IMPACT OF FOREST FIRES ON HYDROLOGICAL P
. .
IMPACT OF FOREST FIRES ON
HYDROLOGICAL PROPERTIES OF A TYPICAL
MEDITERRANEAN FOREST SOIL
O. González, V. Andreu, E. Gimeno-García & J. L. Rubio
Centro de Investigaciones sobre Desertificación-CIDE (CSIC-Universitat de ValenciaGeneralitat Valenciana). Camí de la Marjal, s/n. 46470 - Alba1 (Valencia, SPAm).
Fax: +34 96 127 09 67
Tel.: +34 96 122 05 40
E-mail: [email protected]
ABSTRACT
Fue is one of the most important determining factors in the evolution of
Mediterranean ecosystems. Its effects on soil are plural and diverse, acting on structure,
chemical and physical properties, biota etc. Among them, the induced variations on soil
water dynamics are of key importance for the recovery of the entire ecosystem and in the soil
response to erosion processes. Forest fues are also a factor, triggenng the risk of
desertification. The aim of this study is to assess the impact effect of different fue intensities
on soil hydrology and on runoff generation.
This work was developed in the Permanent Experimental Field Station of La
Concordia with nine plots (4 x 20 m), installed in a calcareous hillside, representative of
Mediterranean shrubland areas. Experimental fires, of two intensity levels were carried out.
Three plots were burned reaching high fue intensity and three other plots were burned with
moderate intensity and the remaining plots were left undisturbed as control. Soil water
content, water retention capacity and pF curves were measured together with runoff
generation dynamics, in the different plots. The intrinsic characteristics of each rain event
occurred up until ayear after the fues and runoff generated on them was monitored.
The data obtained was compared with the state of the plots after seven years,
during the 2000-2002 period. In this period 24 rain events with runoff generation occurred,
with average rainfall intensities (130)around lOmm h". Both fire treatrnents show significant
differences with respect to the Control plots, which are reflected in a value of runoff
production of 76.84% as an average, less than the burned plots. Between fue treatments, the
plots that bumed with high intensity, show the highest values of runoff yield. However,
infiltration rates do not give significant differences between fue treatments.
In the same way, plots that suffer a high intensity f r e show greater values (22.50
cm3 cm'3) on water retention capacity than the other treatments, giving significant differences
with the Moderate intensity plots and Control plots. Differences on this parameter between
plots bumed with Moderate intensity and the Control plots were obsewed but they were not
489
Desertijication in the Mediterrarrean Region: a Securiw Issue, 489-507.
Williarn C. Keprier, Jose L. Rubio, DavidA. Mouat & Fausto Pedrauirri, e&.
O 2006 Spriirger. Printed in tlre Netlrerlands.
O. GONZALEZ, V. ANDREU, E. GIMENO-GARCIA & J.L. RUBIO
statistically significant. The obtained pF curves show the same tendencies, being the greatest
water content retained at pF of 2 in the burned plots and in the Control plots at pF of 4.2.
The effect of f ~ degrading
e
the vegetation cover and by means of the temperature
irnpact on soil, produce changes in its sttuctural characteristics and porosity, affecting soil
water distribution and the effective response on water erosion processes.
Keywords: water retention capacity, pF curve, hydrology, experimental f ~ eMediterranean,
,
desertification.
INTRODUCTION
Wildfires are known to have occurred in the Mediterranean region
through historic and prehistoric times (Naveh, 1990). In the last few
decades, fires of human origin have substantially increased not only in their
nurnber but also in their fi-equency, which have produced reduction in the
recovery periods and could drastically change or even eliminate the
vegetation cover, changing and degrading the ecosystem characteristics
(Andreu, 2002).
Actually, different studies indicate that the tendencies of fires in the
Mediterranean region are to decrease the affected area but increase in
nurnber (European Cornmission, 2002). It could mean that many
regeneration zones could be affected by fire again, breaking its possibilities
of resilience and becoming in regression by alteration of its physical,
chemical and biological functions and processes. Forest fires also increase
desertification risk in the Mediterranean, particularly when there are some
critica1 circumstances, such as: fire on very step slopes and shallow soils
underlying consolidated parent material; fire affecting vulnerable soils or
soil with poor conditions for its re-vegetation; the effects of repeated fires
on the same area and also the incidente of torrential rain in recently bumed
areas.
In the Mediterranean area, as a result of the high rainfall intensities
concentrated after the dry surnrner period, where the fires occur, soil
infiltration and even water storage capacity, mainly in shallow soils, is
easily exceeded, generating important soil losses and high energy runoffs as
a consequence of the relief features (Rubio and Calvo, 1996) that can reach
as much as one to three orders of magnitude (De Bano, 2000; Robichaud, et
al., 2000). This is usually due to soil changes induced by fire, which
depending on its severity or intensity, could be beneficia1 or deleterious to
the ecosystem (Neary, et al., 1999). Between these changes, loss of
vegetation cover and soil structure alterations are factors that produce
amplified hydrological responses (De Bano, 1998).
IMPACT OF FOREST FIRES ON HYDROLOGICAL PROPERTIES
ncrease
e some
Generally, the high intensity fires and the greater structural
breakdown of aggregates, produce formation of microaggregates and
liberation of soil particles, easily removed by runoff (Prosser and Williarns,
1998; Benavides-Solorio and McDonald, 2001). This effect will favour soil
sealing, crust formation, decreasing soil water retention capacity and
reduction in pore size characteristics that incise directly in a reduction on
infiltration and drainage increasing soil erodibility (Le Bissonais, 1996;
Teman and Neller, 1999; Andreu, et al., 200 1).
Post-fire conditions of the surface horizons are of key importance
because they determine the response of soil exposed to raindrop splash,
overland flow and the development of water repellent soil conditions (De
Bano, 1981). These changes in soil characteristics and the degradation of
the vegetation cover and litter layers can lead to very large increases in soil
erosion (Benavides-Solorio and McDonald, 2001).
These facts suggest that one important way to carefully study the
effects of fire on the soil ecosystem is through the performance of fires in
experimental plots (Giovannini and Luchesi, 1997). Then it is possible to
know and measure soil conditions before, during and after the fire
experiment and to improve the knowledge about the hydrology of the zone
affected by fire which could reach different intensities.
In this context, the aim of this study is to evaluate the evolution of
the hydrological characteristics of a typical Mediterranean forest soil
affected by different fire intensities in 1995 and compare it with its state in
2002. Changes in the response of these soils on runoff generation in each
erosive rain event of the time period considered have been also studied.
burned
MATERIALS AND METHODS
ansities
U, soil
oils, is
ioffs as
n reach
iaud, et
which
,ious to
oss of
iroduce
Study Area
This work was developed in the Permanent Experimental Field
Station of La Concordia (Lliria, SPAIN), at 50 lan NW of Valencia city. It
is 575 m above sea leve1 (Figure l), on a forested hillside facing South
South East, with a sclerophyllous shmb cover, regenerated after a previous
wildfire occurred in 1978. The most abundant species include Rosmarinus
ofJicinalis, Ulex pawzjlorus, Quercus coccifera, Rhamnus lycioides, Stipa
tenacissima, Globularia alypum, Cistus clusii and Thymus vulgaris.
O. GONZALEZ, V. ANDREU, E. GIMENO-GARCÍA & J.L. RUBIO
IMPACT
collection.
A rs
three plots c
High and h
different am
', respectivc
Control trea
fire intensit
similar to th
Clin
rainfall evei
transmissior
Soil Anal:
Figure 1. Geographical location of the Permanent Experimental Field Station of
La Concordia (Casinos, Spain)
Climatically the area belongs to the dry ombroclimate of the lower
mesomediterranean belt, according to Thornthwaite's classification. The
average annual precipitation is around 400 rnm, with two maxima, autumn
and spring, and a dry period from June to September with mean monthly
temperatures ranging fiom 13.3"C in January to 25.8OC in August.
The soil is a Rendzic Leptosol (FAO-UNESCO, 1988), or Calcic
Xerochrept type accordiig to Soil Taxonomy classification (Soil Survey Staff,
1990), developed on Jurassic limestone. This soil has a variable depth,
always less than 40 cm, abundant stoniness (E40%) and good drainage.
Experimental Set-Up
The station consists of a set of nine erosion plots, each 4 m wide x
20 m long, with similar characteristics such as soil morphology, slope
gradient, rock outcrops and vegetation cover. The selection of each plot
location was made after intensive surveys of the vegetation, soil and the
morphology pattern, based on across slope transects every two meters.
Plots are oriented parallel to the slope and bounded by bricks. At the
foot of each plot, a 2 m wide collector ran into a 1500 1tank to record al1 the
runoff and sediment produced during each rainfall event. Inside them there
Soil
and immedi
screened to
until their ar
Soil
-33, -300,
membrane I
was calcull
Cameron (1
Wh
gravimetric
density of
percentage.
Wa
with field 1
The pF cur
Ru
al1 plots, d
2002.
IMPACT OF FOREST FIRES ON HYDROLOGICAL PROPERTIES
is a 30 1 tank to concentrate the sediments produced, facilitating its
A random design of three different fue intensity treatments (with
three plots each) was used. Two sets of three plots were bumed reaching
High and Moderate fire intensities, which was achieved by addition of
different amounts of fue1 load to the plots of each treatment, 40 and 20 t harespectively. The remaining three plots were unbumt to be used as a
Control treatment. The necessary quantity of dry biomass to obtain the two
fire intensities was added using vegetation (fiom the surrounding area)
similar to that present in the plots.
Climatic parameters and the intrinsic characteristics of the different
rainfall events were monitored by a logging system of sensors with GSM
transmission of data, placed close to the plots.
',
Soil Analysis and Measurements
Soil samples were taken fiom the top 5 cm of the soil surface before
and immediately after fue, and in the summer of 2002, they were air-dried,
screened to remove the fiaction >2 rnm diameter and stored in plastic boxes
until their analysis.
Soil Water Content (SWC) was calculated for the potentials: 0, -10,
-33, -300, and -1500 KPa, or pF 1, 2, 2.5, 3.5, 4.2, using the pressure
membrane method (Richards, 1947). Soil Water Retention Capacity (WRC)
was calculated for each soil sample using the equations of McLaren and
Cameron (1996).
(a) WRC=(
n wide x
Y, slope
:ach plot
and the
;s. At the
rd al1 the
em there
e10- e1500 )* pb
m) WRC=( e33- elsoO)* pb
Where WRC is soil Water Retention Capacity, 0 10, 0 33, 0 1500 are
gravimetric water volume at - 10 KPa, -33 KPa, and -1500 KPa, and pb bulk
density of soil samples. The results were obtained in volumetric units in
percentage.
Water Retention Capacity with field capacity at -10 KPa (a) and
with field capacity at -33 KPa (b), were calculated for the different plots.
The pF curves were also determined.
Runoff generation dynamics were monitored for each rain event in
al1 plots, during one year after the 1995 fires and in the period of 20002002.
Soil organic matter content was determined by oxidation with
potassium dichromate (Jackson, 1958). Total Nitrogen was determined by
micro-Kjeldahl automatic analyser using the Bremmer method (Black,
1965). Ammonium and nitrate Nitrogen were extracted with 2N KC1
solution and determined by stearn distillation by micro-Kjeldahl automatic
analyzer using the Bremmer method (Black, 1965). Electrical conductivity
was measured in soil saturation extracts by the Richards method (1964).
Available Phosphorous was measured by colorimetry according to the Olsen
and Dean method (Black, 1965). Aggregate stability was assessed using a wetsieving procedure (Pnmo and Carrasco, 1973).
Standard statistical analyses were applied at 95% of signification
level. Analysis of variance and Tukey's test at a=0.05 were used to detect
differences in WRC and pF values, according the different fire treatments.
MAIN RESULTS AND DISCUSSION
Rainfall trends
During the studied period (2000-2002), 24 rain events with runoff
generation were monitored, showing average rainfall intensity (130) of 9.38
mm h-l. The rain behaviour in year 2002, on quantity, characteristics and
distribution, showed a change compared to the previous years (Figure 2).
Since the Experimental Station started in 1995, the trend of the
precipitations in the area has showed substantial variations, among those it
is necessary to highlight the intense period of drought which occurred
during 1998 and 1999. Year 2000 observed a certain tendency of recovery
in the rain regime, reaching 415.90 mm of the annual rainfall. However,
although the total annual rainfall in year 2002 reached 521.50 mrn, a net
decrease of precipitation occurred in 2001 with a total rainfall of only
267.40 mm.
It is also possible to appreciate this variability in the characteristics
and monthly distribution of the rain events. The distribution of rain in 1995
and 2001 could be considered as typical of the Mediterranean region
climatology, with precipitation concentrated mainly in SeptemberNovember and March-May, and long dry periods at other times (Perez
Cueva, 1994). The number of erosive rain events increased in the studied
period, from 38 episodes in 2000 up to 49 in 2002. In general, it is in the
months following the dry season where the larger intensity rains are
concentrated. This tendency strongly changed during 2000 (Figure 2) in
INPACT OF FOREST FIRES ON HYDROLOGICAL PROPERTIES
J.L. RUBIO
495
oxidation with
determined by
iethod (Black,
with 2N KC1
dahl automatic
:al conductivi~
nethod (1964).
ig to the Olsen
:ed using a wet)f signification
used to detect
e treatments.
which after the dry season, 64.10% of the total annual rainfall was collected
in October only (266.60 mm). In this month, 132.8 mm corresponds to one
single event occurred on day 23rd.
600
500
/.
I
400
w
3
300
200
1o0
o
/*it%,.
"S,
Cp"" p
-
1995
its with runoff
(130) ~f 9.38
'acteristics and
ars (Figure 2).
trend of the
imong those it
rhich occurred
cy of recoveq
fall. However,
.50 mm, a net
linfa11 of only
characteristics
~frain in 1995
ranean region
i September' times (Perez
in the studied
-al, it is in the
sity rains are
(Figure 2) in
=
$ 3
--
2000
"""+'
---
ce9~e* vei
2601
oG\',vei*o*"
*vei
.&vef
-2002
Fzgure 2 Curnulative monthly distribution of ram durmg 1995 and 2000-2002
2002 also marked a change in the distribution, volume and duration
of rains compared to previous years. In this year the highest values of
precipitation occurred from April until September, during the summer
period, which is usually affected by water deficit, 224.4 mm were collected.
During the years 2001 and 2002, an increase in the number of
erosive events compared to previous years was observed, with 8 erosive rain
events of 45 rain episodes and 10 of 49, respectively. In 2000, only 6
erosive rains out of 38 events were registered. It is similar to those which
occurred in 1995, but with greater volume and duration of the erosive rains.
The characteristics of the erosive rain events which occurred during the
studied period can be observed in Figure 3.
events, r
surpassec
with an a
facts wei
whose m
2001. In
their du
favoured
the soil 1
samples
Figure 3. Erosive rain events occurred in 1995 (after the fue) and in the period 2000-2002
(A), and duration of these rain events (in minutes) (B)
During 2002 the erosive rains suffered a clear decrease in their
intensity, with a maximum value of 8.6 mm h-' whereas on 1995, 2000 and
2001 were of 35.36 mm h-', 28.6 mm h-' and 30 mm h", respectively. These
l
IMPACT OF FOREST FIRES ON HYDROLOGICAL PROPERTIES
497
events, not surpassing 29 mm. But during 2002, four erosive events
surpassed 35 mm, one of them reaching 96.5 mm of the total rains (15-IV),
with an average intensity that did not exceed 5.63 mm h-' (Figure 3). These
facts were accompanied in 2002 by an increase in the duration of rains,
whose minimum values (1 85 min) have duplicated those for 1995,2000 and
2001. In this way, 2002 showed an increase in the number of rains and in
their duration but with a clear diminution of intensities. These facts
favoured a decrease in the number of erosive rains and the maintenance of
the soil humidity conditions along the year.
Runoff processes
In Table 1 are reported some soil characteristics analyzed in
sarnples taken in 2002.
Table l. Mean values of some soil properties on each fue treatment in 2002
Soil Property
:period
Fire Treatment
High
Moderate
Control
PH
7.88
7.88
7.84
E.C. (dS m-')
0.64
0.65
0.68
A.S. (%)
32.96
33.47
32.03
O.M.(%)
7.92
7.88
7.15
Total-N (%)
0.364
0.356
0.361
NH; -N (mg 100g-')
0.776
0.844
0.888
NO; -N (m g 100 g-')
0.875
0.787
0.536
C.E.C. (cmol,kg")
27.453
26.171
26.45
Na' (%)
0.400
0.386
0.322
K+(%)
3.182
3.050
3.424
2000-2002
)
lecrease in their
1995,2000 and
pectively. These
reduced volume
E.C: electrical conductivity; A.S.: Aggregate stability; O.M.: organic matter;
Total-N: total Nitrogen; NH4f-N: ammonium Nibogen; NO3-- N: nitrate
Nitrogen; C.E.C.: cation exchange capacity. Na, K, Mg and Ca: exchangeable
cations as percentage of the C.E.C.
As it has been observed, the characteristics of rains during 2002,
could have favoured the upholding of a certain degree of soil humidis
throughout the year, without a real period of surnmer water deficit. This
situation could facilitate runoff generation but in tum, rnaintainiq the soil
in a state of better muffling of the impact of rain drops. This situation, with
moderate/low rain intensities, did not favour the generation of aggressive
runoffs and the removal of soil particles. So, for the control soil and burned
ones, the previous highest water content and the low rainfall intensities
during the different seasons, could have improved the infiltration rates and
diminished runoff yields and water repellency (Benavides-Solorio and Mc
Donald, 2001). Robichaud (2000), observed that the hydrophobic conditions
that appeared in the soil surface after fire impact, vary as the soil profile
becomes wetted and eventually responding as a normal infiltrating soil. De
Bano (198 l), note that infiltration increases with time because the
hydrophobic substances responsible for the water repellence are slightl~
water soluble and dissolve slowly, thereby increasing soil wettability. This
fact was confirmed by observations in the field and by the data of runoff
generation dunng the year 2002 (Figure 4).
1995
2000
Fire Treatments
2001
2002
EBHigh UModerate UControl
d
Figure 4. Total values of runoff yield (L m-') corresponding to different treatments during
1995 (after fire) and the 2000-2002 period
1 1
IMI'ACT OF FOREST FIRES ON HYDROLOGICAL PROPERTIES
This great variability in the distribution and characteristics of rains
observed during the period of 2000-2002, has been clearly reflected in the
!general hydrological behaviour and also in the response of the plots
corresponding to the two fue treatments in relation to the incidence of the
water erosion processes. These plots showed significant differences in
respect to the Control plots on their hydrological values during this period.
This fact is reflected in an average value of runoff production in the Control
plots of 76.68 % less than the bumed ones (Figure 4) for the whole period.
However, these differences slightly diminished in 2002 due to the change in
rain characteristics, with a clear reduction in their intensity and an increase
in their duration, but continue being significant (Figure 3 and 4).
Differences observed on the fire treatments in respect to the Control plots
are similar to that observed in 1995 after the fire (84.10%). Only in 2002,
these differences diminished around 14 % respect to 1995.
Between fire treatments, the differences on runoff production have
clearly decreased fiom 14.60% more runoff for plots affected by High
intensity fire in 1995 to 2.73 % in 2000. Actually, the plots bumed with
Moderate intensity present the highest rates of runoff yield. This reflects an
equilibrium of soil conditions and the regeneration of the vegetation cover.
Infiltration rates do not give significant differences between
treatments during the studied period (Figure 5A), which are usually lower
than 1%. Plots not affected by fire always give the maximum values. In
1995 the differences observed between fire treatrnents do not reach 1%, but
in regard to the Control plots, these differences were an average of 5.39%.
In the year 2002, the differences between burned plots and Control plots
were reduced to 1.18 %. It was partly due to the softer rainfall regime
during this year and to an improvement in the vegetation and soil
characteristics.
The data of runoff coefficient (Figure 5B) show a slight decrease in
the differences between the control plots and those affected by fire. As an
average, in 1995 control plots give 79.62% lower runoff coefficients than
the bumed plots, but fiom 2000 until 2002 this difference was gradually
reduced to 73.18 %. Between fire treatments, there were important changes.
In 1995, plots bumed with Moderate intensity presented 12.02 % less runoff
coefficient, meanwhile in 2002 the tendency changed, plots bumed with
Moderate intensity presented the highest values of it.
Fire Treatments
mHigh OModerate OControl
.....................
------
V
1
---
/
Figure 5. Average values of infiltration rate (A) and m o f f coefficient (B) correspondents to
the different treatments during 1995 (afier fue) and 2000-2002 period
The trend observed on both parameters could indicate a slight
amelioration of the hydrological conditions of these soils in respect to those
immediately after the fire impact. Doerr, ef al. (1996), found that the
hydrophobic character of burned and unburned soils, when dry, tend to
disappear as the soil becomes wetter in low intensity events. However,
changes in the rain regime can produce a response similar to that stated in
1995, a clear example is shown by the data corresponding to 2001. The
strong reduction of rainfall volume occurred in that year, with a decrease in
IMPACT OF FOREST FIRES ON HYDROLOGICAL PROPERTIES
the duration and with an increase in the intensity of rains, which was
accompanied by a concentration of the most aggressive events after a long
dry period (April to September), producing a faster and greater runoff
generation, probably due to the incidence of water repellency in the soil and
less receptive soil hydrological conditions. In addition to that, the
degradation of vegetation cover and the changes in soil characteristics
because of fire, could produce an increase in the hydrological response,
especially in the High fire intensity treatment. These effects are in
accordance to the observations of different authors in similar shrubland
environments @e Bano, et al., 1998; Johansen, et al., 2001).
Hydrological evolution
Data obtained on soil Water Retention Capacity (WRC), on samples
of 1995 and 2002 (Table 2), for a matric potential of -10 KF'a (pF 2), show
significant differences between High fire intensity plots and those
corresponding to the Control treatment irnmediately after the fire
experiment (1995). This data is in agreement with the observations of Malik
(1984) on the WRC on superficial soils samples. Just irnmediately after fne,
WRC is greater on bumed areas than on natural soil. This could be
explained by considering that the ash particles in the bumed plot clog the
soil pores in the upper soil layer, thus the density of the larger pores is
reduced and as a consequence, this reduces the rate of water percolation
throughout the soil and increases WRC. This parameter shows its greatest
values in the High intensity treatment, 26.6 % being higher than those of the
Moderate intensity treatment, and 3 1.11 % greater than the control one
Seven years later, in 2002, control plots present higher values of
WRC than bumed ones. Boix-Fayos (1997), found a similar tendency in a
Mediterranean soil that suffered the impact of fire three years before, in
respect to unbumed soil.
The fact that soil affected by High intensity fire, presents greatest
values of WRC than unburned soil irnmediately after buming, could
indicate that changes in physical properties in soil surface appeared
(Andreu, 2001). These changes could be produced in particle-size
distribution and aggregation, by the re-aggregation of clay size particles into
sand-size ones (Giovannini, 1994). This effect could favour higher water
holding at low pF values (Table 2). So, there is an important quantity of
water retained in soil at low pF values, probably stored in the gaps
:rease in
I
O. GONZALEZ, V. ANDREU, E. GIMENO-GARCIA & J.L. RUBIO
502
generated by the re-aggregation of clay particles into sand size ones. When
the WRC is calculated on a basis of a matric potential of -33 KPa, the
possible effect of the water held by the sand-sized particles is eliminated,
and the differences between fire treatments disappear (Table 2B). The fire
-10 KPa and -33 KPa, the held water content is 75% and 55% higher for
High and Moderate intensity treatment in respect to the Control values
(Table 2). Then, there is an important amount of water held in soil at low pF
values for the High intensity treatrnent. Guber et al. (2003) classi@ing
aggregates by size using the average water content at -10 KPa, -33 KPa and
-1500 KPa, found that larger aggregates present the greatest variation on
water content and the greatest values on this parameter.
Table 2. Values of water retention capacity (WRC, cm3 cm") calculated at matric potentials
of -10 KPa and -1500 KPa, and -33KPa and -1500KPa, for the different fue treatments
immediately afier the fue experiment (1995) and in 2002 period
WRC 10 KPa
WRC 33 Kpa
1995
2002
1995
2002
High
0.225 a
0.182 ab
0.105 a
0.131 a
TREATMENTS
Moderate
0.165 b
0.178 a
0.135 a
0.136 a
1
1
Control
0.155 b
0.202 b
0.100 a
0.170 b
Values not sharing the same letter in fíles indicate significant differences for the different
treatment according to the Tukey's test (P
IMPACT OF FOREST FIRES ON
HYDROLOGICAL PROPERTIES OF A TYPICAL
MEDITERRANEAN FOREST SOIL
O. González, V. Andreu, E. Gimeno-García & J. L. Rubio
Centro de Investigaciones sobre Desertificación-CIDE (CSIC-Universitat de ValenciaGeneralitat Valenciana). Camí de la Marjal, s/n. 46470 - Alba1 (Valencia, SPAm).
Fax: +34 96 127 09 67
Tel.: +34 96 122 05 40
E-mail: [email protected]
ABSTRACT
Fue is one of the most important determining factors in the evolution of
Mediterranean ecosystems. Its effects on soil are plural and diverse, acting on structure,
chemical and physical properties, biota etc. Among them, the induced variations on soil
water dynamics are of key importance for the recovery of the entire ecosystem and in the soil
response to erosion processes. Forest fues are also a factor, triggenng the risk of
desertification. The aim of this study is to assess the impact effect of different fue intensities
on soil hydrology and on runoff generation.
This work was developed in the Permanent Experimental Field Station of La
Concordia with nine plots (4 x 20 m), installed in a calcareous hillside, representative of
Mediterranean shrubland areas. Experimental fires, of two intensity levels were carried out.
Three plots were burned reaching high fue intensity and three other plots were burned with
moderate intensity and the remaining plots were left undisturbed as control. Soil water
content, water retention capacity and pF curves were measured together with runoff
generation dynamics, in the different plots. The intrinsic characteristics of each rain event
occurred up until ayear after the fues and runoff generated on them was monitored.
The data obtained was compared with the state of the plots after seven years,
during the 2000-2002 period. In this period 24 rain events with runoff generation occurred,
with average rainfall intensities (130)around lOmm h". Both fire treatrnents show significant
differences with respect to the Control plots, which are reflected in a value of runoff
production of 76.84% as an average, less than the burned plots. Between fue treatments, the
plots that bumed with high intensity, show the highest values of runoff yield. However,
infiltration rates do not give significant differences between fue treatments.
In the same way, plots that suffer a high intensity f r e show greater values (22.50
cm3 cm'3) on water retention capacity than the other treatments, giving significant differences
with the Moderate intensity plots and Control plots. Differences on this parameter between
plots bumed with Moderate intensity and the Control plots were obsewed but they were not
489
Desertijication in the Mediterrarrean Region: a Securiw Issue, 489-507.
Williarn C. Keprier, Jose L. Rubio, DavidA. Mouat & Fausto Pedrauirri, e&.
O 2006 Spriirger. Printed in tlre Netlrerlands.
O. GONZALEZ, V. ANDREU, E. GIMENO-GARCIA & J.L. RUBIO
statistically significant. The obtained pF curves show the same tendencies, being the greatest
water content retained at pF of 2 in the burned plots and in the Control plots at pF of 4.2.
The effect of f ~ degrading
e
the vegetation cover and by means of the temperature
irnpact on soil, produce changes in its sttuctural characteristics and porosity, affecting soil
water distribution and the effective response on water erosion processes.
Keywords: water retention capacity, pF curve, hydrology, experimental f ~ eMediterranean,
,
desertification.
INTRODUCTION
Wildfires are known to have occurred in the Mediterranean region
through historic and prehistoric times (Naveh, 1990). In the last few
decades, fires of human origin have substantially increased not only in their
nurnber but also in their fi-equency, which have produced reduction in the
recovery periods and could drastically change or even eliminate the
vegetation cover, changing and degrading the ecosystem characteristics
(Andreu, 2002).
Actually, different studies indicate that the tendencies of fires in the
Mediterranean region are to decrease the affected area but increase in
nurnber (European Cornmission, 2002). It could mean that many
regeneration zones could be affected by fire again, breaking its possibilities
of resilience and becoming in regression by alteration of its physical,
chemical and biological functions and processes. Forest fires also increase
desertification risk in the Mediterranean, particularly when there are some
critica1 circumstances, such as: fire on very step slopes and shallow soils
underlying consolidated parent material; fire affecting vulnerable soils or
soil with poor conditions for its re-vegetation; the effects of repeated fires
on the same area and also the incidente of torrential rain in recently bumed
areas.
In the Mediterranean area, as a result of the high rainfall intensities
concentrated after the dry surnrner period, where the fires occur, soil
infiltration and even water storage capacity, mainly in shallow soils, is
easily exceeded, generating important soil losses and high energy runoffs as
a consequence of the relief features (Rubio and Calvo, 1996) that can reach
as much as one to three orders of magnitude (De Bano, 2000; Robichaud, et
al., 2000). This is usually due to soil changes induced by fire, which
depending on its severity or intensity, could be beneficia1 or deleterious to
the ecosystem (Neary, et al., 1999). Between these changes, loss of
vegetation cover and soil structure alterations are factors that produce
amplified hydrological responses (De Bano, 1998).
IMPACT OF FOREST FIRES ON HYDROLOGICAL PROPERTIES
ncrease
e some
Generally, the high intensity fires and the greater structural
breakdown of aggregates, produce formation of microaggregates and
liberation of soil particles, easily removed by runoff (Prosser and Williarns,
1998; Benavides-Solorio and McDonald, 2001). This effect will favour soil
sealing, crust formation, decreasing soil water retention capacity and
reduction in pore size characteristics that incise directly in a reduction on
infiltration and drainage increasing soil erodibility (Le Bissonais, 1996;
Teman and Neller, 1999; Andreu, et al., 200 1).
Post-fire conditions of the surface horizons are of key importance
because they determine the response of soil exposed to raindrop splash,
overland flow and the development of water repellent soil conditions (De
Bano, 1981). These changes in soil characteristics and the degradation of
the vegetation cover and litter layers can lead to very large increases in soil
erosion (Benavides-Solorio and McDonald, 2001).
These facts suggest that one important way to carefully study the
effects of fire on the soil ecosystem is through the performance of fires in
experimental plots (Giovannini and Luchesi, 1997). Then it is possible to
know and measure soil conditions before, during and after the fire
experiment and to improve the knowledge about the hydrology of the zone
affected by fire which could reach different intensities.
In this context, the aim of this study is to evaluate the evolution of
the hydrological characteristics of a typical Mediterranean forest soil
affected by different fire intensities in 1995 and compare it with its state in
2002. Changes in the response of these soils on runoff generation in each
erosive rain event of the time period considered have been also studied.
burned
MATERIALS AND METHODS
ansities
U, soil
oils, is
ioffs as
n reach
iaud, et
which
,ious to
oss of
iroduce
Study Area
This work was developed in the Permanent Experimental Field
Station of La Concordia (Lliria, SPAIN), at 50 lan NW of Valencia city. It
is 575 m above sea leve1 (Figure l), on a forested hillside facing South
South East, with a sclerophyllous shmb cover, regenerated after a previous
wildfire occurred in 1978. The most abundant species include Rosmarinus
ofJicinalis, Ulex pawzjlorus, Quercus coccifera, Rhamnus lycioides, Stipa
tenacissima, Globularia alypum, Cistus clusii and Thymus vulgaris.
O. GONZALEZ, V. ANDREU, E. GIMENO-GARCÍA & J.L. RUBIO
IMPACT
collection.
A rs
three plots c
High and h
different am
', respectivc
Control trea
fire intensit
similar to th
Clin
rainfall evei
transmissior
Soil Anal:
Figure 1. Geographical location of the Permanent Experimental Field Station of
La Concordia (Casinos, Spain)
Climatically the area belongs to the dry ombroclimate of the lower
mesomediterranean belt, according to Thornthwaite's classification. The
average annual precipitation is around 400 rnm, with two maxima, autumn
and spring, and a dry period from June to September with mean monthly
temperatures ranging fiom 13.3"C in January to 25.8OC in August.
The soil is a Rendzic Leptosol (FAO-UNESCO, 1988), or Calcic
Xerochrept type accordiig to Soil Taxonomy classification (Soil Survey Staff,
1990), developed on Jurassic limestone. This soil has a variable depth,
always less than 40 cm, abundant stoniness (E40%) and good drainage.
Experimental Set-Up
The station consists of a set of nine erosion plots, each 4 m wide x
20 m long, with similar characteristics such as soil morphology, slope
gradient, rock outcrops and vegetation cover. The selection of each plot
location was made after intensive surveys of the vegetation, soil and the
morphology pattern, based on across slope transects every two meters.
Plots are oriented parallel to the slope and bounded by bricks. At the
foot of each plot, a 2 m wide collector ran into a 1500 1tank to record al1 the
runoff and sediment produced during each rainfall event. Inside them there
Soil
and immedi
screened to
until their ar
Soil
-33, -300,
membrane I
was calcull
Cameron (1
Wh
gravimetric
density of
percentage.
Wa
with field 1
The pF cur
Ru
al1 plots, d
2002.
IMPACT OF FOREST FIRES ON HYDROLOGICAL PROPERTIES
is a 30 1 tank to concentrate the sediments produced, facilitating its
A random design of three different fue intensity treatments (with
three plots each) was used. Two sets of three plots were bumed reaching
High and Moderate fire intensities, which was achieved by addition of
different amounts of fue1 load to the plots of each treatment, 40 and 20 t harespectively. The remaining three plots were unbumt to be used as a
Control treatment. The necessary quantity of dry biomass to obtain the two
fire intensities was added using vegetation (fiom the surrounding area)
similar to that present in the plots.
Climatic parameters and the intrinsic characteristics of the different
rainfall events were monitored by a logging system of sensors with GSM
transmission of data, placed close to the plots.
',
Soil Analysis and Measurements
Soil samples were taken fiom the top 5 cm of the soil surface before
and immediately after fue, and in the summer of 2002, they were air-dried,
screened to remove the fiaction >2 rnm diameter and stored in plastic boxes
until their analysis.
Soil Water Content (SWC) was calculated for the potentials: 0, -10,
-33, -300, and -1500 KPa, or pF 1, 2, 2.5, 3.5, 4.2, using the pressure
membrane method (Richards, 1947). Soil Water Retention Capacity (WRC)
was calculated for each soil sample using the equations of McLaren and
Cameron (1996).
(a) WRC=(
n wide x
Y, slope
:ach plot
and the
;s. At the
rd al1 the
em there
e10- e1500 )* pb
m) WRC=( e33- elsoO)* pb
Where WRC is soil Water Retention Capacity, 0 10, 0 33, 0 1500 are
gravimetric water volume at - 10 KPa, -33 KPa, and -1500 KPa, and pb bulk
density of soil samples. The results were obtained in volumetric units in
percentage.
Water Retention Capacity with field capacity at -10 KPa (a) and
with field capacity at -33 KPa (b), were calculated for the different plots.
The pF curves were also determined.
Runoff generation dynamics were monitored for each rain event in
al1 plots, during one year after the 1995 fires and in the period of 20002002.
Soil organic matter content was determined by oxidation with
potassium dichromate (Jackson, 1958). Total Nitrogen was determined by
micro-Kjeldahl automatic analyser using the Bremmer method (Black,
1965). Ammonium and nitrate Nitrogen were extracted with 2N KC1
solution and determined by stearn distillation by micro-Kjeldahl automatic
analyzer using the Bremmer method (Black, 1965). Electrical conductivity
was measured in soil saturation extracts by the Richards method (1964).
Available Phosphorous was measured by colorimetry according to the Olsen
and Dean method (Black, 1965). Aggregate stability was assessed using a wetsieving procedure (Pnmo and Carrasco, 1973).
Standard statistical analyses were applied at 95% of signification
level. Analysis of variance and Tukey's test at a=0.05 were used to detect
differences in WRC and pF values, according the different fire treatments.
MAIN RESULTS AND DISCUSSION
Rainfall trends
During the studied period (2000-2002), 24 rain events with runoff
generation were monitored, showing average rainfall intensity (130) of 9.38
mm h-l. The rain behaviour in year 2002, on quantity, characteristics and
distribution, showed a change compared to the previous years (Figure 2).
Since the Experimental Station started in 1995, the trend of the
precipitations in the area has showed substantial variations, among those it
is necessary to highlight the intense period of drought which occurred
during 1998 and 1999. Year 2000 observed a certain tendency of recovery
in the rain regime, reaching 415.90 mm of the annual rainfall. However,
although the total annual rainfall in year 2002 reached 521.50 mrn, a net
decrease of precipitation occurred in 2001 with a total rainfall of only
267.40 mm.
It is also possible to appreciate this variability in the characteristics
and monthly distribution of the rain events. The distribution of rain in 1995
and 2001 could be considered as typical of the Mediterranean region
climatology, with precipitation concentrated mainly in SeptemberNovember and March-May, and long dry periods at other times (Perez
Cueva, 1994). The number of erosive rain events increased in the studied
period, from 38 episodes in 2000 up to 49 in 2002. In general, it is in the
months following the dry season where the larger intensity rains are
concentrated. This tendency strongly changed during 2000 (Figure 2) in
INPACT OF FOREST FIRES ON HYDROLOGICAL PROPERTIES
J.L. RUBIO
495
oxidation with
determined by
iethod (Black,
with 2N KC1
dahl automatic
:al conductivi~
nethod (1964).
ig to the Olsen
:ed using a wet)f signification
used to detect
e treatments.
which after the dry season, 64.10% of the total annual rainfall was collected
in October only (266.60 mm). In this month, 132.8 mm corresponds to one
single event occurred on day 23rd.
600
500
/.
I
400
w
3
300
200
1o0
o
/*it%,.
"S,
Cp"" p
-
1995
its with runoff
(130) ~f 9.38
'acteristics and
ars (Figure 2).
trend of the
imong those it
rhich occurred
cy of recoveq
fall. However,
.50 mm, a net
linfa11 of only
characteristics
~frain in 1995
ranean region
i September' times (Perez
in the studied
-al, it is in the
sity rains are
(Figure 2) in
=
$ 3
--
2000
"""+'
---
ce9~e* vei
2601
oG\',vei*o*"
*vei
.&vef
-2002
Fzgure 2 Curnulative monthly distribution of ram durmg 1995 and 2000-2002
2002 also marked a change in the distribution, volume and duration
of rains compared to previous years. In this year the highest values of
precipitation occurred from April until September, during the summer
period, which is usually affected by water deficit, 224.4 mm were collected.
During the years 2001 and 2002, an increase in the number of
erosive events compared to previous years was observed, with 8 erosive rain
events of 45 rain episodes and 10 of 49, respectively. In 2000, only 6
erosive rains out of 38 events were registered. It is similar to those which
occurred in 1995, but with greater volume and duration of the erosive rains.
The characteristics of the erosive rain events which occurred during the
studied period can be observed in Figure 3.
events, r
surpassec
with an a
facts wei
whose m
2001. In
their du
favoured
the soil 1
samples
Figure 3. Erosive rain events occurred in 1995 (after the fue) and in the period 2000-2002
(A), and duration of these rain events (in minutes) (B)
During 2002 the erosive rains suffered a clear decrease in their
intensity, with a maximum value of 8.6 mm h-' whereas on 1995, 2000 and
2001 were of 35.36 mm h-', 28.6 mm h-' and 30 mm h", respectively. These
l
IMPACT OF FOREST FIRES ON HYDROLOGICAL PROPERTIES
497
events, not surpassing 29 mm. But during 2002, four erosive events
surpassed 35 mm, one of them reaching 96.5 mm of the total rains (15-IV),
with an average intensity that did not exceed 5.63 mm h-' (Figure 3). These
facts were accompanied in 2002 by an increase in the duration of rains,
whose minimum values (1 85 min) have duplicated those for 1995,2000 and
2001. In this way, 2002 showed an increase in the number of rains and in
their duration but with a clear diminution of intensities. These facts
favoured a decrease in the number of erosive rains and the maintenance of
the soil humidity conditions along the year.
Runoff processes
In Table 1 are reported some soil characteristics analyzed in
sarnples taken in 2002.
Table l. Mean values of some soil properties on each fue treatment in 2002
Soil Property
:period
Fire Treatment
High
Moderate
Control
PH
7.88
7.88
7.84
E.C. (dS m-')
0.64
0.65
0.68
A.S. (%)
32.96
33.47
32.03
O.M.(%)
7.92
7.88
7.15
Total-N (%)
0.364
0.356
0.361
NH; -N (mg 100g-')
0.776
0.844
0.888
NO; -N (m g 100 g-')
0.875
0.787
0.536
C.E.C. (cmol,kg")
27.453
26.171
26.45
Na' (%)
0.400
0.386
0.322
K+(%)
3.182
3.050
3.424
2000-2002
)
lecrease in their
1995,2000 and
pectively. These
reduced volume
E.C: electrical conductivity; A.S.: Aggregate stability; O.M.: organic matter;
Total-N: total Nitrogen; NH4f-N: ammonium Nibogen; NO3-- N: nitrate
Nitrogen; C.E.C.: cation exchange capacity. Na, K, Mg and Ca: exchangeable
cations as percentage of the C.E.C.
As it has been observed, the characteristics of rains during 2002,
could have favoured the upholding of a certain degree of soil humidis
throughout the year, without a real period of surnmer water deficit. This
situation could facilitate runoff generation but in tum, rnaintainiq the soil
in a state of better muffling of the impact of rain drops. This situation, with
moderate/low rain intensities, did not favour the generation of aggressive
runoffs and the removal of soil particles. So, for the control soil and burned
ones, the previous highest water content and the low rainfall intensities
during the different seasons, could have improved the infiltration rates and
diminished runoff yields and water repellency (Benavides-Solorio and Mc
Donald, 2001). Robichaud (2000), observed that the hydrophobic conditions
that appeared in the soil surface after fire impact, vary as the soil profile
becomes wetted and eventually responding as a normal infiltrating soil. De
Bano (198 l), note that infiltration increases with time because the
hydrophobic substances responsible for the water repellence are slightl~
water soluble and dissolve slowly, thereby increasing soil wettability. This
fact was confirmed by observations in the field and by the data of runoff
generation dunng the year 2002 (Figure 4).
1995
2000
Fire Treatments
2001
2002
EBHigh UModerate UControl
d
Figure 4. Total values of runoff yield (L m-') corresponding to different treatments during
1995 (after fire) and the 2000-2002 period
1 1
IMI'ACT OF FOREST FIRES ON HYDROLOGICAL PROPERTIES
This great variability in the distribution and characteristics of rains
observed during the period of 2000-2002, has been clearly reflected in the
!general hydrological behaviour and also in the response of the plots
corresponding to the two fue treatments in relation to the incidence of the
water erosion processes. These plots showed significant differences in
respect to the Control plots on their hydrological values during this period.
This fact is reflected in an average value of runoff production in the Control
plots of 76.68 % less than the bumed ones (Figure 4) for the whole period.
However, these differences slightly diminished in 2002 due to the change in
rain characteristics, with a clear reduction in their intensity and an increase
in their duration, but continue being significant (Figure 3 and 4).
Differences observed on the fire treatments in respect to the Control plots
are similar to that observed in 1995 after the fire (84.10%). Only in 2002,
these differences diminished around 14 % respect to 1995.
Between fire treatments, the differences on runoff production have
clearly decreased fiom 14.60% more runoff for plots affected by High
intensity fire in 1995 to 2.73 % in 2000. Actually, the plots bumed with
Moderate intensity present the highest rates of runoff yield. This reflects an
equilibrium of soil conditions and the regeneration of the vegetation cover.
Infiltration rates do not give significant differences between
treatments during the studied period (Figure 5A), which are usually lower
than 1%. Plots not affected by fire always give the maximum values. In
1995 the differences observed between fire treatrnents do not reach 1%, but
in regard to the Control plots, these differences were an average of 5.39%.
In the year 2002, the differences between burned plots and Control plots
were reduced to 1.18 %. It was partly due to the softer rainfall regime
during this year and to an improvement in the vegetation and soil
characteristics.
The data of runoff coefficient (Figure 5B) show a slight decrease in
the differences between the control plots and those affected by fire. As an
average, in 1995 control plots give 79.62% lower runoff coefficients than
the bumed plots, but fiom 2000 until 2002 this difference was gradually
reduced to 73.18 %. Between fire treatments, there were important changes.
In 1995, plots bumed with Moderate intensity presented 12.02 % less runoff
coefficient, meanwhile in 2002 the tendency changed, plots bumed with
Moderate intensity presented the highest values of it.
Fire Treatments
mHigh OModerate OControl
.....................
------
V
1
---
/
Figure 5. Average values of infiltration rate (A) and m o f f coefficient (B) correspondents to
the different treatments during 1995 (afier fue) and 2000-2002 period
The trend observed on both parameters could indicate a slight
amelioration of the hydrological conditions of these soils in respect to those
immediately after the fire impact. Doerr, ef al. (1996), found that the
hydrophobic character of burned and unburned soils, when dry, tend to
disappear as the soil becomes wetter in low intensity events. However,
changes in the rain regime can produce a response similar to that stated in
1995, a clear example is shown by the data corresponding to 2001. The
strong reduction of rainfall volume occurred in that year, with a decrease in
IMPACT OF FOREST FIRES ON HYDROLOGICAL PROPERTIES
the duration and with an increase in the intensity of rains, which was
accompanied by a concentration of the most aggressive events after a long
dry period (April to September), producing a faster and greater runoff
generation, probably due to the incidence of water repellency in the soil and
less receptive soil hydrological conditions. In addition to that, the
degradation of vegetation cover and the changes in soil characteristics
because of fire, could produce an increase in the hydrological response,
especially in the High fire intensity treatment. These effects are in
accordance to the observations of different authors in similar shrubland
environments @e Bano, et al., 1998; Johansen, et al., 2001).
Hydrological evolution
Data obtained on soil Water Retention Capacity (WRC), on samples
of 1995 and 2002 (Table 2), for a matric potential of -10 KF'a (pF 2), show
significant differences between High fire intensity plots and those
corresponding to the Control treatment irnmediately after the fire
experiment (1995). This data is in agreement with the observations of Malik
(1984) on the WRC on superficial soils samples. Just irnmediately after fne,
WRC is greater on bumed areas than on natural soil. This could be
explained by considering that the ash particles in the bumed plot clog the
soil pores in the upper soil layer, thus the density of the larger pores is
reduced and as a consequence, this reduces the rate of water percolation
throughout the soil and increases WRC. This parameter shows its greatest
values in the High intensity treatment, 26.6 % being higher than those of the
Moderate intensity treatment, and 3 1.11 % greater than the control one
Seven years later, in 2002, control plots present higher values of
WRC than bumed ones. Boix-Fayos (1997), found a similar tendency in a
Mediterranean soil that suffered the impact of fire three years before, in
respect to unbumed soil.
The fact that soil affected by High intensity fire, presents greatest
values of WRC than unburned soil irnmediately after buming, could
indicate that changes in physical properties in soil surface appeared
(Andreu, 2001). These changes could be produced in particle-size
distribution and aggregation, by the re-aggregation of clay size particles into
sand-size ones (Giovannini, 1994). This effect could favour higher water
holding at low pF values (Table 2). So, there is an important quantity of
water retained in soil at low pF values, probably stored in the gaps
:rease in
I
O. GONZALEZ, V. ANDREU, E. GIMENO-GARCIA & J.L. RUBIO
502
generated by the re-aggregation of clay particles into sand size ones. When
the WRC is calculated on a basis of a matric potential of -33 KPa, the
possible effect of the water held by the sand-sized particles is eliminated,
and the differences between fire treatments disappear (Table 2B). The fire
-10 KPa and -33 KPa, the held water content is 75% and 55% higher for
High and Moderate intensity treatment in respect to the Control values
(Table 2). Then, there is an important amount of water held in soil at low pF
values for the High intensity treatrnent. Guber et al. (2003) classi@ing
aggregates by size using the average water content at -10 KPa, -33 KPa and
-1500 KPa, found that larger aggregates present the greatest variation on
water content and the greatest values on this parameter.
Table 2. Values of water retention capacity (WRC, cm3 cm") calculated at matric potentials
of -10 KPa and -1500 KPa, and -33KPa and -1500KPa, for the different fue treatments
immediately afier the fue experiment (1995) and in 2002 period
WRC 10 KPa
WRC 33 Kpa
1995
2002
1995
2002
High
0.225 a
0.182 ab
0.105 a
0.131 a
TREATMENTS
Moderate
0.165 b
0.178 a
0.135 a
0.136 a
1
1
Control
0.155 b
0.202 b
0.100 a
0.170 b
Values not sharing the same letter in fíles indicate significant differences for the different
treatment according to the Tukey's test (P