Ecological Indicators 60 (2016) 231–236

Contents lists available at ScienceDirect

Ecological Indicators
journal homepage: www.elsevier.com/locate/ecolind

Modeling the ecological niche of long-term land use changes: The role
of biophysical factors
S. Bajocco a,∗ , T. Ceccarelli b , D. Smiraglia a , L. Salvati c , C. Ricotta d
a
Consiglio per la Ricerca in Agricoltura e l’Analisi dell’Economia Agraria, Research Unit for Agricultural Climatology and Meteorology (CRA-CMA), Rome,
Italy
b
Alterra Wageningen UR, Earth Observation and Environmental Informatics Team, Wageningen, Netherlands
c
Consiglio per la Ricerca in Agricoltura e l’Analisi dell’Economia Agraria, Research Centre for Soil-Plant System Studies (CRA-RPS), Rome, Italy
d
University of Rome La Sapienza, Department of Environmental Biology, Rome, Italy

a r t i c l e i n f o a b s t r a c t

Article history: Land use/land cover changes (LULCCs) represent the result of the complex interaction between biophys-
Received 11 February 2015 ical factors and human activity, acting over a wide range of temporal and spatial scales. The aim of this
Received in revised form 20 May 2015 work is to quantify the role of biophysical factors in constraining the trajectories of land abandonment
Accepted 27 June 2015
and urbanization in the last 50 years. A habitat suitability model borrowed from animal ecology was used
Available online 20 July 2015
to analyze the ecological niche of the following LULCC trajectories occurred in Emilia-Romagna (northern
Italy) during 1954–2008: (i) land abandonment (LA) and (ii) urbanization (URB), both from agricultural
Keywords:
areas (URB agr) and from semi-natural areas (URB for). Results showed that the different LULCC trajec-
Biophysical factors
Ecological niche
tories were driven by different combinations of biophysical factors, such as climate, topography and soil
Land abandonment quality. In particular, slope and elevation resulted as the main driving factors for rural processes, while
Land use/land cover change slope and temperatures resulted as the main constraints underlying urban processes. This approach
Urbanization may represent a conceptual and technical step toward the systematic assessment of LULCC processes,
thus providing an effective support tool to inform decision makers about land use transformations, their
underlying causes, as well as their possible implications.
© 2015 Elsevier Ltd. All rights reserved.

1. Introduction
Land use/land cover changes (LULCCs) represent the result of the
complex interaction between socio-economic and cultural condi-
tions, biophysical constraints and land use history (Verburg et al.,
1999); they act over a wide range of temporal and spatial scales
and involve both changes toward different ecological functions
and shifts in intensity within a function, sometimes resulting in a
system’s resilience unbalance (Álvarez Martínez et al., 2011; Shao
et al., 2006; Zurlini et al., 2015). Anthropogenic drivers are rela-
tively active, and are the main determinants of short-term LULCCs.
To the contrary, biophysical factors are relatively stable, and con-
trol the long-term macro-tendency of regional LULCCs (Shao et al.,
2006). While human activity is the major force in shaping LULC,
the underlying biophysical structure of the landscape (i.e. climate,
soil, topography, hydrology, and vegetation) may constrain LULC,
defining the natural capacity or predisposing environmental condi-
tions for land use and influencing the orientations of LULCCs across
∗ Corresponding author. Tel.: +39 06 69531227.
E-mail address: sofia.bajocco@entecra.it (S. Bajocco).
space and time (Jingan et al., 2005; Lambin et al., 2001; Geist et al.,
2006). At finer landscape scales, LULCCs can be related to their con-
straining factors more or less straightforwardly, but aggregating
these changes over broader regions appears as a more difficult task
(Verburg et al., 1999). Therefore, a clear understanding of the driv-
ing forces that cause LULCCs at regional scale is becoming a key
issue (Shao et al., 2006; Van Diggelen et al., 2005; Chaplot et al.,
2005).
In the last decades, the dominant LULCC trajectories observed
at regional scales in the northern Mediterranean Basin are a
biomass increase in marginal areas due to land abandonment
(decreased human pressure) and, on the other hand, the seal-
ing of agricultural soils due to urbanization (increased human
pressure) (Millington et al., 2007; Bajocco et al., 2012). Land aban-
donment (LA) determines a progressive decline in agricultural
land uses due to rural exodus; as a consequence, shrublands and
forests have substantially increased in marginal lands, homoge-
nizing the environmental mosaic, reducing the protection function
of land management, altering the agricultural water use balance
and increasing the biomass available as fuel for potential wildfires
(Moreira et al., 2001; Lozano et al., 2008; Serra et al., 2008; Valipour,
2013, 2014). To the contrary, urbanization (URB) involves the

http://dx.doi.org/10.1016/j.ecolind.2015.06.034
1470-160X/© 2015 Elsevier Ltd. All rights reserved.
232 S. Bajocco et al. / Ecological Indicators 60 (2016) 231–236

wholesale transformation of agricultural and natural ecosystems

to more intensive uses, being among the heaviest anthropogenic
impacts on earth (Seto and Kaufmann, 2003). A pan-European study
highlighted the rapid increase of urbanization particularly in the
form of dispersed urban expansion, considering this as an “ignored
challenge” for the on- and off-site effects on water storage, soil
erosion, loss of organic carbon stock, decrease in biological diver-
sity (Ceccarelli et al., 2014). The environmental consequences of
both LULCCs trajectories entail, on one side, the gradual decrease
in landscape diversity and complexity, and, on the other side, the
increase of vulnerability to certain hazards such as forest fires, ero-
sion, floods, and droughts (Serra et al., 2008).
Understanding the relationships among LULCCs and their con-
straining factors is thus extremely important for enabling scientists,
landscape managers and policy makers to design nature conser-
vation strategies and develop informed and appropriate land use
policies with the aim of preserving some of the unique charac-
teristics of both natural and anthropogenic landscapes (Álvarez
Martínez et al., 2011; Biazin and Sterk, 2013). In this perspective,
regional-scale LULCC modeling represent an essential tool for pre-
dicting the probability of occurrence of a specific change (Röder
et al., 2008) and for identifying which variables provide the most
Fig. 1. Location of the study area.
suitable model (Seabrook et al., 2006; Álvarez Martínez et al., 2011).
A variety of LULCC models have been developed to address dif-
ferent processes, scales of analysis and research questions (Shao 3. Data and methods
et al., 2006; Verburg et al., 2009). Like multiple linear regression,
ordination or clustering methods, multivariate statistical model- 3.1. LULCC data
ing tools can explain the proximate causes of regional LULCCs (Pan
and Bilsborrow, 2005; Purtauf et al., 2005; Lesschen et al., 2005; The land cover maps used in this study (scale 1: 25,000) for
de Freitas et al., 2013). These techniques can be used for two main the years 1954 and 2008 were produced by the Regional Car-
purposes: to project future landscapes under different change sce- tographic Service of Emilia Romagna (http://geoportale.regione.
narios (predictive models) and to explain the relationship between emilia-romagna.it/it/download/). Although the original classifi-
LULCC patterns and driving forces (explorative models) (Millington cation scheme of both maps was not identical, their thematic
et al., 2007; Álvarez Martínez et al., 2011; Millington et al., 2007). In content was harmonized by the Cartographic Service in order
this framework, the present study uses a multivariate exploratory to render them comparable (Ceccarelli et al., 2014). The final
approach borrowed from ecological niche modeling (Hirzel and Le classification scheme follows the third hierarchical level of the
Lay, 2008; Warren and Seifert, 2010) to quantify the role of biophys- CORINE Land Cover nomenclature (http://www.eea.europa.eu/
ical factors in constraining the trajectories of land abandonment publications/COR0-landcover).
and urbanization occurred in the Emilia-Romagna Region (northern For this study the following land cover change trajectories
Italy) in the last 50 years. occurred between 1954 and 2008 were analyzed: (i) land abandon-
ment (LA) and (ii) urbanization (URB), which altogether involved
about 16% of the study area (Fig. 2). LA included all LULCCs
2. Study area from Agricultural areas to more natural land cover types, such as
shrublands and forests (1344.73 km2 ), corresponding to 82.7% of
The Emilia-Romagna region covers an area of 22,446 km2 in the changed surface. URB included all LULCCs from Agricultural
northern Italy and hosts about 4.4 million inhabitants. Nearly areas (URB agr; 230.63 km2 ; 14.2% of the changed surface) and
Emilia Romagna (48%) consists of plains (the Po river valley), while from Semi-natural areas (URB for; 50.98 km2 ; 3.1% of the changed
the remaining 52% is hilly or mountainous (the Apennines range) surface) toward artificial land uses. URB agr and URB for were ana-
stretching for more than 300 km from the north to the south- lyzed separately.
east. Altitude ranges from the sea level along the Adriatic coast,
to 2165 m a.s.l. (Mount Cimone). 3.2. Background data
Artificial areas (17% of the study site) are mainly located in the
lowland. The Po river valley, which is the result of several recla- Seven biophysical background variables that are thought to
mations starting from the Etruscan period and ending just after influence LULCCs in the study area were considered in the analysis:
the World War II, is covered with fruit orchards and annual crops.
The natural landscape has been severely transformed by humans, I. Minimum annual temperature (TMIN)
particularly in the lowlands. At higher elevations the natural veg- II. Maximum annual temperature (TMAX)
etation ranges from oak forests dominated by Quercus pubescens III. Total annual precipitation (PTOT)
and Quercus cerris to Fagus sylvatica forests, sometimes mixed with IV. Aridity index (AI)
Abies alba, and grasslands. V. Elevation (ELE)
The analysis has been performed solely in the hilly-mountain VI. Slope (SLO)
area of the Apennines (10,139.35 km2 with an altitude >500 m VII. Soil quality index (SQI)
a.s.l.) (Fig. 1) in order to limit the impact of socio-economic drivers
related, for instance, to littoralization, tourism, transport and com- The climate variables I–IV (CLIM) were extracted from the
munication routes, that could bias the analysis of the biophysical National Agro-meteorological Database of the Italian Ministry of
factors role over LA, URB agr and URB for trajectories. Agriculture. The database contains daily gridded temperature and
 S. Bajocco et al. / Ecological Indicators 60 (2016) 231–236
Fig. 2. Spatial distribution of the LULCC trajectories analyzed from 1954 to 2008 in the study area.
precipitation data for the period 1960–1990, with a spatial resolu-
tion of 1000 m (see Salvati and Bajocco, 2011). The AI was computed
sensu UNEP (1992); according to this procedure, low index values
indicate arid conditions, whereas high index values indicate humid
conditions.
The topographic variables V–VI (TOPO) were derived from the
digital elevation model (DEM), generated by the Italian Ministry of
Environment, using standard GIS operators at a spatial resolution
of 100 m.
The SQI was derived following the MEDALUS procedure at a spa-
tial resolution of 1000 m integrating information on soil texture,
the nature of parent material, rock cover and soil depth (Kosmas
et al., 1999). Low (high) values of SQI indicate soils of good (bad)
quality.
3.3. Ecological niche factor analysis
The ecological niche factor analysis (ENFA; Hirzel et al., 2002)
was used to analyze the environmental drivers of LULCCs in Emilia
Romagna during 1954–2008. ENFA is a modeling tool formerly used
by ecologists for the exploration of habitat selection by animals
(Gallego et al., 2004; Compton, 2004; Hirzel et al., 2004; Dettki et al.,
2003; Zimmermann, 2004). Generalizing the concept of habitat
selection to every spatially distributed ecological process, the ratio-
nale behind using niche modeling tools for understanding patterns
of LULCCs is based on the observation that a given land use type can
be modeled like a ‘living organism’ with variable preferences for
different resources (i.e. the underlying environmental conditions;
Bajocco et al., 2011; de Angelis et al., 2012).
ENFA is released in the BIOMAPPER package freely available at
http://www.unil.ch/biomapper/. The first type of input data needed
by the ENFA is a set of quantitative raster maps (i.e. the above
biophysical variables I–VII) describing the environmental charac-
teristics of the entire study area (i.e. the available habitat). These
‘background data’ are then compared with the environmental con-
ditions at the locations where the studied species (in our case the
land cover change trajectories LA, URB for and URB agr) have been
detected (i.e. the used habitat).
233
The foremost indicator calculated by ENFA is the so-called
‘marginality’ (Dolédec et al., 2000; Hirzel et al., 2002), a measure of
the distance between the average conditions of the available habitat
and the average conditions of the used habitat. For instance, from a
mathematical viewpoint, the realized niche of a given species is the
density function of the species presences in multivariate ecological
feature space. If the ecological variables are multivariate normally
distributed, the mean vector of the distribution is the point with the
highest species presence and represents the species ecological opti-
mum (Calenge et al., 2005). The marginality is then defined as the
distance of the species optimum from the mean vector of available
habitat conditions over the whole study area, thus measuring the
strength of habitat selection (i.e. the mean deviation, either positive
or negative, of habitat use from availability; Calenge et al., 2005).
The global marginality index usually ranges between 0 and 1
(although in extreme conditions its value can be larger than one)
and its size is proportional to the strength of habitat selection. For
details see Hirzel et al. (2002) and Calenge et al. (2005).
If the landscape cells that are subject to LULCCs were randomly
distributed across the territory, without any clear environmental
driver, the resulting marginality would be zero, meaning that there
is no difference between the mean ecological conditions of the
study area and the ecological conditions at the sites where LUL-
CCs occur. On the other hand, large marginality values indicate that
LULCCs occur preferentially in particular environmental conditions
in relation to the reference area (Basille et al., 2008).
For sake of consistency, to calculate the marginality of the
selected LULCCs in Emilia Romagna, all background variables were
first downscaled to pixel resolution of 1000 m according to the
grain of the variables with coarsest resolution. Next, for analyz-
ing the LA trajectory, since land abandonment typically affects
only areas previously occupied by cultivated fields, the biophysical
characteristics of all landscape cells that changed from agriculture
to natural and semi-natural areas during 1954–2008 (used habi-
tat) were compared with the characteristics of all cells that were
labeled as agriculture in 1954 (available habitat). Likewise, for ana-
lyzing the URB agr trajectory, all cells that changed from agriculture
into artificial areas during 1954–2008 were compared with all cells
234 S. Bajocco et al. / Ecological Indicators 60 (2016) 231–236

Table 1
Mean, minimum and maximum values for each biophysical variable associated to all cells classified as Agriculture in 1954 and for the corresponding LULCC trajectories LA
and URB agr.

Variable Background: agriculture (408,500 ha in 1954) LA trajectory (117,800 ha 1954–2008) URB agr trajectory (19,400 ha 1954–2008)

Mean Min Max Mean Min Max Mean Min Max

AI 1.242 0.992 1.983 1.248 0.999 1.983 1.263 1.005 1.665

ELE (m) 530.2 201 1647 595.282 201 1647 532.938 201 1526
SLO (%) 19.536 0 83.395 23.643 0.236 83.395 15.573 0 58.289
TMIN (◦ C) 6.451 0.982 8.893 6.278 0.982 8.848 6.528 1.543 8.856
TMAX (◦ C) 15.453 11.709 17.617 15.38 11.709 17.53 15.295 12.081 17.478
PTOT (mm) 952.495 791 1371 961.467 793 1371 970.211 795 1223
SQI 1.468 1.149 1.788 1.473 1.172 1.776 1.456 1.216 1.788

Table 2
Mean, minimum and maximum values for each biophysical variable associated to all cells classified as Natural and Semi-natural areas in 1954 and for the corresponding
LULCC trajectory URB for.

Variable Background: natural and semi-natural areas (477,000 ha in 1954) URB for trajectory (5700 ha 1954–2008)

Mean Min Max Mean Min Max

AI 1.286 0.991 1.954 1.291 1.029 1.62

ELE (m) 699.992 201 2046 604.368 206 1555
SLO (%) 26.392 0 106.271 19.778 1.9 43.083
TMIN (◦ C) 6.059 0.977 8.837 6.517 1.947 8.503
TMAX (◦ C) 15.045 11.706 17.59 15.192 12.357 17.133
PTOT (mm) 988.766 791 1358 996.702 851 1218
SQI 1.474 1.121 1.856 1.464 1.216 1.856

labeled as agriculture in 1954. Finally, for analyzing the URB for tra- showed a marginality value of M = 0.264. As shown in Table 3, the
jectory, all cells that changed from natural and semi-natural areas major biophysical drivers of this land cover change trajectory are
into artificial areas during 1954–2008 were compared with all cells slope (SLO = −0.789) and total annual precipitation (PTOT = 0.377).
that were classified as natural and semi-natural areas in 1954. Important variables for the marginality factor are also soil qual-
ity, aridity and maximum temperature while elevation is the least
4. Results contributing variable. Finally, the URB for trajectory showed a
marginality value of M = 0.346. The most relevant biophysical vari-
Tables 1 and 2 show the mean, minimum and maximum val- ables for the marginality factor of the URB for trajectory (Table 3)
ues of each biophysical variable for agriculture and the natural and are minimum temperature (TMIN = 0.507), slope (SLO = −0.706)
semi-natural areas in 1954 and for the analyzed LULCC trajectories. and elevation (ELE = −0.412); while aridity is the least contributing
The coefficients of all ecological variables on the marginality factors variable.
of the three LULCC trajectories are shown in Table 3. The marginal-
ity coefficients range from −1 to +1. The higher the absolute value 5. Discussion
of a coefficient, the larger the difference between habitat use and
availability for the corresponding biophysical variable. Negative In spite of scientific and technological advances, relief, soils and
coefficients mean that LULCCs occur at sites where the biophys- climate still exert a significant influence not only on the agricultural
ical variables have on average lower values than the background and forest uses, but also on their future destinations (Walker, 1977).
mean, while positive coefficients indicate that LULCCs preferen- In this study, the different LULCC trajectories were associated
tially occur at sites where the variables have higher values than the to different combinations of biophysical factors, which reflect rel-
background mean. evant environmental differences between the available and used
The ecological niche factor analysis of the LA trajectory pro- habitat of each trajectory. In this view, not surprisingly, the largest
vided a global marginality of M = 0.274. As shown in Table 3, for overall marginality value is associated to the URB for trajectory,
the LA trajectory, slope (SLO = 0.786) and elevation (ELE = 0.489) meaning that the cells that change from natural to artificial areas
are the biophysical variables with the highest (positive) depar- during 1954–2008 show the largest differences from the mean
ture from the mean background conditions, while minimum and environmental conditions of the corresponding background data
maximum annual temperature are the only variables with a nega- (i.e. the environmental conditions of all cells labeled as natural and
tive departure from the background mean. The URB agr trajectory semi-natural in 1954).
Looking at the marginality coefficients of the single variables,
Table 3 slope represents the most relevant biophysical constraint for all
Marginality values of the LULCC trajectories analyzed in this study, together with LULCC trajectories. This evidence has been confirmed by many
the corresponding coefficients of the input biophysical variables. other studies. For instance, some authors (Felicisimo et al., 2002;
LA URB agr URB for Serra et al., 2008) related the potential distribution of forests to
lithology, latitude, slope, potential insolation and distance from the
Overall marginality 0.274 0.264 0.346
sea, while Álvarez Martínez et al. (2011) found that variables related
TMAX −0.116 −0.261 0.174 to water availability (i.e., distance to rivers, rainfall, slope) and soil
TMIN −0.279 0.128 0.507
PTOT 0.184 0.377 0.113
properties were the most important drivers of forest expansion.
AI 0.077 0.271 0.042 The influence of water availability on vegetation cover patterns
ELE 0.489 0.021 −0.412 was also demonstrated by Vicente-Serrano et al. (2006), who
SLO 0.786 −0.789 −0.706 analyzed the role of aridity on vegetation in a semi-arid Mediter-
SQI 0.109 −0.278 −0.17
ranean region under intense anthropogenic pressure. In terms of
 S. Bajocco et al. / Ecological Indicators 60 (2016) 231–236
topographic variables, Florinsky and Kuryakova (1996) showed that
changes from grassland and shrubland to forest depend on relief
parameters, which controlled the distribution and accumulation of
water and sediment. Bakker et al. (2005) demonstrated that besides
being a major driver of soil erosion, slope is also a direct driver of
land use change because steeper slopes are more difficult to culti-
vate, affecting land abandonment through the loss of nutrients and
water holding capacity.
In detail, the distribution of the abandoned agricultural pix-
els (LA) in multivariate biophysical space is preferentially linked
to annual minimum and maximum temperatures that are both
lower than the background mean. Such thermal conditions are
expressions of cold winters that may, on one hand, harm crops and
cultivations with prolonged frost and, on the other hand, delay the
start of the growing season. Land abandonment tends also to occur
at higher than mean conditions of slope and altitude, where crop-
ping is disadvantaged both for the poor accessibility of the fields
and for the difficulties to cultivate steep soils subjected to increased
erosion and water scarcity. Finally, as expected, the abandoned
agricultural fields are characterized by soils of worse quality (i.e.
higher SQI values) than the mean background conditions.
Urbanization from agriculture (URB agr) tends to occur in more
humid environments and in areas with lower maximum and higher
minimum temperatures than mean background conditions. Fur-
thermore, the URB agr trajectory occupies preferentially flat soils of
better quality with respect to the mean background conditions. To
the contrary, elevation is the least constraining biophysical variable
for this trajectory. These results demonstrate that this urbanization
process is mainly determined by socio-economic choices, which
select the more comfortable and accessible areas at the expense of
the high soil quality that characterizes such locations.
Comparing the two LULCC trajectories starting from the agricul-
tural background, it is interesting to notice that the factors leading
to the different final uses are totally opposite: e.g. cold winters, high
altitudes and steep soils of bad quality are the environmental char-
acteristics underlying the abandonment of cultivated fields; while,
mild winters, cool summers and flat soils of good quality, regard-
less the elevation, represent the main biophysical constraints for
the urbanization process.
Finally, urbanization from deforestation (URB for) showed the
highest global marginality value, meaning that this trajectory
represents the LULCC more constrained by the environmental con-
ditions of the background. In detail, the URB for preferentially
occurs at lower altitudes where minimum and maximum temper-
atures are usually higher than mean background conditions. Like
for URB agr, in semi-natural areas urbanization tends to occur in
the more level and accessible portions of the available habitat, thus
exploiting deeper and higher-quality soils compared to the mean
background conditions.
In summary, as for the agricultural background, the biophysical
conditions, when adverse to crops and cultivations, has a direct role
on the land abandonment process, preventing the long term per-
sistence of the initial land use; to the contrary, even if favorable
to the existence of plantations, the biophysical conditions indi-
rectly drive the change from cultivated fields toward urban uses,
by influencing the land managers choices toward more accessible
and comfortable environments. Also for the urbanization of natural
and semi-natural areas, the role of the biophysical factors is mostly
linked to the accessibility of the background conditions which evi-
dently represents the main driver to the viability of artificial land
uses. In this perspective, given the irreversibility of the soil sealing
processes, the role of the factors guiding urbanization trajectories
should be analyzed at finer scales in order to identify the most
appropriate strategies for a sustainable land management.
In conclusion, the environmental modeling tool used in this
paper enables to explore the main biophysical constraints of
235
landscape dynamics, to quantify to what extent each identified fac-
tor acts and to summarize the main differences between rural and
urban processes. Furthermore, the possibility to repeat the analysis
any time the LULC data are available, with any kind of background
variables, together with the independence of the model from site-
specific information, makes the proposed methodology a potential
standard tool applicable to any type of ecosystem.
The proposed approach can be thus considered a conceptual
and technical step toward the systematic assessment of LULCC pro-
cesses that can be enriched by adding (where available) also data
from the socio-economic domain. It provides a synopsis of major
change processes and their spatial dimension and, therefore, has
the potential to inform decision makers about land use transforma-
tions, their underlying causes, as well as their possible implications.
Future perspectives of the methodology could take into account,
where possible, not only the presence, but also the absence data, in
order to identify the areas where a given LULCC trajectory preferen-
tially occurs, those that are preferentially avoided, and the related
constraining factors.
Acknowledgements
This study has been supported by the Italian Project “Agrosce-
nari – Scenarios of adaptation to climate change in Italian
agriculture” – Research Line 6a.
References
Álvarez Martínez, J.-M., Suárez-Seoane, S., De Luis Calabuig, E., 2011. Modelling the
risk of land cover change from environmental and socio-economic drivers in het-
erogeneous and changing landscapes: the role of uncertainty. Landscape Urban
Plan. 101, 108–119.
Bajocco, S., De Angelis, A., Perini, L., Ferrara, A., Salvati, L., 2012. The impact of land
use/land cover changes on land degradation dynamics: a mediterranean case
study. Environ. Manage. 49, 980–989.
Bajocco, S., Pezzatti, G.B., de Angelis, A., Conedera, M., Ricotta, C., 2011. Bootstrap-
ping wildfire selectivity for the forest types of canton ticino (Switzerland). Earth
Interact. 15, 1–11.
Bakker, M.M., Govers, G., Kosmas, C., Vanacker, V., van Oost, K., Rounsevell, M., 2005.
Soil erosion as a driver of land-use change. Agric. Ecosyst. Environ. 105, 467–481.
Basille, M., Calenge, C., Marboutin, E., Andersen, R., Gaillard, J.-M., 2008. Assessing
habitat selection using multivariate statistics: some refinements of the
ecological-niche factor analysis. Ecol. Model. 211, 233–240.
Biazin, B., Sterk, G., 2013. Drought vulnerability drives land-use and land cover
changes in the Rift Valley dry lands of Ethiopia. Agric. Ecosyst. Environ. 164,
100–113.
Calenge, C., Dufour, A., Maillard, D., 2005. K-select analysis: a new method to analyse
habitat selection in radio-tracking studies. Ecol. Model. 186, 143–153.
Ceccarelli, T., Bajocco, S., Perini, L.L., Salvati, L., 2014. Urbanisation and land take of
high quality agricultural soils – exploring long-term land use changes and land
capability in Northern Italy. Int. J. Environ. Res. 8, 181–192.
Chaplot, V., Giboire, G., Marchand, P., Valentin, C., 2005. Dynamic modelling for lin-
ear erosion initiation and development under climate and land-use changes in
northern Laos. Catena 63, 318–328.
Compton, R.C., (Master’s thesis) 2004. Predicting Key Habitat and Potential Distribu-
tion of Northern Bottlenose Whales (Hyperoodon ampullatus) in the Northwest
Atlantic Ocean. University of Plymouth.
de Angelis, A., Bajocco, S., Ricotta, C., 2012. Modelling the phenological niche of large
fires with remotely sensed NDVI profiles. Ecol. Model. 228, 106–111.
de Freitas, M., Santos, J., Alves, D., 2013. Land-use and land-cover change processes in
the Upper Uruguay Basin: linking environmental and socioeconomic variables.
Landsc. Ecol. 28, 311–327.
Dettki, H., Lofstrand, R., Edenius, L.L., 2003. Modeling habitat suitability for moose
in coastal northern Sweden: empirical vs process-oriented approaches. Ambio
32, 549–556.
Dolédec, S., Chessel, D., Gimaret-Carpentier, C., 2000. Niche separation in community
analysis: a new method. Ecology 81, 2914–2927.
Felicisimo, A.M., Frances, E., Fernandez, J.M., Gondalez-Diez, A., Varas, J., 2002. Mod-
eling the potential distribution of forests with a GIS. Photogramm. Eng. Remote
Sens. 68, 455–462.
Florinsky, I.V., Kuryakova, G.A., 1996. Influence of topography on some vegetation
cover properties. Catena 27, 123–141.
Gallego, D., Cánovas, F., Esteve, M.A., Galián, J., 2004. Descriptive biogeogra-
phy of Tomicus (Coleoptera: Scolytidae) species in Spain. J. Biogeogr. 31,
2011–2024.
236 S. Bajocco et al. / Ecological Indicators 60 (2016) 231–236
Geist, H., McConnell, W., Lambin, E., Moran, E., Alves, D., Rudel, T., 2006. Causes and
trajectories of land-use/cover change. In: Lambin, E., Geist, H. (Eds.), Land-Use
and Land-Cover Change. Springer, Berlin, Heidelberg, pp. 41–70.
Hirzel, A.H., Hausser, J., Chessel, D., Perrin, N., 2002. Ecological-niche factor analysis:
how to compute habitat-suitability maps without absence data? Ecology 83,
2027–2036.
Hirzel, A.H., Le Lay, G., 2008. Habitat suitability modelling and niche theory. J. Appl.
Ecol. 45, 1372–1381.
Hirzel, A.H., Posse, B., Oggier, P.-A., Crettenand, Y., Glenz, C., Arlettaz, R., 2004. Eco-
logical requirements of reintroduced species and the implications for release
policy: the case of the bearded vulture. J. Appl. Ecol. 41, 1103–1116.
Jingan, S., Jiupai, N., Chaofu, W., Deti, X., 2005. Land use change and its corresponding
ecological responses: a review. J. Geogr. Sci. 15, 305–328.
Kosmas, C., Kirkby, M., Geeson, N., 1999. Manual on key indicators of desertifica-
tion and mapping environmentally sensitive areas to desertification. In: The
MEDALUS Project – Mediterranean Desertification and Land Use. European Com-
mission, Brussels.
Lambin, E.F., Turner, B.L., Geist, H.J., Agbola, S.B., Angelsen, A., Bruce, J.W., Coomes,
O.T., Dirzo, R., Fischer, G., Folke, C., George, P.S., Homewood, K., Imbernon, J.,
Leemans, R., Li, X., Moran, E.F., Mortimore, M., Ramakrishnan, P.S., Richards, J.F.,
Skånes, H., Steffen, W., Stone, G.D., Svedin, U., Veldkamp, T.A., Vogel, C., Xu, J.,
2001. The causes of land-use and land-cover change: moving beyond the myths.
Global Environ. Chang. 11, 261–269.
Lesschen, J.P., Verburg, P.H., Staal, S.J., 2005. Statistical Methods for Analyzing the
Spatial Dimension of Changes in Land Use and Farming Systems. The Interna-
tional Livestock Research Institute/LUCC Focus 3 Office, Nairobi/Wageningen.
Lozano, F.J., Suárez-Seoane, S., Kelly, M., Luis, E., 2008. A multi-scale approach for
modeling fire occurrence probability using satellite data and classification trees:
a case study in a mountainous Mediterranean region. Remote Sens. Environ. 112,
708–719.
Millington, J.A., Perry, G.W., Romero-Calcerrada, R., 2007. Regression techniques for
examining land use/cover change: a case study of a Mediterranean landscape.
Ecosystems 10, 562–578.
Moreira, F., Rego, F., Ferreira, P., 2001. Temporal (1958–1995) pattern of change in
a cultural landscape of northwestern Portugal: implications for fire occurrence.
Landsc. Ecol. 16, 557–567.
Pan, W.K.Y., Bilsborrow, R.E., 2005. The use of a multilevel statistical model to analyze
factors influencing land use: a study of the Ecuadorian Amazon. Global Planet.
Change 47, 232–252.
Purtauf, T., Roschewitz, I., Dauber, J., Thies, C., Tscharntke, T., Wolters, V., 2005. Land-
scape context of organic and conventional farms: influences on carabid beetle
diversity. Agric. Ecosyst. Environ. 108, 165–174.
Röder, A., Hill, J., Duguy, B., Alloza, J.A., Vallejo, R., 2008. Using long time series of
Landsat data to monitor fire events and post-fire dynamics and identify driving
factors. A case study in the Ayora region (eastern Spain). Remote Sens. Environ.
112, 259–273.
Salvati, L., Bajocco, S., 2011. Land sensitivity to desertification across Italy: past,
present, and future. Appl. Geogr. 31, 223–231.
Seabrook, L., McAlpine, C., Fensham, R., 2006. Cattle, crops and clearing: regional
drivers of landscape change in the Brigalow Belt, Queensland, Australia,
1840–2004. Landsc. Urban Plan. 78, 373–385.
Serra, P., Pons, X., Saurí, D., 2008. Land-cover and land-use change in a Mediterranean
landscape: a spatial analysis of driving forces integrating biophysical and human
factors. Appl. Geogr. 28, 189–209.
Seto, K.C., Kaufmann, R.K., 2003. Modeling the drivers of urban land use change in the
Pearl River Delta, China: integrating remote sensing with socioeconomic data.
Land Econ. 79, 106–121.
Shao, J., Wei, C., Xie, D., 2006. An insight on drivers of land use change at regional
scale. Chin. Geogr. Sci. 16, 176–182.
UNEP, 1992. World Atlas of Desertification. Edward Arnold, London.
Valipour, M., 2013. Necessity of irrigated and rainfed agriculture in the world. Irrig.
Drain. Syst. Eng. S9, e001, http://dx.doi.org/10.4172/2168-9768.S9-e001
Valipour, M., 2014. Land use policy and agricultural water management of the previ-
ous half of century in Africa. Appl. Water Sci., http://dx.doi.org/10.1007/s13201-
014-0199-1
Van Diggelen, R., Sijtsma, F.J., Strijker, D., van den Burg, J., 2005. Relating land-
use intensity and biodiversity at the regional scale. Basic Appl. Ecol. 6,
145–159.
Verburg, P.H., de Koning, G.H.J., Kok, K., Veldkamp, A., Bouma, J., 1999. A spatial
explicit allocation procedure for modelling the pattern of land use change based
upon actual land use. Ecol. Model. 116, 45–61.
Verburg, P.H., van de Steeg, J., Veldkamp, A., Willemen, L., 2009. From land cover
change to land function dynamics: a major challenge to improve land charac-
terization. J. Environ. Manage. 90, 1327–1335.
Vicente-Serrano, S.M., Cuadrat-Prats, J.M., Romo, A., 2006. Aridity influence on vege-
tation patterns in the middle Ebro Valley (Spain): evaluation by means of AVHRR
images and climate interpolation techniques. J. Arid Environ. 66, 353–375.
Walker, M., 1977. Agricultural Location. Basil Blackwell, Oxford.
Warren, D.L., Seifert, S.N., 2010. Ecological niche modeling in Maxent: the impor-
tance of model complexity and the performance of model selection criteria. Ecol.
Appl. 21, 335–342.
Zimmermann, F., (PhD Thesis) 2004. Conservation of the Eurasian lynx (Lynx lynx) in
a Fragmented Landscape—Habitat Models, Dispersal and Potential Distribution.
Faculte de biologie et de medecine de l’Universite de Lausanne, Departement
d’Ecologie et Evolution, Lausanne.
Zurlini, G., Li, B.L., Zaccarelli, N., Petrosillo, I., 2015. Spectral entropy, ecological
resilience, and adaptive capacity for understanding, evaluating, and managing
ecosystem stability and change. Global Change Biol. 21, 1377–1378.