addition, the high below-ground biomass of Prairie vegetation serves to con-
tinually add organic matter to the soil thereby functioning as a positive feed-
back mechanism to mcrease fertihty and water holding capacity (in part due
to the surface area provided by the additional organic material). Conversely,
sandy soils have larger pores that allow water to drain more easily They not
only dry earlier during dry periods but "the more water that percolates through
more washed out—
the the nutrients are particularly nitrogen, potassium
soil,
and
sulfur" (Tucker 1999). Therefore, soils that are high in sand, like those of the
Woodlands, are often poor for plant growth since they are relatively infertile
and well-dramed (Vankat
often too 1979).
Confounding
factors
have
Several other factors the potential to operate
tion to the soil-mediated frequency hypothec
fire
between grass-dominated and woody Of importance
vegetation. particular
is
Van Auken
the grazing regime (2000), tor example, has identified high levels
herbuoi by domestic animals as the primary cause of brush encroachment
ot y
m
the scmiai id giasslands of the southwestern U.S. This influence probably
is
the Rsuhol combination oi disturbance and reduced frequency. The graz-
fire
1
K^
mt
n^ during picscttlcmcnt times was certainly very different than at
pu
SL ni nstt \d of domestic iniin rls continuously confined to limited areas of
I
K
pisti the piimary heibn 01 es were large dense herds of migratory bison free
m
distances The azmg regime would thus have been extremely
ast gi
^ -IN c >\ L \
t
1
K and While move
lioth tempoially spatially a large herd could through
i,ul
and
in ickland Praine crop the vegetation very short (thus prevent-
lie
131
1
I
m
some would
in^ liitsj seasons oi years a given area probably be missed en-
mg
tiuK k u large amounts of standing biomass (heavy fuel load for fires).
L ndct such conditions \\ould not be expected every nor would be
fire year,
it
maintenance would
csstnti c\ct year the of the grassland. Rather, be
for fire
\1
)
cntical only w ith enough legularity to prevent trees from growing large enough
become
invulnerable grass
to to fires
Another grazing related influence the role of dense herds of grazers on
is
woody and
controlling vegetation stimulating Savory has
grasses. (1998) stressed
the impoi rncc of hci ding giazers on maintaining grasslands and has empha-
t
cd cxttcmel) didcicnt consequences human-controlled
the of typical graz-
si
Uong
ing tcginus pel lods of cxposuic to low densities of animals) and natural
s\ stems (ht^c mi^i atoi hcidsat high density due to threat of predation). He
\
h IS n ittd thu LxticmcK hi^h densities of grazers (and thus damage to woody
m
pi mtsj be of ct itic impoitincc in preventing encroachment of woody
i\ il
m
md
A
shiltm^ hihncc grazing adapted
\i-^(.t >n the favor of grasses. dif-
\ti
impKt may
fcicnt eis is th seed dispersal be of major importance in
elf \ it
j,y
the in\ ision of giasslands b) cert \m woody species. Data of Brown and Archer
may
and
(1999) suggest that rates patterns of seed dispersal be the primary de-