Aridification determines changes in forest growth in Pinus halepensis forests
under semiarid Mediterranean climate conditions
Sergio M. Vicente-Serrano1,*, Teodoro Lasanta1 and Carlos Gracia2
1
Instituto Pirenaico de Ecología, CSIC (Spanish Research Council), Campus de Aula Dei, P.O. Box
202, Zaragoza 50080, Spain
2
Departament d'Ecologia, Facultat de Biologia, Universitat de Barcelona and CREAF (Centre de
Recerca Ecològica i Aplicacions Forestals). Universitat Autònoma de Barcelona. Barcelona, Spain.
* svicen@ipe.csic.es
Abstract. Using a set of 16 Landsat TM and Landsat ETM images from 1984 to 2006, we tested
whether climate trends in the last three decades differentially controlled the vegetal activity of eight
Pinus halepensis forests located across a marked bioclimatic gradient. Our results show spatial
differences in trends in the normalized difference vegetation index (NDVI) between 1984 and 2006,
which were highly related to the spatial distribution of aridity. There was a strong correlation
between the change in the NDVI values and the climatic water balance in each forest (R = 0.86, p =
0.006). A large increment of the NDVI was observed in forests located in the most humid areas,
whereas those located in the most arid areas showed a slight decrease in the NDVI. We used a
forest growth simulation model (GOTILWA+) to estimate the leaf area index (LAI) in each of the
eight analyzed forests, and found similar results to those obtained for the NDVI. Significant
correlation was found between the spatial pattern of NDVI and LAI trends between 1984 and 2006
(R = 0.82, p = 0.01). The climate evolution in the last four decades explains the observed and
modeled changes in the NDVI and the LAI. In agreement with other studies, we showed that the
general response of the forests to more favorable conditions (in terms of temperature increment and
CO2 fertilization) is increased leaf activity and biomass. Nevertheless, in forests located in the most
arid areas the positive trend observed in the potential evapotranspiration rates increased water
stress, and had a negative effect on forest growth. Given the future predictions of warming and
declining precipitation from global climate models for the Mediterranean region, an increase in
stress conditions affecting these forests is expected, which will affect their growth and survival,
mainly in the most arid areas.
Key words: forest activity, forest change, aridity, Pinus halepensis, NDVI, leaf area index,
Landsat, climatic change, potential evapotranspiration, climatic trends, Mediterranean.
Introduction
Forest biomass is one of the major reservoirs of organic carbon (Wofsy et al., 1993; Dixon et al.,
1994; Barford et al., 2001). Given the large dependence of carbon storage on the physiological
status and activity of forests, a priority in environmental science is to assess the potential impacts of
global change processes on the activity, health and growth of forest ecosystems (Saxe et al., 2001;
Nemani et al., 2003; Boisvenue and Running, 2006), as they could markedly affect future rates of
carbon storage.
1
Various studies have identified changes over recent decades in forest activity and growth at
continental (e.g. Zhou et al., 2001; Slayback et al., 2003; Delbart et al., 2006) and local scales (e.g.
Tardif et al., 2003; Andreu et al., 2007; Martínez-Villalta et al., 2008). Most of the changes have
been caused by human activities, particularly deforestation (Achard et al., 2002; De Fries et al.,
2002) and forest fires (Giglio et al., 2006; Riaño et al., 2007), but land marginalization and rural
abandonment have contributed to natural revegetation processes in some regions (Zahawi et al.,
1999; Vicente-Serrano et al., 2004; Sluiter and de Jong, 2007).
Some changes in vegetation cover are due to climate variability and change. At a global scale, net
primary production of forests increased 6% in the last two decades, mainly in tropical ecosystems
and driven by climatic factors including decreased cloud cover and the resulting increase in solar
radiation (Nemani et al., 2003). In temperature-limited ecosystems, temperature increase is
contributing to greater forest activity and growth. At high latitudes of the northern hemisphere, an
increase in the rate of forest activity that is directly related to temperature increase has been
reported (Myneni et al., 1998; Goetz et al., 2005), with implications for carbon sequestration by
forests (Pioao et al., 2008). Similarly, some studies have shown an increase during the 20th century
in the upper altitudinal limit of trees in alpine mountain areas, mainly as a consequence of increased
temperatures (e.g. Camarero and Gutierrez, 2004; Lenoir et al., 2008).
In regions where the productivity and activity of forests is water-limited, variations in precipitation
can markedly influence forest changes. Orwing and Abrams (1997) and Abrams et al. (1998)
analyzed the effect of drought on radial growth of different tree species in Pennsylvania, USA, and
showed that trees located in the most arid environments were more sensitive to droughts. In
semiarid and subhumid regions, where climate variability is very high, the occurrence of droughts
markedly reduces physiological processes, vegetation activity and leaf biomass (e.g. Borguetti et
al., 1998; Bréda et al., 2006; Vicente-Serrano, 2007), and wood production and carbon storage by
forests (e.g. Barber et al., 2000; Ogaya et al., 2003; Voronin et al., 2005).
2
Nevertheless, in water-limited environments including most of the Mediterranean region, changes
in precipitation may not be the main factor driving changes to forests; temperature may also play a
significant role, as it interacts with precipitation to determine water availability. Increased
temperature commonly has a positive effect on the activity and growth of trees, but the effect can be
negative if there is no corresponding increase in precipitation, as water stress may occur. In Spain,
Andreu et al. (2007) have shown a progressive increase in the effect of temperature on tree growth
during the second half of the 20th century and a decrease in the effect of precipitation. This suggests
that in this period there were more years in which climate conditions limited tree growth as a
consequence of induced temperature-related water stress. These results are consistent with those of
Martínez-Villalta et al. (2008) for northeast Spain, who showed that increased annual temperature
favored growth during humid years, and had the opposite effect in dry years. Some studies have
shown reduced forest growth in Mediterranean ecosystems located near the distribution limit for
some forest species. Sarris et al. (2007) reported a decline in the radial growth of Pinus cembra on
the island of Samos (Greece), associated with a decline in precipitation. In Spain, Jump et al. (2006)
found a decline since 1975 in Fagus sylvatica forests near the distribution limit for this species,
related to increasing temperature and greater water stress. In Italy, Maselli (2004) noted a decrease
of activity in coniferous and broadleaf forests between 1986 and 2000, associated with a decrease in
winter rainfall.
Nevertheless, the spatial pattern of forest growth and decline can be very complex. Large variations
in forest responses to climate are related to local climatic gradients. For example, Martínez-Villalta
et al. (2008) showed that, in the driest parts of northeast Spain, the effect of increased temperature
was always detrimental to growth, whereas the effect was positive in humid areas. Vicente-Serrano
et al. (2006) showed a direct spatial relationship between vegetation activity and aridity in broadleaf
and coniferous forests in Spain. Lloret et al. (2007) have also shown that the diversitystability
relationship between forest activity and drought in Spain follows a climatic gradient, with the driest
3
localities being more affected; this finding is in agreement with the results of Vicente-Serrano
(2007) in the center of the Ebro Valley (northeast Spain). The large spatial differences found at
local scales in the response of forests to drought conditions suggests the importance of analyzing
recent changes and trends in forest activity in relation to local climatic gradients, and the need to
consider forests near their limit of distribution, as the first impacts of climate change processes are
expected to be observed in ecotones (Neilson, 1993).
Although there have been some studies of trends in tree growth in European environments, based on
tree ring analysis in a range of environmental and climatic gradients (e.g. Mäkinen et al., 2002;
Dittmar et al., 2003; Martínez-Villalta et al. 2008), few studies have analyzed trends in other forest
variables related to photosynthetic activity and leaf biomass production. Studies of this nature are
very relevant to understanding the effects of global change on ecosystems, as there have been
reports of marked variability in roots, and above ground wood and green parts of trees, related to
changes in water availability (Lapenis et al., 2005). Changes in leaf biomass and leaf activity are
more difficult to identify than changes in tree growth, where tree ring analysis is commonly used.
However, while analysis of leaf-related processes poses some difficulties, remote sensing can be
used to measure the activity and trends in leaf biomass over large areas and varying climatic
gradients. Moreover, the availability of long time-series archives of satellite images provides the
opportunity for analysis of trends and changes over recent decades.
Few studies have analysed the possible evolution of the P. halepensis forest under future climate
projections. Gaucherel et al. (2008) have expected an increase of the P. halepensis growth in South
France during the twenty-first century as a response to CO2 increase. The results agree with the
outputs of a biogeochemistry model in Provence considering both temperature and precipitation
changes (Rathgeber et al., 2003). Nevertheless, there are no analyses focussed on the impacts of
climate change processes on the P. halepenesis forests located near the distribution limits. Sabaté et
4
al. (2002) suggested that in these areas the growth could be reduced as a consequence of water
constraints associated to precipitation decrease and temperature increase in Mediterranean areas.
In the present study, we analyzed trends in vegetation activity over a climate gradient (346626 mm
average annual precipitation) in the central Ebro Valley (northeast Spain), which is in the
northernmost semiarid region of Europe. We focused on Pinus halepensis, a forest species that is
highly adapted to aridity. It is able to withstand drought periods and soil drying because it rapidly
responds to decreasing predawn water potential by stomatal closure, which reduces transpiration
(Borguetti et al., 1998) and prevents severe tissue dehydration and foliage dieback (Chaves et al.,
1993). Consequently, a decline in the photosynthetic activity of P. halepensis forests would be a
sign of widespread aridification in the region, given the high degree of adaptation of this species to
common water stress conditions.
The aims of the study were to investigate: i) changes in leaf activity in P. halepensis forests, ii)
spatial differences in leaf activity trends over the past three decades, iii) the likely climatic drivers
of temperature and precipitation trends, and spatial patterns of aridity and iv) the predicted changes
in the leaf activity according to the climate change projections. We addressed these aims using: i)
an exploratory step based on high resolution remote sensing data, and ii) a confirmatory step using a
forest growth model to simulate the leaf area index (LAI) for the period covered by the remote
sensing data to project LAI for the entire twenty-first century according to different climate change
scenarios. We tested the hypothesis that trends in forest activity and LAI are controlled in
contrasting ways in the most arid and humid areas by recent temperature trends, and with a strong
influence of aridity. The study was highly relevant as Mediterranean forests are highly sensitive and
vulnerable to change as a consequence of: i) centuries of human use and modification of the land
(e.g. fires) that have affected the resilience of forests to climate perturbations (e.g. drought); ii) the
high temporal variability of climate, characterized by frequent and severe drought periods; and iii)
the increased water stress predicted for the region by current climate change models for the end of
5
the 21st century, resulting from a large decrease in precipitation and increased evapotranspiration
rates (Giorgi and Lionello, 2008). Thus, temperature in the study region increased markedly (about
2ºC) during the 20th century (Brunet et al. 2006), aridity is very pronounced (Vicente-Serrano et
al., 2006), and there is low precipitation (Cuadrat et al., 2007) and high rates of evapotranspiration
during summer (Vicente-Serrano et al., 2007).
Study area
The study area corresponds to the 199/31 Landsat scene located in the middle Ebro Valley, which is
one of the most arid regions of the Iberian Peninsula. Relief determines the climate, as the valley is
surrounded by mountain chains that cause continental climatic characteristics and dry conditions
(Cuadrat et al., 2007). Temperature is strongly seasonal, with frequent frost periods in winter that
markedly limit vegetation development. In the center of the valley the average winter temperature is
10ºC, but minimum temperatures below 10ºC are common. The average temperature in summer is
24ºC, although maximum temperatures above 40ºC are common. Seasonal variability in
precipitation is less pronounced, although summer is the predominant dry season. In the center of
the valley the annual precipitation is less than 350 mm. Throughout the majority of the study area
there is a negative water balance (precipitation minus evapotranspiration), as a consequence of the
high potential evapotranspiration (PET) that occurs in summer. Thus, annual PET reaches 1300 mm
in some sectors of the valley, which causes a very negative water balance (> 900 mm). In contrast,
in the northernmost mountain areas the annual water balance is usually positive, as a consequence
of greater precipitation and lower PET rates. Figure 1 shows the location of the study area and the
spatial distribution of climatic water balance, obtained from GIS modeling and PET calculations
(see below). Moreover, in the middle Ebro Valley the high temporal variability in precipitation
introduces more limitations to vegetation growth, as severe drought periods occur very frequently
(Vicente-Serrano and Cuadrat, 2007), and periods of more than 80 days without precipitation are
common (Vicente-Serrano and Beguería, 2003). The lithology is characterized by millstones and
6
gypsums (Peña et al., 2002), which contribute to aridity because there is poor retention of water by
the soils (Navas and Machín, 1998).
Forested landscapes comprising P. sylvestris, Quercus faginea, P. nigra and Q. ilex dominate the
mountainous areas of the north and southwest. However, because of their adaptation to arid
conditions the unique forests located in the most arid regions at center of the valley are composed of
P. halepensis, with a few areas of Juniperus turiphera.
The P. halepensis forests in the center of the Ebro Valley commonly occur on the top and slopes of
structural platforms developed on Miocene carbonate and marl sediments, which are arranged on
horizontal strata but also occur on the margin of the limestone and conglomerate structures that
characterize the Pre-Pyrenean chains in the north. They occupy 177,266 ha in the area covered by
the 199/31 Landsat image (MAPA, 1978). Commercial exploitation of the forest is marginal, as the
traditional extraction of wood for construction, firewood and resin was abandoned after the 1950s
and 1960s. Although the structure of these forests shows some variability, they are comprised
mostly of P. halepensis with underlying vegetation dominated in the most humid areas by Buxus
sempervirens, Juniperus communis and Q. coccifera, which form a dense underlying cover.
Thermophilic shrubs occur in the most arid forests, dominated by Rosmarinus officinalis, Thymus
sp. and Genista scorpius. Communities dominated by Ononis tridentata, Gypsophila hispanica and
Helianthemum squamatum occur on some gypsum outcrops.
Material and methods
Remote sensing data
Remote sensing data enables the spatial and temporal dynamics of the vegetation cover to be
identified. The energy recorded by satellite sensors for different regions of the electromagnetic
spectrum has been widely used since the 1970s to monitor vegetation status. The degree of
vegetation activity, coverage, total biomass and water content affect the canopyenergy
interactions, and thus the total reflected energy detected by the sensors. Data on various biophysical
7
parameters can be extracted from satellite imagery, including the LAI, primary biomass and
percentage of vegetation cover (Sellers, 1985; Myneni, 1997; Gower et al., 1999). Nevertheless,
given the complexity involved in accurately applying models to different vegetation types and
conditions, it is common to use spectral indices (vegetation indices) to determine the vegetation
status and activity. Calculation of vegetation indices is based on the reflective properties of the
active green vegetation: high absorption of energy is recorded in the visible part of the
electromagnetic spectrum and low absorption is recorded in the near infrared region (Knipling,
1970). As most Earth observation satellites record information in the visible and near-infrared
regions, it is possible to calculate vegetation indices that facilitate vegetation monitoring, and the
analysis of spatial and temporal patterns in vegetation activity (Baret and Guyot, 1991; Bannari et
al., 1997). The most widely used vegetation index used is the normalized difference vegetation
index (NDVI), which is calculated according to (Rouse et al., 1974):
NDVI 
 IR   R
 IR   R
where IR is the reflectivity in the near-infrared region, and R is the reflectivity in the red region of
the electromagnetic spectrum. The NDVI is determined by absorption in the red region (which is
proportional to the chlorophyll density) and the reflectivity in the near infrared (which is
proportional to the vegetation density). High NDVI values are indicative of high vegetation activity.
In reality the NDVI is a measure of the photosynthetic capacity of the canopy (Ruimy et al., 1994;
Gallo et al., 1985) and the stomatal resistance to water vapor transfer (Tucker and Sellers, 1986).
The main limitation of the NDVI is related to its saturation in high biomass and LAI environments
(Gillies et al., 1997; Choudhury, 1987), as the NDVI saturates before maximum biomass is reached
(Carlson et al., 1990). This implies that the relationship between the LAI and the NDVI is strongly
non-linear (e.g., Baret and Guyot, 1991; Wang et al., 2005) and that the LAI estimation using
remote sensing data is commonly based on non-linear approaches (González-Sanpedro et al., 2008;
Ganguly et al., 2008a and 2008b). Nevertheless, several studies have shown that the relationship
8
between the NDVI, the leaf biomass and the LAI is linear up to a threshold LAI, which Carlson and
Ripley (1997) determined to be between 2 and 4. The LAI of the P. halepensis forests of the
Mediterranean region commonly oscillates between 1 and 2, and on this basis a linear association
between changes in the forest NDVI, changes in leaf biomass, and the LAI can be assumed.
Among the various satellite sensors providing data for analysis of the Earth's surface characteristics,
including vegetation activity and change, the Landsat TM has proved invaluable over the past three
decades (Cohen and Goward, 2004). The systematic archiving of Landsat data makes this
information very useful for retrospective analysis of land surface characteristics, and the spatial (30
m since 1984) and spectral resolution of the images facilitates detection of both abrupt and gradual
changes in vegetation cover, as well as monitoring of a number of forest processes (e.g. Cohen et
al., 2002; Millington et al., 2003; Song and Woodcock, 2003). For analysis of small or gradual
changes in forests the sensor calibration performance and capability must be maintained, and the
U.S. Geological Survey has devoted considerable effort to this task in the Landsat program
(Chander and Markham, 2003; Chander et al., 2004; Chander et al., 2007).
For analysis of changes in the leaf activity of the P. halepensis forests in the central Ebro Valley we
used a database of Landsat TM and Landsat ETM+ images for the period 19842006. The database
for August during the analysis period comprises 16 images free of cloud cover; the dates of the
images used are shown Table 1. Images in summer were selected to avoid the noise caused by
seasonal variations in vegetation activity. In addition, as a consequence of the aridity the majority of
the herbaceous and shrub cover was not active in the region in summer, and the activity recorded by
the satellite was mostly that of the P. halepensis trees. Moreover, it is easier to detect possible
effects of climate change and variability on forests in summer, as irradiance or temperature limit the
photosynthetic capacity of plants (Valladares and Pearcy, 1997). Thus, Mediterranean summer
drought stress causes a generalized decrease in chlorophyll, as a common regulatory mechanism
(Kyparissis et al., 1995).
9
To ensure that gradual changes recorded in the forests were really responses to changes in
vegetation activity or forest parameters, and not to external noise (related to sensor degradation,
changes in illumination conditions or atmospheric effects), the database of Landsat images was
processed using a procedure involving: i) geometric correction using ground control points and
elevation (Palá and Pons, 1995) with a root mean square error < 0.5 pixels, which results in a
perfect match between images, ii) calibration and cross-calibration of the TM and ETM+ images
(Vogelmann et al., 2001), iii) atmospheric correction using a radiative transfer model (6S) that
included external atmospheric information (Vermote et al., 1997), iv) a non-Lambertian
topographic correction to avoid errors caused by differences in illumination conditions (Riaño et al.,
2003), and v) a relative normalization between images (Du et al., 2002). This procedure allowed
accurate measurements of physical surface reflectance units to be obtained. The correction applied
to the images ensured temporal homogeneity of the dataset, the absence of artificial noise caused by
sensor degradation and atmospheric conditions, and spatial comparability among areas, given the
accurate topographic normalization applied. Details of the correction procedure applied to the
images, and a complete description of the data set and its validation can be found in VicenteSerrano et al. (2008).
Selection of the P. halepensis forests
Among the 177,266 ha occupied by P. halepensis forests in the study area, we selected areas that
had not been affected by forest fires and/or deforestation and reforestation activities. From the
official national land use and cultivation maps developed for 1978 (MAPA, 1978), six years before
the first Landsat image, 79,177 ha of P. halepensis forests were selected for analysis during the
19842006 period. The forests were grouped into eight areas (Figure 1). Areas 1 and 4 were
affected by forest fires in 1986 and 1989, respectively, but the surface area involved represented
less than 5% of the total surface of the forests. Figure 2 shows the mean monthly PET and
precipitation in the eight analyzed forests, obtained from GIS-interpolation (see below). The
10
climatic conditions in the various forests showed similar patterns: i) a precipitation peak in spring
and in autumn, and ii) very high PET rates in summer. The values representing water deficit and
surplus showed differences among areas, with marked contrasts between the most humid (1 and 3)
and the driest (6, 7 and 8) forests. Table 2 shows the mean annual values of precipitation, PET and
water balance in each forest. Differences in the water balance (average 400.5 mm) between the
driest (6) and the most humid (1) forests were substantial.
Analysis of NDVI change
To improve spatial and temporal comparability, the NDVI series were converted to series of
standardized anomalies, considering the long-term mean and standard deviation of all the available
Landsat scenes.
To determine the patterns of change in NDVI standardized anomalies and the statistical
signification of their trends between 1984 and 2006 in the various P. halepensis forests, we used a
non-parametric coefficient (Spearman rho), as it is less affected by non-normality of the series than
are parametric coefficients (Siegel and Castelan, 1988). The Spearman rho coefficient identifies
statistically significant trends, but not the magnitude of change; for this purpose we used a least
square regression to calculate the regression slope between each of the series and the series of years.
This parameter indicated the annual rate of increase or decrease in the standardized NDVI for
19842006.
Climate data
To analyze the climate factors affecting changes in leaf activity in P. halepensis forests we used two
different datasets based on: i) an accurate representation of the spatial variability of the average
climate conditions in the region, and ii) a homogeneous quantification of the climate evolution in
recent decades. The first dataset was derived from the Digital Climatic Atlas of Aragón (Cuadrat et
al., 2007), obtained by spatial modeling of average climate variables at a resolution of 100 m, using
11
different potential predictors; the dataset comprised data on precipitation and PET. The calculation
procedure for the PET maps used the methodology of Vicente-Serrano et al. (2007), which is based
on a geographic information system (GIS) interpolation of the average monthly maximum and
minimum temperatures using a linear regression approach, spatial modeling of potential incoming
solar radiation using a digital terrain model following the method of Pons and Niyerola (2008), and
calculation of the PET following the Hargreaves’ equation. Monthly precipitation was also directly
interpolated using a linear regression approach, following the method of Ninyerola et al. (2000).
The errors in the monthly maximum temperature, minimum temperature and the precipitation maps
for the region were less than 1.1C, 1.0C and 8.7 mm, respectively. The goodness of predictions
were very good for the different variables, as the coefficient of agreement D (Willmott, 1982)
oscillated between 0.95 and 0.97 for the monthly precipitation maps, between 0.87 and 0.96 for the
minimum temperature maps, and between 0.91 and 0.96 for the maximum temperature maps (a
value of D = 1 represents complete agreement between the observed and predicted data). Maps of
average annual precipitation and PET were obtained from the sum of the different monthly maps,
and an annual climate water balance (precipitation minus PET) was obtained as a measure of the
average aridity. Uncertainty in the PET spatial estimates was low according to the used approach.
Vicente-Serrano et al. (2007) showed that calculations of solar radiation from a DTM and GIS
modeling provide realistic spatial distribution of PET and also that it is preferable to model in
advance the variables involved in the calculation of PET (temperature and solar radiation) and,
subsequently, to calculate PET by means of layer algebra in the GIS rather than modeling the local
PET calculations directly.
To explain the possible changes in vegetation activity, and to also apply a forest growth simulation
model to validate the remote sensing results, we used daily precipitation, and maximum and
minimum temperature series. Series were reconstructed and quality controlled, and temporal
homogeneity was tested by means of a protocol (Vicente-Serrano et al., 2009) that eliminated
12
nonclimate noise from the data. The noise commonly includes collection, transcription and
digitizing, but also instrument errors. Reconstruction relied on the assumption that the cessation of
data recording at one observatory, and the establishment of one or more new observatories close to
the previous one, results in two or more data series which are usually not useful for climate analysis
as a consequence of their short duration. Following a reconstruction approach, the shorter series can
be combined into a single series, which is ascribed to the last observatory that collected data. Also
temporal inhomogeneities are common in the climate series as a consequence of the deterioration or
replacement, variations in the time of observations, and changes in the surrounding environment.
The protocol detailed above removes all the noise related to data errors and inhomogeneities but
maintains the original climate signal without filtering the resulting series. Observatories closest to
each forest, and with data for 19652006, were used for validation and explanatory purposes (see
location in Figure 1).
The evolution of the drought conditions in the region was analyzed with the purpose of identifying,
possible driving factors in the forest activity and LAI. For this purpose, we obtained the
Standardized Precipitation Evapotranspiration Index (SPEI), which is based on a monthly climatic
water balance (precipitation minus potential evapotranspiration) adjusted using a 3-parameter loglogistic distribution to take into account common negative values. The values are accumulated to
different time scales, following a similar approach to other drought indicators, and then converted to
standard deviations with respect to average values. The complete methodology used for calculations
is described by Vicente-Serrano et al. (2010).
Forest growth simulation model
A forest growth simulation model was used to assess: (i) whether the patterns of changes in the leaf
activity of the forests, obtained from remote sensing data, agreed with outputs from a theoretical
biophysical model that included the temporal variability of climate variables in the study area, and
13
(ii) whether model predictions were sufficiently reliable to extend the evolution of vegetation
activity parameters some decades, since homogeneous records were available (1965). Different
models have been developed to represent ecophysiological processes (e.g. Kramer, 2001; Morales
et al., 2005). In this study we used a terrestrial biochemical model (GOTILWA+), which simulates
forest growth processes and predicts the evolution of several physiological parameters at a daily
time scale. GOTILWA+ simulates forest growth in different environments and different species,
taking into account changing climatic conditions. The model, which has been used and validated in
Europe (Morales et al., 2005) and Mediterranean regions (Gracia et al., 1998; Sabaté et al., 2002),
accurately predicts different ecosystem processes (Morales et al., 2005). A broad description of the
model can be found in Gracia et al. (1999) and at http://www.creaf.uab.es/gotilwa+. We focused on
the model output of the LAI (m2 m2), due to the strong and linear relationship with the NDVI in the
LAI range of 04 (Carlson and Ripley, 1997).
In GOTILWA+, the leaf area of each tree and, consequently, the leaf area index (LAI) of the forest
are all highly dependent on water availability in the model. Contributing to the formation of new
biomass components, there is a metabolic cost which constitutes the growth respiration. The
balance among the maintenance respiration, the Net Primary Production and the metabolic cost
associated to the formation of new biomass determines the processes of leaf formation and leaf fall.
In the leaf energy balance, GOTILWA+ accounts for both the fraction of the reflected radiation
arriving to the leaf and the leaf surface characteristics. For example, pubescence or waxes can
increment this term. Thus, the physiological characterization of the leaves of a species has several
parameters involved in the model. These leaves can be amphistomatous or hypostomatous and
obviously, deciduous or evergreen. The leaf area index determines the amount of intercepting
surface and leaf angle, which changes the amount of light that the leaf absorbs and, thus, it is the
most influential factor involved in the different rates of light extinction through canopy.
14
The model requires a number of inputs to simulate forest growth parameters at a daily time scale.
We considered each of the eight analyzed forests as homogeneous in terms of the tree diameter at
breast height classes, which were defined according to the third Spanish national forest inventory
(http://www.mma.es/portal/secciones/biodiversidad). Leaf photosynthesis and stomatal conductance
parameters for P. halepensis were included in the GOTILWA+ model, based on measured data
from
experimental
plots
in
Collserola
(Catalonia)
(http://www.creaf.uab.es/gotilwa+/SParameters.htm). Soil parameters for some of the forests were
obtained from previous studies (Badía, 1989; Machín and Navas, 1994; Badía et al., 2000; Pardina
and Badía, 2004). In those forests for which parameters were not available, approximate values of
hydrological soil properties and organic matter were assigned from the Spanish National Map of
Soil Types (Ministerio de Fomento, 2007), according to the soil type. The climate variables
required by GOTILWA+ were daily precipitation, and maximum and minimum temperature; these
were obtained for 1965 and 2006 from the observatories indicated in Figure 1. Monthly atmospheric
CO2
levels
(ppmv)
were
obtained
from
Mauna
Loa
station
(http://cdiac.ornl.gov/ftp/trends/co2/maunaloa.co2) and were interpolated daily. Wind speed data
were obtained from the observatory at Zaragoza, in the center of the study area. The vapor pressure
deficit (ea minus ed) is usually determined from air temperature and relative humidity. The mean
saturation vapor pressure, ea, was estimated from:
e o (Tmax )  e o (Tmin )
ea 
2
where eo(T) is the saturation vapor pressure at temperature T, and is estimated from:
e (T )  0.6018e
o
 17.27T 


 T  237.7 
The mean daily vapor pressure, ed, was estimated from:
ed 
hr
ea
100
15
where hr is the mean daily relative humidity. The parameter ed, can also be taken as the saturation
vapor pressure at the mean daily dew-point temperature, Td. Td is usually relatively constant during
the day, and the nightly minimum temperature, Tmin, tends to come to equilibrium with Td. Thus, ed
can be approximated as:
ed  e o (Tmin )
This assumption has been shown to be acceptable in a range of climates (Kimball et al., 1997).
Global radiation at the Earth’s surface was estimated according to the method of Hargreaves et al.
(1985), who showed that this parameter can be estimated from the maximum and minimum
temperature by:
Rs  aRa Tmax  Tmin  b
where Rs is the global radiation at the surface in MJ m2. Ra is the extraterrestrial radiation (MJ
m2), and is determined from the latitude and date using the methods detailed by Allen et al.,
(1998). Tmax and Tmin are the maximum and minimum air temperatures (°C), respectively. a and b
are coefficients that, according to Hargreaves et al. (1985) and based on the latitude of the study
area, are 0.16 and 0,86, respectively. This method for solar radiation estimation has been
successfully tested in other locations of the Iberian Peninsula (Castellvi, 2001).
Finally, in the GOTILWA+ simulations we did not consider understorey biomass or forest
management. The monthly LAI values were obtained for August (coinciding with the NDVI data)
in the years from 1965 to 2006 for each one of the eight analyzed forests.
Climate change projections
To determine the possible future evolution of the forest activity in the analyzed Pinus halepensis
forests, we used the climate projections of a multi-model approach in the frame of the European
Project PRUDENCE (http://prudence.dmi.dk). The models HIHAM, HADRM3, RCAO, REGCM
and PROMES driven by the same Global Circulation Model: HadAM3H were used to determine
16
the average changes in the seasonal maximum and minimum temperature and total precipitation in
the period 1070-2100 over the analyzed region. We analyzed two emission scenarios: A2 and B2,
which consider average-high and average-low greenhouse gasses emissions, respectively
(Nakicenovick et al., 1998). In the study area, an increase of 4,1ºC and 3.6ºC in the average annual
maximum and minimum temperature, respectively, in the A2 scenario and an increase of 2.5ºC and
2.2ºC, respectively, in the B2 scenario are expected. Moreover, annual precipitation is expected to
reduce by 18% and 13% in the A2 and B2 scenarios, respectively. However, these changes have
expected seasonal differences with the most important warming and precipitation decrease in
summer months.
We generated random daily temperature and precipitation series between 2006 and 2100 that follow
the expected climate evolution in each one of the eight observatories used according to the A2 and
B2 scenarios. For this purpose, we used the Gamma distribution to fit the precipitation series and
the Normal distribution to fit the maximum and minimum temperature series. Independent
distribution parameters were obtained for the three 30 year-periods (2010-2040, 2040-2070 and
2070-2100), considering a linear precipitation and temperature evolution from the present to the
2070-2100 projections. Parameters were obtained independently for each season to consider the
projected seasonal differences in the evolution of precipitation and temperature.
GOTILWA+ was run from 1959 to 2100 to determine time series of LAI for each summer and each
one of the eight analyzed forests considering the multi-model precipitation and temperature changes
in the A2 and B2 scenarios and the evolution of the CO2 concentrations estimated for both scenarios
(Solomon et al., 2007).
Results
Table 3 shows the percentage of surface area with positive and negative trends for each of the
analyzed P. halepensis forests. Large differences were found among forests. Forests 1, 3, 4 and 5
were dominated by positive trends. It is remarkable that 33% of the surface area of forest 1 showed
17
positive and significant trends. High percentages of positive trends were also found for forests 3 and
4 (12.8% and 21.2%, respectively). In contrast, forests 6, 7 and 8 predominantly showed statistically
significant negative trends, between 12.6% and 17.5% of the total surface area. In these three
forests less than 3% of the surface showed positive and significant trends.
Figure 3 shows the spatial distribution of the magnitude of change in each forest. Forests 1 and 4
showed the generalized spatial patterns of forest growth. In contrast, forests 6, 7 and 8 showed a
dominant pattern characterized by a general decrease of the standardized NDVI values. In general,
there were no large spatial differences in the change within each region, and a homogeneous pattern
was observed for each of the analyzed forests.
Figure 4a shows the relationship between the change in the standardized NDVI (least square slope
parameter) for each forest and the annual aridity conditions (climatic water balance), obtained from
GIS modeling. The relationship between the average values for each forest was very strong and
statistically significant (R = 0.83, p = 0.01). The greatest increases in the NDVI value were recorded
in the most humid forests. In contrast, forests located in the driest areas showed a slight decrease
between 1984 and 2006 in the summer NDVI. Thus, there was a direct and linear relationship
between aridity and the recorded NDVI changes, reinforcing the central role of water availability in
explaining forest dynamics in these semiarid environments. Figure 4b shows the relationship
between the NDVI changes and the average NDVI at the beginning of the period (average NDVI
values for 1984, 1985 and 1987). This figure is included to show that the NDVI increase was not
related to the age of the forests (young trees commonly have lower NDVI values than trees in
mature forests), as the growth of pine trees and the rate of the LAI increase is commonly greater for
young trees than mature trees (Trasobares et al., 2004; Condés and Sterba, 2008). We did not
observe this in our study, as all the forest were well established at the beginning of the analysis
period (1984), and although the relationship was not statistically significant, a greater increase in
the NDVI was recorded in the areas with higher NDVI values at the beginning of the study period.
18
This indicates that the control of NDVI trends by aridity was not biased or determined by other
factors, such as the leaf biomass in the forest.
Figure 5a shows the relationship between the average NDVI values for each forest between 1984
and 2006, and the average LAI modeled using GOTILWA+. As expected, there was a strong
correlation between the two variables (R = 0.82, p = 0.01), showing that the forests with the highest
(lowest) NDVI values recorded the highest (lowest) values of LAI. Figure 5b shows that there was
also strong agreement between the average change in the NDVI for each forest between 1984 and
2006, and the modeled change in the LAI; the correlation was positive and statistically significant
(R = 0.82, p = 0.01). The relationship between the LAI trends and the water balance was also very
strong (R = 0.85, p = 0.006; Figure 5c), demonstrating that very similar and comparable results
were obtained from remote sensing data and the forest growth simulation.
Figure 6 shows the evolution of the modeled LAI (black line) and the NDVI for each year since
1965. The values represent standardized anomalies regarding the 1984-2006 period. In forests 1 and
3 the general positive trend in the NDVI and LAI values observed between 1984 and 2006 was the
result of a continuous positive trend since 1965. In contrast, the stability or slight decrease in the
LAI and the NDVI in the driest forests (6, 7 and 8) since 1965 is also evident from the LAI
simulation. In the majority of forests (especially forests 1, 2, 4, 5 and 8) there was a major increase
in the LAI values until the end of the 1970s, following which there was a large drop in the LAI
values. After this, forests 1 to 5 showed a progressive increase in the LAI, but in forests 6, 7 and 8
the LAI remain unchanged or had a slight trend of decrease.
The climate evolution during the last four decades explains the observed trends in the LAI and the
NDVI, and the spatial differences in the eight forests analyzed. Drought conditions have noticeably
increased in the region, with a higher persistence in the drought duration during the last decades
(Figure 7). Tthe most extreme droughts episodes were recorded during the decades of the 1990 and
2000. The evolution of droughts would explain the different behavior observed in the eight
19
analyzed forests, in relation to the moisture availability and the warming processes. Figure 8 shows
the average evolution of precipitation and annual PET in various periods of the year (calculated
according to the method of Hargreaves), based on an average series for the eight analyzed forests.
The annual regional series was correlated to the eight independent series, which oscillated between
0.78 and 0.89 for precipitation and between 0.91 and 0.98 for PET; this can be considered
representative of the climate evolution for the entire study area between 1965 and 2006. Various
periods of the year were considered as we did not know a priori in which season climate affects the
activity of forests. Table 4 shows the trend in these variables for the periods 19842006 and
19652006, using the Spearman rho non-parametric coefficient. Precipitation did not show a
significant trend in either period independent of the monthly aggregation of the data. However, for
the 19652006 period the coefficients were negative, indicating a slight trend toward lower
precipitation in the region. Moreover, it is noteworthy that, since 1980, a higher frequency of dry
years has been recorded, including some very extreme years (e.g., 1986, 1995 and 2006). In
contrast, the PET during spring and summer showed a positive and significant trend for the two
analyzed periods. Since 1980, in the periods of high evapotranspiration (spring and summer) the
average PET increased about 100 mm. This has increased the water demand by vegetation, with the
most negatively affected forests being those located in the most arid environments, where an
increase in PET will increase water stress, and limit leaf activity and the development of leaf
biomass. In contrast, forests located in the most humid areas (e.g. forest 1) would be favored by the
greater respiratory activity associated with high PET, as water availability is enough to maintain
normal ecophysiological processes.
Figure 9 shows the LAI evolution in the eight analyzed forests in the twenty-first century according
to the predicted changes of precipitation and temperature in the A2 and B2 emission scenarios. The
forest 1, located in the most humid area, shows a general increase in the LAI in the A2 scenario. On
the contrary, considering the B2 scenario the LAI values in forests 1 tend to stabilize regarding the
20
2000-2006 values. This is the unique forest that shows a noticeable increase in the LAI considering
the A2 scenario. The remaining forests show a opposite behavior, with a predicted progressive
decrease of the LAI during the twenty-one century, more pronounced under the A2 scenario. The
decrease is more apparent in the forests 3, 4 and 5, which are not located in the most arid areas.
These forests tend to reduce the LAI values by 9% to 13% regarding the 2006 values. In the most
arid forests (2, 6, 7 and 8), the decrease is less pronounced (5%).
Discussion
In the central Ebro Valley, we showed a general increase in the NDVI values for P. halepensis
forests since 1984. The LAI, simulated using a tree growth model, showed the same pattern as the
NDVI, that is, a general increase in the LAI since 1965. Very similar results were obtained using
the NDVI from Landsat images and the LAI simulations from the GOTILWA+ model; this was
evident in terms of the spatial relationships between the NDVI and the LAI, trends in both
variables, and relationships with aridity. The outputs of forest growth models are commonly
validated using other variables, mainly obtained from tree ring records and indicators of wood
production (e.g. Landsberg and Waring, 1997; Gaucherel et al., 2008). However, the LAI is
difficult to validate because of the large spatial variability, and difficulties in quantification over
large areas. Although the purpose of using both measurements was not to validate either the NDVI
measurements or the LAI simulations, the very similar spatial and temporal results obtained confer
significant robustness to the trends found in the leaf biomass/activity during the past three decades.
The general trend towards greater vegetation activity and higher LAI observed in the majority of
analyzed P. halepensis forests in the central Ebro Valley is consistent with most observations at
continental and global scales (Zhou et al., 2001; Slayback et al., 2003; Goetz et al., 2005). In most
forests of the Mediterranean region, measurements of the NDVI and the LAI have indicated an
increase in vegetation activity (Lanfredi et al., 2004; Alcaraz et al., 2006; Hill et al., 2008). This
evolution is consistent with the general increase in forest growth in this region, as measured with
21
tree rings (Briffa, 1992; Spiecker et al., 1996; Martínez-Villalta et al., 2008). As P. halepensis is
highly adapted to water stress conditions and has limited water requirements (e.g. Baquedano and
Castillo, 2007), temperature increase is expected to have a positive effect on forest activity and
growth. For Pinus species this has been the dominant pattern during recent decades in the
Mediterranean area (e.g. Vila et al., 2008; Martínez-Villalta et al., 2008). Forest growth is
benefiting from CO2 fertilization, which promotes increased water use efficiency and allows trees to
reduce water stress (Hättenschwiler et al., 1997; Rathgeber et al., 2000), but also from nitrogen
fertilization and temperature increase, which favor respiration processes and plantatmosphere gas
interchange.
However, our results also demonstrated very clear spatial differences in P. halepensis forests in the
region, even though the recent climate evolution in terms of temperature and precipitation
variability has been spatially very homogeneous. The greatest increase in both the LAI and the
NDVI has occurred in the most humid forests, while these parameters have not increased in forests
in the most arid parts of the region, where a significant decrease in the NDVI has occurred over
substantial areas (from 12.6% to 17.5%). Thus, over recent decades there has been a direct
relationship between the aridity in each forest and the magnitude and direction of change. This
pattern is not restricted to the analyzed area, as vegetation at the limits of its environmental
distribution is more vulnerable to climatic variability than that located in areas where climatic
conditions more optimal for forest activity (Fritts, 1976). Some studies have also shown that
Mediterranean forests located near their water availability distribution limit show no changes, or a
decrease, in leaf activity, primary production and secondary growth (Maselli, 2004; Jump et al.,
2006; Sarris et al., 2007; Martínez-Villalta et al., 2008). This demonstrates that aridity is the main
constraint on forest growth in this region, determining the distinct spatial distribution of forests as
well as trends in forest activity.
22
The evolution of precipitation during the last four decades in the study area did not show a
significant change, although between 1965 and 2006 a negative coefficient was observed,
reinforced by long dry periods in the 1990s and 2000s. This is consistent with the generalized
decrease in annual precipitation reported for the entire Mediterranean basin (Hoerling et al., 2006;
Alpert et al., 2008). The most common finding for the Iberian Peninsula is a trend of less
precipitation (López-Moreno et al., 2009). De Luis et al. (2009) showed that for Mediterranean
Spain a greater decrease in precipitation occurs in summer and spring. In the central Ebro Valley
the intensity and magnitude of droughts have increased in recent decades (Vicente-Serrano and
Cuadrat, 2007), with 1995 and 2005 being the driest years in the region since the 1940s.
More important than decreased precipitation has been the increase in PET losses during spring and
summer, which is directly related to increased temperatures. The Earth’s average temperature has
been increasing since the 19th century, especially since the 1920s (Jones and Moberg, 2003), and
the Mediterranean has followed the global trend (Repapis and Philastras, 2004). Since the 1950s all
studies for the region indicate a trend toward higher temperatures. Thus, the increase in temperature
in the last 100 years for the entire basin has oscillated between 1 and 4.5ºC (Brunetti et al., 2004;
Alpert et al., 2008). At a more local scale, increased temperatures have been reported for the Iberian
Peninsula (Brunet et al., 2006). The generalized increase in temperature in the Spanish
Mediterranean region has particularly affected maximum temperatures in summer (Abaurrea et al.,
2007), increasing water stress in this season. In the central Ebro Valley the increase in the
springsummer PET was more than 100 mm between the end of the 1970s and 2006. The evolution
of precipitation and PET during the last four decades has been the major influence on trends in the
NDVI and the LAI, and responsible for the spatial differences observed in the P. halepensis forests
of the central Ebro Valley.
The decreased precipitation, the concentration of dry years in recent decades, and particularly the
increase of PET rates in the central Ebro Valley explain the stability of the NDVI and LAI, or
23
trends of slight decrease in these parameters, observed in the most arid forests. A temperature
increase can favor the activity and growth of trees by increasing respiration processes, but it also
interacts with water balance. Thus, depending on water availability, the effect of a temperature
increase can be positive or negative for vegetation. In water-limited ecosystems, the response of
trees to rising CO2 and temperature is determined by water availability: an increase in temperature
without a corresponding increase in precipitation will enhance water stress, limiting leaf activity
and growth. This explains the spatial differences in the NDVI and LAI trends for P. halepensis in
the central Ebro Valley. The increased temperature increases the vapor pressure deficit in the air
over the entire region, leading to increases in the vegetation transpiration rates. In humid areas,
where water is available, warming favors atmosphereplant gas exchange, photosynthesis and the
fixation of carbon by plants (Rustad et al., 2001). In dry sites the same process increases water
stress, affecting stomatal conductance and closure, and can even cause xylem embolism and plant
die-off. Breshears et al. (2005) showed that droughts under warming conditions cause a more
general tree die-off in semiarid regions than similarly severe droughts under non-warming
conditions. In dry areas of Catalonia (northeast Spain), Martínez-Villata et al. (2008) showed that
increased temperature was always detrimental to growth, but particularly so in dry years. Therefore,
although P. halepensis is highly adapted to aridity, it adopts the strategy of reducing leaf biomass
under severe stress conditions. In a recent experiment in southeast Spain, Belmonte et al. (2008)
showed that a response in P. halepensis to severe water stress was increased litter production, which
reduced the LAI and leaf activity. Moreover, in the central Ebro Valley, repeated severe droughts in
the 1990s and 2000s can have increased the impact of warming processes to explain the slight
decrease in the NDVI and LAI values in the most arid areas, reducing the capacity of vegetation to
recover, because the damage to forests caused by droughts can persist for several years (Orwing and
Abrams, 1997).
24
Although projections of the future evolution of climate and land usevegetation change are subject
to uncertainty, the best available tools for modeling climate indicate marked changes in the
Mediterranean region in coming decades. For temperature, the studies show a consistent trend to
warmer conditions (e.g. Gibelin and Dequé 2003; Giorgi, 2006), with the magnitude of increase for
the area in the 21st century predicted to be 16ºC. More extreme hot events are also expected
(Giorgi and Lionelo, 2008). These conditions will increase evapotranspiration rates and water stress
in the region.
For precipitation the projected changes exhibit more variability among models than for temperature
(López-Moreno et al., 2008), but most studies project a general tendency toward less precipitation
in the next century (Giorgi and Lionello, 2008). As for temperature, the magnitude of change
depends on the model and scenario used. An increase in the interannual variability in precipitation
is expected (Giorgi et al., 2004; Alpert, 2008) in line with recent observations (De Luis et al.,
2009), as is an increase in the length of the dry season (Evans, 2008). Droughts of a severity with a
present predicted 100-year occurrence will recur every 10 years in the northern Mediterranean
(Wei et al., 2007). Gao and Giorgi (2008) have shown that by the end of the 21st century the
Mediterranean region may experience a substantial increase in the northward extension of dry and
arid lands, due to a substantial increase in temperature and a pronounced decrease in precipitation.
This will be particularly evident in spring and summer, and in the central and southern portions of
the Iberian, Italian, Greek and Turkish peninsulas and the major islands (Corsica, Sardinia and
Sicily).
The impact of global climate change and vegetationclimate feedbacks makes it very difficult to
predict future vegetation patterns, but there is strong consensus in relation to the positive role of
future CO2 fertilization on forest growth and activity (Bazzaz, 1990; Norby et al., 1999). Gaucherel
et al. (2008) used a validated ecophysiological model based on tree ring data to assess the future
impact of climate change on the production of P. halepensis in the Mediterranean regions of France.
25
They showed an increase in production of about 26% was related to CO2 fertilization; their results
are consistent with the findings of Rathgeber et al. (2000) for P. halepensis forests. This pattern is
dependent on water being available to maintain forest activity but, as indicated above, most climate
models predict a large reduction in water availability in the Mediterranean region. As Sabaté et al.
(2002) indicated, the positive effects of climate change on forest ecophysiological functioning may
be offset if a temperature increase is not matched by an increase in water availability, and in the
long-term such negative effects may not be compensated for by an increase in CO2 concentrations.
We have shown that forests located in the most water-limited areas are not favored by temperature
increase and CO2 fertilization. Thus, the GOTILWA+ model has shown that under two different
climate change emission scenarios (A2 and B2), an important decrease of the LAI in most of the P.
halepensis forests of the Ebro valley is predicted providing consistent evidence on water availability
as the main constraint factor of forest growth in the region. Thus, it is expected that by aridity
increase in the region, these forests will show a more pronounced decay in the LAI and NDVI
values, and even die-off may occur, as being recorded in other regions as a response to warming
processes and drought (Breshears et al. 2005; Van Mantgem and Stephenson, 2007). The
disappearance of these forests would have major implications for soil degradation, erosion and
water balance. The soils of the center of the valley are poor and are unable to retain the water as a
consequence of the high hydraulic conductivity (Navas and Machín, 1998), and the frequent intense
precipitation events in the region (Beguería et al., 2009) could increase soil losses in deforested
areas. Moreover, semiarid forests are important sites for carbon sequestration in the Mediterranean
region (Grünzweig et al., 2003), and reduced forest activity or the disappearance of large forested
areas will have major implications for the carbon balance in the region.
Conclusions
In the northernmost semiarid region of Europe, there are large spatial differences in the response of
P. halepensis forests to the recent evolution of precipitation and temperature. The climate evolution
26
in this region has been very spatially homogeneous, but a similar evolution in the NDVI and the
LAI, modeled as a function of aridity in the last three decades, has not been observed, although
consistency has been observed in the NDVI and LAI trends in P. halepensis forests in the region.
The increase in PET and the slight decrease in precipitation has exacerbated water stress conditions
in the most arid parts of the region, which explains the stability of, or slight decrease in, the NDVI
and LAI values in recent decades. In contrast, forests located in the most humid areas have shown a
large increase in these indices, as water is not the main limiting factor. This research highlights the
great spatial diversity in the ecological impact of climate change processes in regions where climate
variability has been very homogeneous, but within which contrasting climatic conditions occur.
Climate change models predict an increase in water stress conditions in the Mediterranean region,
characterized by a decrease in precipitation and an increase in temperature with a noticeable impact
on the most arid forests. This pattern could threaten the survival of P. halepensis forests located in
the most arid regions, as chronic limiting water conditions, caused by warming, and the severe
and/or prolonged droughts that are common in the region, could cause extensive forest dieback.
Acknowledgements
This work has been supported by the research projects CGL2005-04508/BOS, CGL200801189/BTE, CGL2006-11619/HID and CGL2008-1083/CLI financed by the Spanish Commission
of Science and Technology and FEDER, EUROGEOSS (FP7-ENV-2008-1-226487) and ACQWA
(FP7-ENV-2007-1- 212250) financed by the VII Framework Programme of the European
Commission, “Las sequías climáticas en la cuenca del Ebro y su respuesta hidrológica” Financed by
“Obra Social La Caixa” and the Aragón Government and “Programa de grupos de investigación
consolidados” financed by the Aragón Government.
References
27
Abaurrea J, Asín J, Cebrián AC, Centelles A (2007) Modeling and forecasting extreme hot events in
the central Ebro valley, a continental-Mediterranean area. Global and Planetary Change, 57,
43-58.
Abrams MD, Ruffuer MC, Morgan TA (1998) Tree-ring responses to drought across species and
contrasting sites in the ridge and valley of central Pennsylvania. Forest Science, 44, 550558.
Achard F, Eva HD, Stibig HJ, Mayaux P, Gallego J, Richards T, Malingreau JP (2002)
Determination of deforestation rates of the world's humid tropical forests. Science, 297, 9991002.
Alcaraz D, Paruelo J, Cabello J (2006) Identification of current ecosystem functional types in the
Iberian Peninsula. Global Ecology and Biogeography, 15, 200-212.
Allen RG, Pereira LS, Raes D, Smith M, (1998). Crop evapotranspiration. Guidelines for
computing crop water requirements. Irrigation and Drainage Paper No. 56, FAO, Rome,
Italy.
Alpert P, Krichak SO, Shafir H, Haim D, Osetinsky I (2008) Climatic trends to extremes employing
regional modeling and statistical interpretation over the E. Mediterranean. Global and
Planetary Change, 63, 163-170.
Andreu L, Gutiérrez E, Macia M, Ribas M, Bosch O, Camarero JJ (2007) Climate increases
regional tree-growth variability in Iberian pine forests. Global Change Biology, 13, 804-815.
Badía D (1989) Los suelos en Fraga. Cartografía y evaluación. Instituto de Estudios
Altoaragoneses.
Badía D, Martí C, Cuchí JA, Casanova J (2000) Suelos de la sierra de Salinas al río Vero
(Somontano de Barbastro, Altoaragón). Edafología, 7, 241-249.
Bannari A, Morin D, Bonn F, Huete AR (1995) A review of vegetation indices. Remote Sensing
Reviews, 13, 95-120.
Baquedano FJ, Castillo FJ (2007) Drought tolerance in the Mediterranean species Quercus
coccifera, Quercus ilex, Pinus halepensis, and Juniperus phoenica, Photosynthetica, 45,
229-238.
Barber VA, Juday GP, Finney BP (2000) Reduced growth of Alaskan white spruce in the twentieth
century from temperature-induced drought stress. Nature, 405, 668-673.
Baret F, Guyot G (1991) Potentials and limits of vegetation indices for LAI and APAR assessment.
Remote Sensing of Environment, 35, 161-173.
Barford CC, Wofsy SC, Munger JW, Goulden ML, Pyle HE, Urbanski SP, Hutyra L et al. (2001)
Factors controlling long- and short-term sequestration of atmospheric CO2 in a mid-latitude
forest. Science, 294, 1688-1691.
Bazzaz FA (1990) The response of natural ecosystems to the rising global CO2 levels. Annual
Review of Ecology and Systematics, 21, 167-196.
Beguería S, Vicente-Serrano SM, López-Moreno JI, García-Ruiz JM (2009) Annual and seasonal
mapping of peak intensity, magnitude and duration of extreme precipitation events across a
climatic gradient, North-east Iberian Peninsula. International Journal of Climatology. In
press.
Belmonte F, López-Bermúdez F, Romero A (2008) Reducción de la biomasa del pino carrasco
(Pinus halepensis) en un área del sureste semiárido peninsular como estrategia para evitar el
estrés hídrico. Papeles de Geografía, 47-48, 25-34.
Boisvenue C, Running SW (2006) Impacts of climate change on natural forest productivity Evidence since the middle of the 20th century. Global Change Biology, 12, 862-882.
Borguetti M, Cinnirella S, Magnani F, Saracino A (1998) Impact of long-term drought on xylem
embolism and growth in Pinus halepensis Mill. Trees, 12, 187-195.
28
Bréda N, Huc R, Granier A, Dreyer E (2006) Temperate forest trees and stands under severe
drought: A review of ecophysiological responses, adaptation processes and long-term
consequences. Annals of Forest Science, 63, 625-644.
Breshears D, Cobb NS, Rich PM, Price KP, Allen CD, Balice RG, Romme WH, Kastens JH, Floyd
ML et al. (2005) Regional vegetation die-off in response to global-change-type drought
PNAS, 102, 15144-15148.
Briffa KR (1992) Increasing productivity of natural growth conifers in Europe over the last century.
Lunqua Report, 34, 64-71.
Brunet M, Saladié O, Jones P, Sigró J, Aguilar E, Moberg A, Lister D et al. (2006) The
development of a new dataset of Spanish daily adjusted temperature series (SDATS) (18502003). International Journal of Climatology, 26, 1777-1802.
Brunetti M, Buffoni L, Mangianti F, Maugeri M, Nanni T (2004) Temperature, precipitation and
extreme events during the last century in Italy. Global and Planetary Change, 40, 141-149.
Camarero JJ, Gutiérrez E (2004) Pace and pattern of recent treeline dynamics: Response of
ecotones to climatic variability in the Spanish Pyrenees. Climatic Change, 63, 181-200.
Carlson TN, Perry EM, Schumugge TJ (1990) Remote estimation of soil moisture availability and
fractional vegetation cover for agricultural fields. Agricultural and Forest Meteorology, 52,
45-69.
Carlson TN, Ripley DA (1997) On the relation between NDVI, fractional vegetation cover, and leaf
area index. Remote Sensing of Environment, 62, 241-252.
Castellvi F (2001) A new simple method for estimating monthly and daily solar radiation.
Performance and comparison with other methods at Lleida (NE Spain); a semiarid climate.
Theor. Appl. Climatol., 69, 231-238.
Chander G, Markham B (2003) Revised Landsat-5 TM radiometric calibration procedures and
postcalibration dynamic ranges. IEEE Transactions on Geoscience and Remote Sensing, 41,
2674−2677.
Chander G, Helder DL, Markham BL, Dewald JD, Kaita E, Thome KJ, Micijevic E, Ruggles TA
(2004). Landsat-5 TM reflective-band absolute radiometric calibration. IEEE Transactions
on Geoscience and Remote Sensing, 42, 2746−2760.
Chander G, Markham BL, Barsi JA (2007) Revised Landsat 5 Thematic Mapper radiometric
calibration. IEEE Transactions on Geoscience and Remote Sensing, 4, 490−494.
Chaves MM, Pereira JS, Maroco J (2003) Understanding plant response to drought – from genes to
the whole plant. Functional Plant Biology, 30, 239-264.
Choudhury BJ (1987) Relationships between vegetation indices, radiation absorption and net
photosynthesis evaluated by sensitivity analysis. Remote Sensing of Environment, 22, 209233.
Cohen W, Goward S (2004) Landsat's role in ecological applications of remote sensing. BioScience,
54, 535−545.
CohenWB, Spies TA, Alig RJ, Oetter DR, Maiersperger TK, Fiorella M (2002) Characterizing 23
years (1972–95) of stand replacement disturbance in Western Oregon forests with Landsat
imagery. Ecosystems, 5, 122−137.
Condés S, Sterba H (2008) Comparing an individual tree growth model for Pinus halepensis Mill. in
the Spanish region of Murcia with yield tables gained from the same area. European Journal
of Forest Research, 127, 253-261.
Cuadrat JM, Saz MA, Vicente-Serrano SM (2007) Atlas Climático de Aragón. Gobierno de Aragón.
229 pp.
De Luis M, González-Hidalgo JC, Longares LA, Stepanek P (2009) Seasonal precipitation trends in
the Mediterranean Iberian Peninsula in second half of 20th century. International Journal of
Climatology, in press.
29
DeFries RS, Houghton RA, Hansen MC, Field CB, Skole D, Townshend J (2002) Carbon emissions
from tropical deforestation and regrowth based on satellite observations for the 1980s and
1990s. Proceedings of the National Academy of Sciences of the United States of America,
99, 14256-14261.
Delbart N, Le Toan T, Kergoat L, Fedotova V (2006) Remote sensing of spring phenology in boreal
regions: A free of snow-effect method using NOAA-AVHRR and SPOT-VGT data (19822004). Remote Sensing of Environment, 101, 52-62.
Dittmar C, Zech W, Elling W (2003) Growth variations of Common beech (Fagus sylvatica L.)
under different climatic and environmental conditions in Europe - A dendroecological study.
Forest Ecology and Management, 173, 63-78.
Dixon RK, Brown S, Houghton RA, Solomon AM, Trexler MC, Wisniewski J (1994) Carbon pools
and flux of global forest ecosystems. Science, 263, 185-190.
Du Y, Teillet P, Cihlar J (2002) Radiometric normalization of multitemporal high resolution
satellite images with quality control and land cover change detection. Remote Sensing of
Environment, 82, 123−134.
Evans JP (2008) 21st century climate change in the Middle East. Climatic Change: 1-16.
Fritts HC (1976) Tree rings and climate. Academic Press. Londres.
Gallo KP, Daughtry CST, Bauer ME (1985) Spectral estimation of absorbed photosinthetically
active radiation in corn canopies. Remote Sensing of Environment,17. 221-232.
Ganguly S, Schull MA, Samanta A, Shabanov NV, Milesi C, Nemani RR, Knyazikhin Y, Myneni
RB (2008) Generating vegetation leaf area index earth system data record from multiple
sensors. Part 1: Theory. Remote Sensing of Environment, 112. 4333-4343.
Ganguly S, Samanta A, Schull MA, Shabanov NV, Milesi C, Nemani RR, Knyazikhin Y, Myneni,
RB (2008) Generating vegetation leaf area index Earth system data record from multiple
sensors. Part 2: Implementation, analysis and validation. Remote Sensing of Environment,
112. 4318-4332.
Gao X, Giorgi F (2008) Increased aridity in the Mediterranean region under greenhouse gas forcing
estimated from high resolution simulations with a regional climate model. Global and
Planetary Change, 62, 195-209.
Gaucherel C, Guiot J, Misson L (2008) Changes of the potential distribution area of French
Mediterranean forests under global warming. Biosgeosciences, 5, 1493-1504.
Gibelin A-L, Déqué M (2003) Anthropogenic climate change over the Mediterranean region
simulated by a global variable resolution model. Climate Dynamics, 20, 327-339.
Giglio L, Csiszar I, Justice CO (2006) Global distribution and seasonality of active fires as observed
with the Terra and Aqua Moderate Resolution Imaging Spectroradiometer (MODIS)
sensors.
Journal
of
Geophysical
Research
G:
Biogeosciences
111,
doi:10.1029/2005JG000142.
Gillies RR, Carlson TN, Cui J, Kustas WP, Humes KS (1997) A verification of the triangle method
for obtaining surface soil water content and energy fluxes from remote measurements of the
Normalized Difference Vegetation Index (NDVI) and surface radiant temperature.
International Journal of Remote Sensing, 18, 3145-3166.
Giorgi F, Bi X, Pal J (2004) Mean, interannual variability and trends in a regional climate change
experiment over Europe. II: Climate change scenarios (2071-2100). Climate Dynamics, 23,
839-858.
Giorgi F (2006) Climate change hot-spots. Geophysical Research Letters, 33 (8): L08707
Giorgi F, Lionello P (2008) Climate change projections for the Mediterranean region. Global and
Planetary Change, 63, 90-104.
Goetz SJ, Bunn AG, Fiske GJ, Houghton RA (2005) Satellite-observed photosynthetic trends across
boreal North America associated with climate and fire disturbance PNAS, 102, 1352113525.
30
González-Sanpedro MC, Le Toan T, Moreno J, Kergoat L, Rubio E (2008) Seasonal variations of
leaf area index of agricultural fields retrieved from Landsat data. Remote Sensing of
Environment, 112. 810-824
Gower ST, Kucharik CJ, Norman JM (1999) Direct and indirect estimation of leaf area index,
f(APAR), and net primary production of terrestrial ecosystems. Remote Sensing of
Environment, 70, 29-51.
Gracia CA, Sabaté S, Tello E (1998) Modelling the responses to climate change of a Mediterranean
forest managed at different thinning intensities: Effects on growth and water fluxes. Impacts
of Global change on Tree Physiology and Forest Ecosystems 52: 243-252
Gracia C, Tello E, Sabaté S et al. (1999) GOTILWA: an integrated model of water dynamics and
forest growth. In: Ecology of Mediterranean Evergreen Oak Forests (eds. Rodà F, Gracia
CA, Retana J, Bellot J), pp. 163–180. Springer-Verlag, Heidelberg.
Grünzweig JM, Lin T, Rotenberg E, Schwartz A, Yakir D (2003) Carbon sequestration in arid-land
forest. Global Change Biology, 9, 791-799.
Hargreaves GH, Samani ZA (1985) Reference crop evapotranspiration from temperature. Appl.
Eng. Agric., 1, 96-99.
Hättenschwiler S, Miglietta F, Raschi A, Körner C, (1997) Thirty years of in situ tree growth under
elevated CO2: a model for future forest responses? Global Change Biology, 3, 463–471.
Hill J, Stellmes M, Udelhoven T, Röder A, Sommer S (2008) Mediterranean desertification and
land degradation. Mapping related land use change syndromes based on satellite
observations. Global and Planetary Change, 64, 146-157.
Hoerling M, Hurrell J, Eischeid J (2006) Mediterranean climate change and Indian Ocean warming.
Nuovo Cimento della Societa Italiana di Fisica C, 29, 99-104.
Jones PD, Moberg A (2003) Hemispheric and large-scale surface air temperature variations: An
extensive revision and an update to 2001. Journal of Climate, 16, 206-223.
Jump AS, Hunt JM, Peñuelas J (2006) Rapid climate change-related growth decline at the southern
range edge of Fagus sylvatica. Global Change Biology, 12, 2163-2174.
Knipling EB (1970) Physical and physiological basis for the reflectance of visible and near-infrared
radiation from vegetation. Remote Sensing of Environment, 1, 155-159.
Kramer K (2001) Process-based models for scaling up to tree and stand level. In: Long term effects
of climate change on carbon budgets of forests in Europe. (eds Kramer K, Mohren GMJ),
ALTERRA Report 194. ALTERRA, Wageningen, The Netherlands, pp 61-78.
Kyparissis A, Petropoulou Y, Manetas Y (1995) Summer survival of leaves in a soft-leaved shrub
(Phlomis fruticosa L., Labiatae) under Mediterranean field conditions: Avoidance of
photoinhibitory damage through decreased chlorophyll contents. Journal of Experimental
Botany, 46, 1825-1831.
Landsberg JJ, Waring RH (1997) A generalised model of forest productivity using simplified
concepts of radiation-use efficiency, carbon balance and partitioning. Forest Ecology and
Management, 95, 209-228.
Lanfredi M, Simoniello T, Macchiato M (2004) Temporal persistence in vegetation cover changes
observed from satellite: Development of an estimation procedure in the test site of the
Mediterranean Italy. Remote Sensing of Environment, 93, 565-576.
Lapenis A, Shvidenko A, Shepaschenko D, Nilsson S, Aiyyer A (2005) Acclimation of Russian
forests to recent changes in climate. Global Change Biology, 11, 2090-2102.
Lenoir J, Gégout JC, Marquet PA, de Ruffray P, Brisse H (2008) A Significant Upward Shift in
Plant Species Optimum Elevation During the 20th Century. Science 320, 1768 - 1771
Lloret F, Lobo A, Estevan H, Maisongrande P, Vayreda J, Terradas J (2007) Woody plant richness
and NDVI response to drought events in Catalonian (northeastern Spain) forests. Ecology,
88, 2270-2279.
31
López-Moreno JI, Goyette S, Beniston M. (2008) Climate change prediction over complex areas:
Spatial variability of uncertainties and predictions over the Pyrenees from a set of regional
climate models. International Journal of Climatology, 28, 1535-1550.
López-Moreno JI, Vicente-Serrano SM, Gimeno L, Nieto R (2009) The stability of the seasonality
of precipitation in the Mediterranean region: observations since 1950 and projections for the
twenty- first century. Geophysical Research Letters. In press.
Machín J Navas A (1994) The soils of La Plana (Zaragoza), suitable agricultural and forestry uses.
An.Estac. Exp. Aula Dei, 21, 165-172.
MAPA (1978) Mapa de cultivos y aprovechamientos de España. Ministerio de Agricultura, Pesca y
Alimentación. Madrid.
Mäkinen H, Nöjd P, Kahle H-P, Neumann U, Tveite B, Mielikäinen K, Röhle H, Spiecker H (2002)
Large-scale climatic variability and radial increment variation of Picea abies (L.) Karst. in
central and northern Europe. Trees - Structure and Function, 17, 173-184.
Martínez-Villalta J, López BC, Adell N, Badiella L, Ninyerola M (2008) Twentieth century
increase of Scots pine radial growth in NE Spain shows strong climate interactions. Global
Change Biology, 14, 2868-2881.
Maselli F (2004) Monitoring forest conditions in a protected Mediterranean coastal area by the
analysis of multiyear NDVI data. Remote Sensing of Environment, 89, 423-433.
Millington AC, Velez-Liendo XM, Bradley AV (2003) Scale dependence in multitemporal mapping
of forest fragmentation in Bolivia: Implications for explaining temporal trends in landscape
ecology and applications to biodiversity conservation. ISPR Journal of Photogrammetry &
Remote Sensing, 57, 289−299.
Ministerio de Fomento (2007) Geología, geomorfología y edafología. Ministerio de Fomento.
Madrid. 193 pp.
Morales P, Sykes MT, Prentice IC, Smith P, Smith B, Bugmann H, Zierl B et al. (2005) Comparing
and evaluating process-based ecosystem model predictions of carbon and water fluxes in
major European forest biomes. Global Change Biology, 11, 2211-2233.
Myneni RB (1997) Estimation of global leaf area index and absorbed par using radiative transfer
models. IEEE Transactions on Geoscience and Remote Sensing, 35, 1380-1393.
Myneni RB, Tucker CJ, Asrar G, Keeling CD (1998) Interannual variations in satellite-sensed
vegetation index data from 1981-1991. Journal of Geophysical Research D: Atmospheres,
103, 6145-6160.
Nakiceovick N, Grübler A, McDonalds A (1998). Global Energy Perspectives. Cambridge
University press. Cambridge, 299 pp.
Navas A, Machín J (1998) Spatial analysis of gypsiferous soils in the Zaragoza province (Spain),
using GIS as an aid to conservation, Geoderma, 87, 57–66.
Neilson RP (1993) Transient ecotone response to climatic change: some conceptual and modelling
approaches. Ecological Applications, 3, 385-395.
Nemani RR, Keeling CD, Hashimoto H, Jolly WM, Piper SC, Tucker CJ, Myneni RB, Running SW
(2003) Climate-driven increases in global terrestrial net primary production from 1982 to
1999. Science, 300, 1560-1563.
Ninyerola M, Pons X, Roure JM (2000) A methodological approach of climatological modelling of
air temperature and precipitation through GIS techniques. International Journal of
Climatology 20, 1823–1841.
Norby RJ, Wullschleger SD, Gunderson CA, Johnson DW, Ceulemans R (1999) Tree responses to
rising CO2 in field experiments: Implications for the future forest. Plant, Cell and
Environment, 22, 683-714.
Ogaya R, Peñuelas J, Martínez-Vilalta J, Mangirón M (2003) Effect of drought on diameter
increment of Quercus ilex, Phillyrea latifolia, and Arbutus unedo in a holm oak forest of NE
Spain. Forest Ecology and Management, 180, 175-184.
32
Orwing DA, Abrams MD (1997) Variation in radial growth responses to drought among species,
site and canopy strata. Trees, 11, 474-484.
Palà V, Pons X (1995) Incorporation of relief in polynomial-based geometric corrections.
Photogrammetric Engineering and Remote Sensing, 61, 935−944.
Pardina S, Badía D (2004) Propiedades físicas de los suelos de "El Plano (Somontano de
Barbastro)". Geórgica: revista del espacio rural, 10, 39-52.
Peña JL, Pellicer F, Julián A, Chueca J, Echeverría MT, Lozano MV, Sánchez M (2002) Mapa
geomorfológico de Aragón. Consejo de Protección de la Naturaleza de Aragón. 54 pp + 3
mapas.
Piao S, et al. (2008) Net carbon dioxide losses of northern ecosystems in response to autumn
warming. Nature 451. 49-52
Pons X, Ninyerola M (2008) Mapping a topographic global solar radiation model implemented in a
GIS and refined with ground data. International Journal of Climatology, 28, 1821-1834.
Rathgeber C, Nicault A, Guiot J, Keller T, Guibal F, Roche P (2000) Simulated responses of Pinus
halepensis forest productivity to climatic change and CO2 increase using a statistical model.
Global and Planetary Change, 26, 405-421.
Rathgeber C, Nicault A, Kaplan JO, Guiot J (2003) Using a biogeochemistry model in simulating
forests productivity responses to climatic change and [CO2] increase: example of Pinus
halepensis in Provence (south-east France). Ecological Modelling. 166. 239–255.
Repapis CC, Philastras CM (2004) A note on the air temperature trends of the last 100 years as
evidenced in the eastern Mediterranean time series. Theoretical and Applied Climatology,
39, 93-97.
Riaño D, Chuvieco E, Salas J, Aguado I (2003) Assessment of different topographic corrections in
landsat-TM data for mapping vegetation types. IEEE Transactions on Geoscience and
Remote Sensing, 41, 1056−1061.
Riaño D, Moreno Ruiz JA, Isidoro D, Ustin SL (2007) Global spatial patterns and temporal trends
of burned area between 1981 and 2000 using NOAA-NASA Pathfinder. Global Change
Biology, 13, 40-50.
Rouse JW, Hass RH, Schell JA, Deering DW, Harlan JC (1974) Monitoring the the vernal
advancement and retrogradation (greenwave effect) of natural vegetation. NASA/GSFC
type III Final Report, Greenbelt, MD.
Rustad LE, Campbell JL, Marion GM, Norby RJ, Mitchell MJ, Hartley AE, Cornelissen JHC et al.
(2001) A meta-analysis of the response of soil respiration, net nitrogen mineralization, and
aboveground plant growth to experimental ecosystem warming. Oecologia, 126, 543-562.
Ruimy A, Sangier B, Dedieu G (1994) Methodology for the estimation of terrestrial net primary
production from remotely sensed data. Journal of Geophysical Research, D: 5263-5283.
Sabaté S, Gracia CA, Sánchez A (2002) Likely effects of climate change on growth of Quercus ilex,
Pinus halepensis, Pinus pinaster, Pinus sylvestris and Fagus sylvatica forests in the
Mediterranean region. Forest Ecology and Management, 162, 23-37.
Sarris D, Christodoulakis D, Körner C (2007) Recent decline in precipitation and tree growth in the
eastern Mediterranean. Global Change Biology, 13, 1187-1200.
Saxe H, Cannell MGR, Johnsen Ø, Ryan MG, Vourlitis G (2001) Tree and forest functioning in
response to global warming. New Phytologist, 149, 369-400.
Sellers PJ (1985) Canopy reflectance, photosynthesis and transpiration. International Journal of
Remote Sensing, 6, 1335-1372.
Siegel S, Castelan NJ (1988) Nonparametric Statistics for the Behavioral Sciences, McGraw-Hill,
New York.
Slayback DA, Pinzon JE, Los SO, Tucker CJ (2003) Northern hemisphere photosynthetic trends
1982-99. Global Change Biology, 9, 1-15.
33
Sluiter R, De Jong SM (2007) Spatial patterns of Mediterranean land abandonment and related land
cover transitions. Landscape Ecology, 22, 559-576.
Solomon, S. et al., (2007) Climate Change 2007. The physical science basis.. Cambridge University
Press. Cambridge.
Song C, Woodcock CE (2003) Monitoring forest succession with multitemporal Landsat images:
Factors of uncertainty. IEEE Transactions on Geoscience and Remote Sensing, 41,
2557−2567.
Spiecker H, Mielikänen K, Kölh M, Skovsgaard JP (1996) Growth trends in European forests:
studies from 12 countries. European Forest Institute Research Report Nº 5, Springer-Verlag.
Heidelberg, 372 p.
Tardif J, Camarero JJ, Ribas M, Gutiérrez E (2003) Spatiotemporal variability in tree growth in the
Central Pyrenees: Climatic and site influences. Ecological Monographs, 73, 241-257.
Trasobares A, Tomé M, Miina J (2004) Growth and yield model for Pinus halepensis Mill. in
Catalonia, north-east Spain. Forest Ecology and Management, 203, 49-62.
Tucker CJ, Sellers PJ (1986) Satellite Remote Sensing of primary production. International Journal
of Remote Sensing, 7, 1395-1416.
Valladares F, Pearcy RW (1997) Interactions between water stress, sun-shade acclimation, heat
tolerance and photoinhibition in the sclerophyll Heteromeles arbutifolia. Plant, Cell and
Environment, 20, 25-36.
Van Mantgem PJ, Stephenson NL (2007) Apparent climatically induced increase of tree mortality
rates in a temperate forest. Ecology Letters, 10, 909-916.
Vermote EF, Tanré D, Deuzé JL, Herman M, Morcrette JJ (1997) Second simulation of the satellite
signal in the solar spectrum, 6S: an overview. IEEE Transactions on Geoscience and
Remote Sensing, 35, 675−686.
Vicente-Serrano SM, Beguería S (2003) Estimating extreme dry-spell risk in the middle Ebro valley
(Northeastern Spain): A comparative analysis of partial duration series with a General
Pareto distribution and Annual maxima series with a Gumbel distribution. International
Journal of Climatology, 23, 1103-1118.
Vicente-Serrano SM, Lasanta T, Romo A (2004) Analysis of the spatial and temporal evolution of
vegetation cover in the Spanish central Pyrenees: the role of human management.
Environmental Management, 34, 802-818.
Vicente-Serrano SM, Cuadrat JM, Romo A (2006) Aridity influence on vegetation patterns in the
middle Ebro valley (Spain): evaluation by means of AVHRR images and climate
interpolation techniques. Journal of Arid Environments, 66, 353-375.
Vicente-Serrano SM (2007) Evaluating The Impact Of Drought Using Remote Sensing In A
Mediterranean, Semi-Arid Region, Natural Hazards, 40, 173-208.
Vicente-Serrano SM, Cuadrat JM (2007) Trends in drought intensity and variability in the middle
Ebro valley (NE Spain) during the second half of the twentieth century. Theoretical and
Applied Climatology, 88, 247-258.
Vicente-Serrano SM, Lanjeri S, López-Moreno JI (2007) Comparison of different procedures to
map reference evapotranspiration using geographical information systems and regressionbased techniques. International Journal of Climatology, 27, 1103-118.
Vicente-Serrano SM, Pérez-Cabello F, Lasanta T (2008) Assessment of radiometric correction
techniques in analyzing vegetation variability and change using time series of Landsat
images. Remote Sensing of Environment, 112, 3916-3934.
Vicente-Serrano SM, Beguería S, López-Moreno JI, García-Vera MA, Stepanek P (2009) A
complete daily precipitation database for North-East Spain: reconstruction, quality control
and homogeneity. International Journal of Climatology. In press.
34
Vicente-Serrano SM, Beguería S, López-Moreno JI, (2010) A Multi-scalar drought index sensitive
to global warming: The Standardized Precipitation Evapotranspiration Index – SPEI.
Journal of Climate DOI: 10.1175/2009JCLI2909.1
Vila B, Vennetier M, Ripert C, Chadioux O, Liang E, Guibal F, Torre F (2008) Has global change
induced divergent trends in radial growth of Pinus sylvestris and Pinus halepensis at their
bioclimatic limit? The example of the Sainte-baume forest (South-east France). Annals of
Forest Science, 65, 709.
Vogelmann JE, Helder D, Morfitt R, Choate MJ, Merchant JW, Bulley H (2001) Effects of Landsat
5 Thematic Mapper and Landsat 7 enhanced Thematic Mapper Plus radiometric and
geometric calibrations and corrections on landscape characterization. Remote Sensing of
Environment, 78, 55−70.
Voronin PY, Konovalov PV, Bolondinskii VK, Kaipiainen LK, Mao Z-J (2005) Decline in the
primary productivity of Northwestern European forests as a consequence of climate
aridization. Russian Journal of Plant Physiology, 52, 513-517.
Wang Q, Adiku S, Tenhunen J, Granier A (2005) On the relationship of NDVI with leaf area index
in a deciduous forest site. Remote Sensing of Environment 94: 244-255.
Weiß M, Flörke M, Menzel L, Alcamo J (2007) Model-based scenarios of Mediterranean droughts.
Advances in Geosciences, 12, 145-151.
Willmott CJ (1982) Some comments on the evaluation of model performance. Bulletin of the
American Meteorological Society 63: 1309–1313.
Wofsy SC, Goulden ML, Munger JW, Fan S-M, Bakwin PS, Daube BC, Bassow SL, Bazzaz FA
(1993) Net exchange of CO2 in a mid-latitude forest. Science 260: 1314-1317.
Zahawi RA, Augspurger CK (1999) Early plant succession in abandoned pastures in Ecuador.
Biotropica, 31, 540-552.
Zhou L, Tucker CJ, Kaufmann RK, Slayback D, Shabanov NV, Myneni RB (2001) Variations in
northern vegetation activity inferred from satellite data of vegetation index during 1981 to
1999. Journal of Geophysical Research D: Atmospheres, 106, 20069-20083.
35
Date
08/20/1984
08/07/1985
08/13/1987
08/02/1989
08/24/1991
08/10/1992
08/29/1993
08/03/1995
08/24/1997
08/14/1999
08/08/2000
07/26/2001
08/30/2002
08/27/2004
08/14/2005
08/01/2006
Sensor
TM
TM
TM
TM
TM
TM
TM
TM
TM
TM
ETM+
ETM+
ETM+
TM
TM
TM
Table 1. Dates of the Landsat 5-TM and Landsat 7-ETM+ images used in this study.
36
Forest
1
2
3
4
5
6
7
8
Surface area
12991.7
15196.4
5529.4
5178.6
4589.1
10387.8
7182.6
18132.8
Precipitation
626.4
459.9
516.4
432.9
397.8
346.7
377.1
391.7
PET
1118.6
1086.3
1160.6
1111.5
1058.8
1239.3
1174.2
1110.4
Water balance
-492.2
-626.5
-644.3
-678.6
-661.1
-892.7
-797.1
-718.6
Table 2. Surface area (ha) of each P. halepensis forest in the study, and the average values of annual
precipitation, PET and climatic water balance (precipitation minus evapotranspiration).
37
1
2
3
4
5
6
7
8
Negative
(<0.05)
1.1
5.1
5.3
1.7
2.4
17.5
16.9
12.6
Negative
(n.s.)
15
45.4
33.6
16.5
38
59.8
58.8
56.3
Positive
(n.s.)
50.9
43.2
47.8
60.6
51
21.2
21.6
29
Positive
(<0.05) Total
33
6.1
12.8
21.2
8.1
1.4
2.3
2.1
100
100
100
100
100
100
100
100
Table 3. Percentage of the surface area of each forest with positive and negative NDVI trends,
based on the Spearman rho coefficient (19842006). The nonsignificant (n.s.) and
significant trends (p < 0.05) are indicated.
38
Precipitation
September-August
December-August
March-August
PET
April-August
June-August
1984-2006
Rho P-value
0.14
0.52
0.07
0.73
0.12
0.58
R P-value
0.72
0.001
0.52
0.01
1965-2006
Rho P-value
-0.22
0.16
-0.17
0.29
-0.13
0.40
R P-value
0.60 0.0002
0.51
0.004
Table 4. Trends in precipitation and PET in the periods 19842006 and 19652006, calculated
according Spearman rho test.
39
Figure 1. Location of the Ebro valley and the 199-31 Path/Row Landsat image within the Iberian
Peninsula, and the spatial distribution of climatic water balance (precipitation minus PET) in
the region. The eight P. halepensis forests analyzed and the location of the precipitation and
temperature observatories used are also shown.
40
Figure 2. Mean monthly precipitation and PET in each one of the eight analyzed forests. Gray line: average precipitation. Black line: average
PET. Dark gray surface: water deficit. Light gray surface: water surplus.
41
Figure 3. Spatial distribution of the magnitude of the NDVI trends per year in the eight analyzed P.
halepensis forests, obtained from the slope of the least square linear regression analysis
between the NDVI for each pixel and the series of time.
42
b)
a)
Water balance
-400
-600
-800
R = 0.83
P-value = 0.01
-1000
-0.10
-0.05
0.00
0.05
Slope (Standardized anomalies/year)
0.10
Average NDVI (1984-1987)
0.7
0.6
0.5
0.4
0.3
R = 0.58
P-value = 0.13
0.2
-0.10
-0.05
0.00
0.05
0.10
Slope (Standardized anomalies/year)
Figure 4. Relationships among the change in NDVI (in standardized values) in each forest (from the
slope regression), the climatic water balance (precipitation minus PET), and the average
NDVI at the beginning of the period (19841987). Whiskers show one standard deviation
values in each forest.
43
Average NDVI
0.56
a)
0.52
0.48
0.44
0.40
0.36
0.8
R = 0.82
P-value = 0.01
1.0
1.2
1.4
1.6
NDVI Trend (St. anomaly/year)
0.60
1.8
0.15
b)
0.10
0.05
0.00
-0.05
-0.10
-0.10
Modeled LAI
R = 0.71
P-value = 0.05
-0.05
0.00
0.05
0.10
0.15
LAI Trend (St. anomaly/year)
-400
Water balance
c)
-500
-600
-700
-800
-900
-0.05
R = 0.78
P-value = 0.02
0.00
0.05
0.10
0.15
LAI Trend (St. anomaly/year)
Figure 5. Relationships: a) between the LAI modeled by GOTILWA+ and the average NDVI in
each forest between 1984 and 2006, b) between the trends observed in the modeled LAI and
the NDVI, in standardized values, between 1984 and 2006, and c) between the trends in the
LAI and the average annual water balance, obtained by means of the GIS interpolation.
44
-4
-6
2
1970
1980
1990
0
-2
-4
1960
1970
1980
1990
0
-2
1960
2000
Forest 4
Forest 2
2
2
1980
1990
2000
Forest 5
0
-2
-4
1960
2000
1970
St. anomalies
-2
1960
St. anomalies
St. anomalies
0
1970
St. anomalies
4
Forest 1
St. anomalies
St. anomalies
2
1980
1990
2000
1980
1990
2000
2
Forest 3
0
-2
-4
-6
-8
-10
1960
1970
2
1980
1990
2000
1980
1990
2000
Forest 6
0
-2
-4
1960
1970
2
Forest 7
St. anomalies
St. anomalies
4
0
-2
Y vs zona1
-4
1960
1970
1980
1990
2000
2
Forest 8
0
-2
-4
1960
1970
Figure 6. Evolution of the LAI standardized anomalies modeled by GOTILWA+ in the various
forests (19652006). Black points indicate the NDVI standardized anomalies for each year.
45
3
6-months SPEI
SPEI
2
1
0
-1
-2
-3
1960
1970
1980
1990
2000
3
12-months SPEI
SPEI
2
1
0
-1
-2
-3
1960
1970
1980
1990
2000
Figure 7: Evolution of the Standardised Precipitation Evapotranspiration Index at the time scales of
6 and 12 months, obtained from the average series of PET and precipitation in the region.
46
Precipitation (mm.)
PET (mm.)
700
600
500
400
300
200
1960
600
1970
1980
1990
2000
2010
December-August
500
400
300
200
100
0
1960
400
Precipitation (mm.)
September-August
PET (mm.)
Precipitation (mm.)
800
1970
1980
1990
2000
2010
900
875
850
825
800
775
750
725
700
1960
650
625
600
575
550
525
500
475
450
1960
April-August
1970
1980
1990
2000
2010
June-August
1970
1980
1990
2000
2010
March-August
300
200
100
0
1960
1970
1980
1990
2000
2010
Figure 8: Evolution of precipitation and PET in the study area for various periods. Dotted line:
19651984. Black line 19842006. Series were obtained from the average values of the
eight analyzed observatories.
47
1.4
1960
1.7
1980
2000
2020
2040
2060
2080
Forest 3
1.5
1.4
1.50
1.45
1980
2000
2020
2040
2060
2080
1.40
1.35
1.30
1980
2000
1.50
1.45 Forest 7
1.40
1.35
1.30
1.25
1.20
1960 1980 2000
2020
2020
2040
2040
2060
2060
2080
2080
1.15
1.15
2100
Forest 5
1.25
1960
1.50
2100
1.6
1.3
1960
Leaf Area Index
1.6
Leaf Area Index
1.8
1.10
Leaf Area Index
Forest 1
Leaf Area Index
Leaf Area Index
Leaf Area Index
Leaf Area Index
Leaf Area Index
2.0
2100
2100
Forest 2
1.05
1.00
0.95
0.90
1960
1.45
1980
2000
2020
2040
2060
2080
2100
2000
2020
2040
2060
2080
2100
2000
2020
2040
2060
2080
2100
2000
2020
2040
2060
2080
2100
Forest 4
1.40
1.35
1.30
1.25
1960
1980
Forest 6
1.10
1.05
1.00
0.95
1960
1.10
1980
Forest 8
1.05
1.00
0.95
0.90
1960
1980
Figure 9. Leaf Area Index simulations in the eight analysed forests considering observations (19652006) and simulated series according to A2 and B2 scenarios. Black line: observed data. Dark gray:
B2 scenario. Light gray: A2 scenario.
48