Ecological Modelling xxx (xxxx) xxx

Contents lists available at ScienceDirect

Ecological Modelling
journal homepage: www.elsevier.com/locate/ecolmodel

High spatial resolution bioclimatic variables to support ecological

modelling in a Mediterranean biodiversity hotspot
Erika Bazzato a, Leonardo Rosati b, Simona Canu c, Michele Fiori d, Emmanuele Farris e,
Michela Marignani a, *
a
Department of Life and Environmental Sciences - Botany Division, University of Cagliari, Via Sant’Ignazio da Laconi, 13, 09123 Cagliari CA, Italy
b
School of Agriculture, Forestry, Food and Environment, Via dell’Ateneo Lucano, 10, University of Basilicata, 85100 Potenza, Italy
c
Territorial Department of Sassari and Gallura, Environmental Protection Agency of Sardinia (ARPAS), Via Rockfeller 58/60, 07100 Sassari SS, Italy
d
Department of Meteorology and Climatology, Environmental Protection Agency of Sardinia (ARPAS), Viale Porto Torres 119, 07100 Sassari SS, Italy
e
Department of Chemistry and Pharmacy, University of Sassari, Via Piandanna, 4, 07100 Sassari SS, Italy

A R T I C L E I N F O A B S T R A C T

Keywords: Understanding the effects of climate on biodiversity and its different levels of response to climatic variation is
Sardinia important for addressing conservation-based questions: the use of bioclimatic variables and species modelling
WorldClim tools is common in environmental, agricultural and biological sciences. Unfortunately, most of the ecological
Species distribution model
local studies are limited to the use of global data with coarse spatial resolutions, while fine-grain climate data are
Seasonal climatic variations
Data reliability
necessary to capture environmental variability and perform reliable modelling. We propose a high-resolution
dataset (40 m grid) of the suite of original coarse-grain bioclimatic variables proposed by WorldClim 2 for the
island of Sardinia (Italy); variations amongst our dataset and WorldClim 2 were calculated and mapped to show
the spatial distribution of differences between all pairs of variables.
We observed relevant differences for the bioclimatic variables related to rainfall (mean RMSE = 39.79; mean
nRMSE = 0.21) compared to the temperature ones (mean RMSE = 4.81; mean nRMSE = 0.11). Moreover,
discrepancies are not evenly distributed in the territory: the greater differences correspond to the areas char­
acterized by complex orographic systems.
Results recommend caution in making ecological assessments based on bioclimatic variables derived from
global data with coarse spatial resolutions in physiographically complex landscapes, especially in the Mediter­
ranean regions, characterized by seasonal climatic variations and high levels of biodiversity and biogeographical
complexity.
These new data will support a new generation of research studies in a broad array of ecological applications at
a much finer scale than previously possible.

1. Introduction Svenning, 2015), especially for those with high dispersal capacities like
marine invertebrates, birds, and insects (Parmesan, 2006).
Climate varies across space and species can shift their distribution in Bioclimatic variables, unlike climate data, are developed focusing on
order to find appropriate climatic conditions where they can live suit­ relevant combination of variables, considering biotic thresholds; hence
ably (Bellard et al., 2012). In the same way, climate fluctuations drive they better describe, and predict, the response of living organisms
the ecological changes in species, populations, ecological networks and (Jennings and Harris, 2017; Rivas-Martínez et al., 2011).
ecosystems functions and processes (Parmesan, 2006). Climate variation In the middle 1980s, the earliest computer-based methods were
over time, including year-to-year variability, has been linked to a shift in developed for estimating mean climate conditions of a given site on
phenology and physiology of plants and animals (Bellard et al., 2012; Earth’s surface, by using point location data sets (Sutherst and May­
Parmesan, 2006); moreover, also latitudinal and altitudinal range shifts wald, 1985) or spatially local gridded climate data (e.g. Booth et al.,
are well documented for a wide number of species (Lenoir and 1987). Following the development of more sophisticated and complex

* Corresponding author.
E-mail address: marignani@unica.it (M. Marignani).

https://doi.org/10.1016/j.ecolmodel.2020.109354
Received 11 August 2020; Received in revised form 17 October 2020; Accepted 1 November 2020
0304-3800/© 2020 Published by Elsevier B.V.

Please cite this article as: Erika Bazzato, Ecological Modelling, https://doi.org/10.1016/j.ecolmodel.2020.109354
E. Bazzato et al. Ecological Modelling xxx (xxxx) xxx

spatial interpolation methods (Hutchinson and Gessler, 1994), mod­ suite of bioclimatic variables proposed by WorldClim 2 (Fick and Hij­
ellers have rapidly built spatially gridded climatologies, appropriately mans, 2017), one of the most used dataset in ecological modelling
scaled on land elevation (Hutchinson, 1995). Subsequently, spatially (Marchi et al., 2019).
interpolated gridded climate data have become available for re­ We calculated the suite of 19 bioclimatic variables using a high-
searchers, improving environmental information in sites where there resolution monthly climatologies of temperature and precipitation of
was a lack of local data (Hijmans et al., 2005). Sardinia, based on long-term climate time series and local topography.
Open data on gridded bioclimate datasets, which differ in their To assess the differences amongst our fine-grain dataset and the
quality over time, space and resolution (from 30 s ~1 km2 to 10 min original coarse-grain bioclimatic variables of WorldClim 2, we per­
~340 km2 at the equator), are for example WorldClim (Fick and Hij­ formed a quantitative comparison and spatial distribution of errors be­
mans, 2017), MerraClim (C. Vega et al., 2017), CHELSA (Karger et al., tween all pairs of variables of these datasets.
2017), CliMond (Kriticos et al., 2012), EuMedClim (Fréjaville and The high-resolution data produced can be particularly suited for
Benito Garzón, 2018) and ENVIREM (Title and Bemmels, 2018). Most of studying species distributions under current conditions, improving
these global datasets consist of monthly average temperature (mini­ ecological studies at finer spatial scales.
mum, maximum and medium), monthly precipitation and solar radia­
tion assessed across a large temporal range, as well as bioclimatic 2. Study area
variables. Bioclimatic variables, originally devised by Nix (1986) and
deriving from the monthly temperature and rainfall values, describe The island of Sardinia, one of the two largest Mediterranean islands,
annual trends (e.g., mean annual temperature and precipitation), sea­ is located in the middle of the western Mediterranean Basin and covers a
sonal trends (e.g., annual range in temperature and precipitation) and surface area of around 24,000 km2 with a coastline of about 1900 km,
extreme or limiting environmental factors (e.g., temperature of the marked by a variety of landforms (cliffs, sandy dunes, long or pocket
coldest and warmest month, and precipitation of the wet and dry beaches). Due to its large extension, the territory is characterized by a
quarters). complex orographic pattern with hilly lands, plateaus, mountain and
Considering their peculiar characteristics, bioclimatic variables were plains (Fig. 1), placed on heterogeneous geological substrata for age and
considered suitable for studying species distributions, under current or typology.
possible future conditions, using species distribution modelling (SDM) More than 600 formations and more second-rank lithostratigraphic
tools (Kriticos et al., 2012). The use of bioclimatic variables and species units have been recognized (Carmignani et al., 2016): Palaeozoic
modelling tools have thus found a widespread use in environmental, magmatic intrusive units and metamorphic complexes related to Her­
agricultural and biological sciences (Booth et al., 2014; Di Febbraro cynian Orogenesis; sedimentary successions linked to Mesozoic and
et al., 2018; Guisan and Thuiller, 2005; Pecchi et al., 2019): assessing Tertiary marine transgression; volcano-sedimentary successions related
the environmental niche of species or their invasion and proliferation; to the opening of the Tyrrhenian Sea; Quaternary deposits of various
quantifying the impact of climate and other environmental changes on origin (alluvial, aeolian, lacustrine, littoral and slope movement-related)
species distributions; modelling species assemblages from individual covering the previous geological formations.
species predictions; testing biogeographical, ecological and evolu­
tionary hypotheses; identifying sites of high potential of occurrence for
rare species; developing strategies and action plans to ensure a
long-term conservation of species.
For many applications, fine spatial grain climate data is considered
necessary to capture environmental variability, especially in physio­
graphically complex landscapes (Hijmans et al., 2005); for example,
they are preferable to study distribution of species with low-dispersal
ability (Chust et al., 2004; Franklin et al., 2013; Guisan et al., 2007),
species corridors and effects of barriers, or for others detailed ecological
or conservation studies (Elith and Leathwick, 2009; Hess et al., 2006;
Nezer et al., 2017). Fine-grain climate grids are able to detect potential
microrefugia (Hannah et al., 2014; Meineri and Hylander, 2017), i.e.
sites with peculiar microclimates that support populations of species
outside their main distribution area. Microrefugia are thus particularly
relevant to understand the spatial distribution of species in response to
climate change (Dobrowski, 2011) and the demographic and genetic
performance of populations at the periphery of their range (Papuga
et al., 2018; Pironon et al., 2017).
Unfortunately, fine-grain climate grids are only available for limited
parts of the world (Hijmans et al., 2005) and most of the ecological local
studies are limited to and by the use of global data with coarse spatial
resolutions.
The development of high spatial resolution bioclimatic data is
particularly important in the Mediterranean basin, one of the 35
terrestrial biodiversity hot spots of the world (Medail, 2017), where
climate-driven habitat loss was recognized as a major threat to biodi­
versity (Barredo et al., 2016). Nevertheless, as for Sardinia, the
second-largest island of the Mediterranean basin, several studies to
assess the distribution of plants (e.g. Casazza et al., 2014; Fois et al.,
2018a, 2018b; Ongaro et al., 2018) or animals (e.g. Iannella et al., 2019;
Russo et al., 2014; Sýkora et al., 2017) relied on coarse-grain bioclimatic
open data such as Worldclim. To fill this gap in Sardinia, we propose a Fig. 1. Sardinia is the second main island in the Mediterranean and it is
novel high-resolution dataset (40 m grid, equal to ~ 1.69 arcsec) of the characterised by a complex orographic pattern.

2
E. Bazzato et al. Ecological Modelling xxx (xxxx) xxx

The climate is typically Mediterranean, with mild and poorly rainy Table 1
winters, warm and dry summers. Recent detailed bioclimate mapping, Bioclimatic variables, types, descriptions and units of the new high-resolution
using the bioclimatic classification of Rivas-Martínez et al. (2011), bioclimatic variables of Sardinia (Italy).
identified that the island is characterized by two macrobioclimates Filename Type of variable Description Unit
(Mediterranean pluviseasonal oceanic and Temperate oceanic), four BIO01.tif Temperature-related Annual Mean Temperature ◦
C
classes of continentality (from weak semihyperoceanic to weak sub­ variable
continental), eight thermotypic horizons (from lower thermomedi­ BIO02.tif Temperature-related Mean Diurnal Range (Mean of ◦
C
terranean to upper supratemperate) and seven ombrothermic horizons variable monthly (max temp - min
temp))
(from lower dry to lower hyperhumid), whose combination resulted in
BIO03.tif Temperature-related Isothermality (BIO02/BIO07) (x Index
43 different isobioclimates (Canu et al., 2015). variable 100)
The heterogeneous climate, morphology and geological substrata of BIO04.tif Temperature-related Temperature Seasonality Index
the island determine a high rate of endemism (Fois et al., 2017) and a variable (standard deviation x 100)
wide variety of Potential Natural Vegetation sensu Farris et al. (2010), BIO05.tif Temperature-related Maximum Temperature of ◦
C
variable Warmest Month
described in detail by Bacchetta et al. (2009). BIO06.tif Temperature-related Minimum Temperature of ◦
C
variable Coldest Month
3. Methods BIO07.tif Temperature-related Temperature Annual Range ◦
C
variable (BIO05-BIO06)
BIO08.tif Temperature-related and Mean Temperature of Wettest C
Monthly average temperatures (minimum, maximum and mean) and
◦

rainfall-related variable Quarter

precipitations were originally interpolated to produce the bioclimatic BIO09.tif Temperature-related and Mean Temperature of Driest ◦
C
map of Sardinia (Canu et al., 2015). The data at 40 m resolution were rainfall-related variable Quarter
created using high quality meteorological data from 203 rain gauges and BIO10.tif Temperature-related Mean Temperature of Warmest ◦
C
68 temperature gauges of the regional climatic database of the Weather variable Quarter
BIO11.tif Temperature-related Mean Temperature of Coldest ◦
C
and Climate Department (ARPA Sardegna) for the time period variable Quarter
1971–2000. Monthly average temperature and precipitation were BIO12.tif Rainfall-related variable Annual Precipitation mm
interpolated by Regression Kriging, combining a Multiple Linear BIO13.tif Rainfall-related variable Precipitation of Wettest Month mm
Regression with an Ordinary Kriging of the regression residuals. Factors BIO14.tif Rainfall-related variable Precipitation of Driest Month mm
BIO15.tif Rainfall-related variable Precipitation Seasonality Index
such as latitude, longitude, altitude, sea distance and local topography
(Coefficient of Variation)
were considered as independent geographic variables to account for BIO16.tif Rainfall-related variable Precipitation of Wettest Quarter mm
topographic effects (Canu et al., 2015). BIO17.tif Rainfall-related variable Precipitation of Driest Quarter mm
Starting from this baseline data, we calculated bioclimatic variables BIO18.tif Temperature-related and Precipitation of Warmest mm
at 40 m resolution using the C++ code included in the System for rainfall-related variable Quarter
BIO19.tif Temperature-related and Precipitation of Coldest Quarter mm
Automated Geoscientific Analysis (SAGA) version 7.5.0 (Conrad et al.,
rainfall-related variable
2015). The free and open-source Geographical Information System
SAGA under the GNU public license was specifically developed for
regional climate and environmental modelling applications (Conrad Spearman’s correlation coefficient (ρ), the root mean square error
et al., 2015). (RMSE) and the normalized root mean square error by the mean
In order to evaluate the extreme or limiting environmental factors, (nRMSE) between all pairs of variables.
we defined the quarterly parameters by following the definitions pro­ For each variable we mapped the spatial distribution of errors, by
vided by WorldClim (Hijmans et al., 2005) and ANUCLIM (Xu and means of the difference between the two datasets, namely the new high
Hutchinson, 2013). spatial resolution dataset minus WorldClim 2.
Three types of bioclimatic variables were evaluated (Table 1): vari­ All data manipulation and geographic analyses were performed with
ables related to temperature (BIO01-BIO07 and BIO10-BIO11); variables R (R Core Team, 2020), using raster (Hijmans, 2020) and gdalUtils
related to rainfall (BIO12-BIO17); variables related to both temperature (Greenberg and Mattiuzzi, 2020) packages. Metadata of rasters were
and rainfall (BIO08-BIO09 and BIO18-BIO19). added using ArcGIS software by Esri.
The calculation of bioclimatic variables related to temperature was
performed using average monthly maximum, minimum and mean 4. Results
temperatures. Cell-by-cell calculations of bioclimatic variables related to
rainfall were conducted using monthly average precipitation. We generated a high-resolution suite of 19 bioclimatic variables of
Some descriptive statistics such as mean, minimum and maximum Sardinia: all rasters are provided at roughly 1.69 arcsec (40 m cell size)
were used to describe the results. To assess the rate of dispersion of data, resolution and in the WGS84 geographic coordinate system (EPSG
for each bioclimatic variable we calculated the coefficient of variation 4326). GeoTIFF rasters of all 19 bioclimatic variables of Sardinia (Italy)
(in percentage). are included in Annex I. Metadata files include file name, thumbnail,
tags and description for all rasters.
3.1. Comparisons with the WorldClim 2 bioclimatic variables The coefficient of variation (in%) of bioclimatic variables was lower
in the temperature-related variables (mean CV = 11.39), higher for the
Comparison amongst WorldClim 2 and the new high-resolution precipitation-related ones (mean CV = 23.79) and intermediate in the
bioclimatic variables of Sardinia was possible because they were based variables related to both temperature and precipitation (mean CV =
on the same temporal range: 1970–2000 for WorldClim 2 (Fick and 21.33). In particular, the maximum coefficient of variation of the tem­
Hijmans, 2017) and 1971–2000 for Sardinia (Canu et al., 2015). perature ones amounts to 33.78 for the Minimum Temperature of
Variables were compared following three main steps: (i) at first, we Coldest Month (BIO06), with BIO10 having the minimum value (CV =
resampled WorldClim 2 data to the resolution of our variables (40 m) 4.02, Table 2). Within rainfall-related bioclimatic variables, the Pre­
using the nearest neighbour method; (ii) then we checked errors in raster cipitation of Driest Month (BIO14) had a higher variation (CV = 44.88)
alignment and adjusted alignment using the nearest neighbour method than other variables, representing the highest level of variability in all
and one of our raster as snap raster, to ensure all cells were properly the dataset. Regarding variables related to both temperature and pre­
aligned; (iii) finally we performed the quantitative comparison analyses. cipitation the Precipitation of Warmest Quarter (BIO18) showed the
To assess if and where the two datasets are different, we calculated highest level of variation (CV = 33.77), while the Mean Temperature of

3
E. Bazzato et al. Ecological Modelling xxx (xxxx) xxx

Table 2 (ρ > 0.60). On the contrary, the precipitation of the driest and warmest
Mean, minimum, maximum and coefficient of variation (CV) values for each of quarters highlighted a low correlation (ρ < 0.60).
the new high spatial resolution bioclimatic variables of Sardinia (Italy). The spatial distribution of errors, calculated as the difference be­
Variable Name of variable Mean Minimum Maximum CV tween the two datasets i.e. new high-resolution dataset minus World­
(%) Clim2, showed the heterogeneous distribution of the spatial
BIO01 Annual Mean 15.44 8.61 18.12 8.51 discrepancies of the variables (Figs. 5–7) and a specific pattern accord­
Temperature ing to the different bioclimatic variable analysed. In these figures red
BIO02 Mean Diurnal Range 9.69 4.06 13.46 13.64 colours indicate areas where a given variable was overestimated by
(Mean of monthly (max
WorldClim 2, blue colours the underestimated ones.
temp - min temp))
BIO03 Isothermality (BIO02/ 37.08 22.71 43.69 7.99 For example, the spatial distributions of the differences for the
BIO07) (x 100) Annual Mean Temperature (BIO01) highlighted lower values modelled
BIO04 Temperature 571.91 461.06 655.64 4.98 by WorldClim 2 compared to our dataset in the mountain areas, with
Seasonality (standard differences of more than 2 ◦ C (Fig. 5). Maximum/minimum tempera­
deviation x 100)
BIO05 Maximum Temperature 30.59 24.16 33.96 4.40
tures (BIO05/06/10) are generally underestimated by WorldClim 2,
of Warmest Month with peaks up to 7 ◦ C for the maximum temperature of the warmest
BIO06 Minimum Temperature 4.57 − 2.13 9.79 33.78 month (BIO05). Accordingly, the annual range of extreme temperature
of Coldest Month conditions (BIO07) was underestimated by WorldClim 2 in the internal
BIO07 Temperature Annual 26.02 17.45 31.16 7.25
and mountain areas of the island, being overestimated in coastal areas in
Range (BIO05-BIO06)
BIO08 Mean Temperature of 11.85 3.22 17.36 22.54 the north, south and eastern coast (Fig. 5).
Wettest Quarter Regarding rainfall-related bioclimatic variables, a markedly different
BIO09 Mean Temperature of 23.54 17.43 26.16 5.27 spatial distribution was observed for the Annual precipitation (BIO12),
Driest Quarter showing a gradient moving from north-west to south-east: in the NW
BIO10 Mean Temperature of 24.30 18.90 26.16 4.02
areas WorldClim 2 overestimated, while the SE areas are under­
Warmest Quarter
BIO11 Mean Temperature of 8.60 0.98 12.29 17.98 estimated, with strong differences (higher than 500 mm) (Fig. 6). Pre­
Coldest Quarter cipitation Seasonality (Coefficient of Variation, BIO15) is strongly
BIO12 Annual Precipitation 690.58 418.54 1376.38 19.22 overestimated by WorldClim 2 in the internal and mountain areas and
BIO13 Precipitation of Wettest 100.73 57.82 209.09 19.23
slightly underestimated along the S-E coast. The Precipitation of Driest
Month
BIO14 Precipitation of Driest 8.21 0.46 22.32 44.88 Quarter (BIO17) is generally overestimated, in particular in the NW
Month zones. An asymmetry similar to the one observed for BIO12 was detected
BIO15 Precipitation 50.23 39.75 60.48 7.28 for the Mean Temperature of Wettest Quarter (BIO08), with under­
Seasonality estimated values in the western part of the island and overestimated
(Coefficient of
ones in the eastern ones (Fig. 7). Furthermore, a general overestimation
Variation)
BIO16 Precipitation of Wettest 288.68 164.44 582.96 19.26 was observed for the Precipitation of Warmest Quarter (BIO18) in all
Quarter western areas of the island, from north to south, up to 40 mm (Fig. 7).
BIO17 Precipitation of Driest 36.71 11.24 82.44 32.89
Quarter
5. Discussions
BIO18 Precipitation of 39.97 11.24 90.93 33.77
Warmest Quarter
BIO19 Precipitation of Coldest 222.29 113.18 495.01 23.73 Bioclimatic variables are fundamental for understanding and
Quarter modelling the ecological processes and the distribution of biodiversity of
earth (Jennings and Harris, 2017; Rivas-Martínez et al., 2011). Never­
theless, our study demonstrated that comparing the global dataset to a
Driest Quarter (BIO09) showed the lowest one (CV = 5.27).
high spatial resolution one, revealed that we should pay attention on the
The spatial distribution of the calculated bioclimatic variables is
accuracy of coarse spatial resolutions data, especially in areas of high
shown according to the three groups temperature (Fig. 2), rainfall
heterogeneity where weather stations are few and sparsely distributed
(Fig. 3), temperature and rainfall related variables (Fig. 4).
(Sandoval et al., 2020), like the Mediterranean area.
We observed that the discrepancies existing amongst our high spatial
4.1. Comparisons with the WorldClim 2 bioclimatic variables resolution dataset and WorldClim 2 are evident and each bioclimatic
variable behaved in a different way: we did not detect a general over/
The comparison of our high-resolution dataset vs. WorldClim 2 in underestimation pattern (or trend) of the bioclimatic variables, but we
terms of Spearman’s correlation coefficient (ρ) showed significant linear rather observed variable-specific patterns mainly linked to the local
correlations (all p-values < 0.001) for all the 19 bioclimatic variables orographic conditions and to the direction of the dominant winds
(Table 3) with the highest correlation amongst BIO11 values and the driving weather perturbations.
lowest for BIO15. The new high spatial resolution dataset compared to WorldClim 2
The normalized root mean square error (nRMSE) revealed relevant showed that the larger discrepancies were spotted in the bioclimatic
differences (Table 3), in particular for the bioclimatic variables related variables related to precipitation. Those inconsistencies are not evenly
to rainfall showing a higher discrepancy (mean RMSE = 39.79; mean distributed in the territory: the greater differences between the two
nRMSE = 0.21) compared to the temperature ones (mean RMSE = 4.81; datasets correspond to the areas characterized by complex orographic
mean nRMSE = 0.11). systems. Moreover, since the dominant air mass perturbations in Sar­
All bioclimatic variables related to temperature showed a high cor­ dinia come from west and the rain shadow effect is not considered in the
relation (ρ > 0.70) with WorldClim 2 (Table 3). model, the global dataset, probably limited by an uneven distribution of
Rainfall-related bioclimatic variables were less strongly correlated meteorological stations, strongly underestimated the annual precipita­
with WorldClim 2 than temperature-related ones; Seasonality trend of tion in the eastern zones and overestimated the annual and the summer
precipitation (BIO15) was poorly correlated (ρ = 0.42). precipitations (i.e. driest and warmest quarter) in the western zones (up
With regard to bioclimatic variables related to both temperature and to 40 mm). These discrepancies, if applied, for example, to vascular
precipitation, mean Temperature of Wettest and Driest quarters showed plant species distribution models, can cause biases in the comprehension
good correlations with corresponded WorldClim 2 bioclimatic variables of the distribution of thermo-xerophilous species particularly (or

4
E. Bazzato et al. Ecological Modelling xxx (xxxx) xxx

Fig. 2. Temperature-related bioclimatic variables (BIO01-BIO07 and BIO10-BIO11) of Sardinia (Italy).

Fig. 3. Rainfall-related bioclimatic variables (BIO12-BIO17) of Sardinia (Italy).

exclusively) abundant in the western coast of the island (like Chamaerops especially in isolated mountainous areas, were indicated for the first and
humilis, Polygala rupestris, Viola arborescens, amongst the others (Biondi second versions of this dataset by Hijmans et al. (2005) and Fick and
et al., 2001), and, contrarily, of mesophilous species of Hijmans (2017), respectively. In particular for Italy, Pesaresi et al.
non-Mediterranean origin that can surprisingly colonize low elevation (2014, 2017) highlighted that the lower accuracy in precipitation spa­
(down to the sea level) in the eastern coast, like Ostrya carpinifolia tialization of WorldClim, could be explained by the scarcity of meteo­
(Bacchetta et al., 2004a) and Taxus baccata (Farris et al., 2012). rological stations density respect to the topographic complexity and
Similar limitations of the WorldClim spatial dataset accuracy, heterogeneity of the Italian territories. Bedia et al. (2013) highlighted

5
E. Bazzato et al. Ecological Modelling xxx (xxxx) xxx

Fig. 4. Bioclimatic variables of Sardinia (Italy) related to both temperature and precipitation (BIO08-BIO09 and BIO18-BIO19).

of climate change scenarios at regional or local scales. This can even­

Table 3 tually compromise the successful use of models in biodiversity conser­
Pairwise comparison of the new high-resolution bioclimatic variables and
vation and management actions.
WorldClim 2 bioclimatic variables in terms of Spearman’s correlation coefficient
The maximum/minimum temperature of the warmest/coldest
(ρ) and normalized root mean square error (nRMSE).
month are underestimated, i.e. compared to the high spatial resolution
Variable Name of variable Spearman’s RMSE nRMSE
dataset, the global dataset generally models cooler summer maximum
rho
(up to 7 ◦ C) and colder winter minimum (more than 4 ◦ C). Accordingly,
BIO01 Annual Mean Temperature 0.94 0.49 0.03 the coarse scale dataset models a smaller temperature annual range in
BIO02 Mean Diurnal Range (Mean of 0.76 1.00 0.10
monthly (max temp - min temp)
the mountains and a wider one on the coasts, underrating the con­
BIO03 Isothermality (BIO02/BIO07) (x 0.72 2.22 0.06 tinentality values in the internal areas and exaggerating it on the coasts.
100) These discrepancies are particularly important in the Mediterranean
BIO04 Temperature Seasonality 0.76 31.10 0.05 climate, since the seasonal distribution of rainfall and the extreme
(standard deviation x 100)
temperatures determine the limits for species survival. Since Sardinia
BIO05 Maximum Temperature of 0.77 3.00 0.10
Warmest Month shows many plains and depressed areas in the internal parts of the is­
BIO06 Minimum Temperature of 0.93 2.04 0.45 land, often surrounded by hills or mountains, it is of crucial importance
Coldest Month to discriminate areas with higher temperature annual range (i.e. more
BIO07 Temperature Annual Range 0.79 1.47 0.06 continental) from those characterized by a smaller temperature annual
(BIO05-BIO06)
range (i.e. more oceanic). In the more continental areas, no matter the
BIO08 Mean Temperature of Wettest 0.77 1.90 0.16
Quarter altitude above the sea level, species like Arbutus unedo L., Laurus nobilis
BIO09 Mean Temperature of Driest 0.68 1.20 0.05 L. and Myrtus communis L., amongst the others, are very rare if not
Quarter completely absent (Bacchetta et al., 2007; Farris et al 2007a), whereas
BIO10 Mean Temperature of Warmest 0.87 1.53 0.06
species more tolerant to continentality like Quercus gr. pubescens are
Quarter
BIO11 Mean Temperature of Coldest 0.96 0.41 0.05 relatively abundant even at lower elevation (Bacchetta et al., 2004b).
Quarter The observed differences related to the precipitation of driest/
BIO12 Annual Precipitation 0.60 137.00 0.20 warmest periods also influence the delimitation between Mediterranean
BIO13 Precipitation of Wettest Month 0.61 19.57 0.19 and Temperate macro-bioclimates (Rivas-Martínez et al., 2011). The
BIO14 Precipitation of Driest Month 0.75 2.41 0.29
definition of this ecological boundary can be particularly important in a
BIO15 Precipitation Seasonality 0.42 4.90 0.10
(Coefficient of Variation) Mediterranean island where zones with a Temperate bioclimate are
BIO16 Precipitation of Wettest Quarter 0.56 64.76 0.22 crucial for the conservation of small, isolated populations of plant spe­
BIO17 Precipitation of Driest Quarter 0.81 10.09 0.27 cies of boreal-temperate origin, often living at their rear edge and
BIO18 Precipitation of Warmest 0.55 21.75 0.54
therefore with important conservation concerns such as Daphne laureola,
Quarter
BIO19 Precipitation of Coldest Quarter 0.54 52.59 0.24
Isopyrum thalictroides, Lotus alpinus and Sanicula europaea (Farris et al.,
2018; Rosati et al., 2020) but also characterized by a high evolutionary
potential (Hampe and Petit, 2005). In the same way, those Temperate
that the discrepancy in precipitation-related variables between local and areas in a Mediterranean context host non-sclerophyllous plant com­
WorldClim datasets could determine a lack of robustness of the species munities like woods with Quercus gr. pubescens, Ostrya carpinifolia, Taxus
distribution models leading, for example, to artifacts in the projections baccata and Ilex aquifolium, as shrubs with Sorbus torminalis, Malus

6
E. Bazzato et al. Ecological Modelling xxx (xxxx) xxx

Fig. 5. Spatial distribution of the differences between all pairs of the new high spatial resolution dataset and WorldClim 2 temperature-related bioclimatic variables
(BIO01-BIO07 and BIO10-BIO11). Red colours indicate overestimated areas by WorldClim 2, blue colours the underestimated ones.

pumila, Pyrus communis and Juniperus nana (Bacchetta et al., 2009; Mediterranean territory. According to our results, in Sardinia the most
Farris et al., 2012) and perennial pasturelands with Anthoxanthum consistent indices regard temperature, with the Annual Mean Temper­
odoratum and Cynosurus cristatus (Farris et al., 2013), identified as ature being the most reliable one. Yet, the spatial distribution of the
habitats of European concern. variables highlights that mountain areas are difficult to model; in fact,
Treating the bioclimatic indices individually helps us to understand even the annual mean temperature shows some variations.
which are more reliable, indicating the critical issues to be faced when On the contrary, WorldClim 2 does not seem to be reliable on the
one is forced to use global datasets such as WorldClim 2 in a precipitation indices, influencing the combined indices too: amongst the

7
E. Bazzato et al. Ecological Modelling xxx (xxxx) xxx

Fig. 6. Spatial distribution of the differences between all pairs of the new high spatial resolution dataset and WorldClim 2 rainfall-related bioclimatic variables
(BIO12-BIO17). Red colours indicate overestimated areas by WorldClim 2, blue colours the underestimated ones.

least performing variables we can identify Minimum Temperature of composition and richness in Europe is shaped and strongly influenced by
Coldest Month (BIO06) and the Precipitation of Warmest Quarter both historical and environmental conditions, in particular climate
(BIO18). (Svenning and Skov, 2005): high levels of divergence have been high­
Given the high discrepancy identified in several sectors of our study lighted, particularly on islands, which have been attributed to the
area, we recommend being cautious in making ecological assessments combined effects of climatic changes, current ecological conditions, and
based on bioclimatic variables derived from global data with coarse anthropogenic factors, that have originated a long history of population
spatial resolutions. The high degree of variability of the new high- isolation (González-Martínez et al., 2010).
resolution bioclimatic variables of the island underlined the need to These new data will support a new generation of research studies in a
use fine spatial resolution data to capture the ecological response in broad array of ecological applications at a much finer scale than previ­
physiographically complex landscapes (Hijmans et al., 2005). ously possible. This sharpening of analysis is particularly urgent in those
areas considered as climate-change hotspots (Giorgi, 2006), like the
6. Conclusions Mediterranean basin (Giorgi and Lionello, 2008): in southern European
mountains boreo-temperate species are suspected to undergo a serious
In this paper we present and make available the first high spatial decline in future decades, as a consequence of the climatic change
resolution dataset for the second largest island in the Mediterranean (Erschbamer et al., 2009; Normand et al., 2007; Stanisci et al., 2005).
(Sardinia, Italy), including the 19 bioclimatic variables proposed in Coarse-scale data is certainly useful for studying patterns on a global
WorldClim and widely used for ecological studies (e.g. Iannella et al., scale, but to model in order to obtain reliable results for planning con­
2019; Sýkora et al., 2017). servation actions and biodiversity management, we need data with good
Increasing the availability of high spatial resolution data to improve spatial resolution (Sandoval et al., 2020), showing the variability of our
ecological understanding of variation at finer scales is extremely territories.
important, especially in the Mediterranean regions where past
geographical and climatic changes and current environmental hetero­ Author contributions
geneities have determined high levels of biodiversity and biogeo­
graphical complexity (Medail, 2017; Thompson, 2020). Tree species EB designed the study, developed the new high-resolution

8
E. Bazzato et al. Ecological Modelling xxx (xxxx) xxx

References

Bacchetta, G., liriti, G., Mossa, L., Pontecorvo, C., Serra, G., 2004a. A phytosociological
study of Ostrya carpinifolia Scop. woods in Sardinia (Italy). Fitosociologia 41 (1).
Bacchetta, G., Biondi, E., Filigheddu, R., Farris, E., Mossa, L., 2004b. A phytosociological
study of the deciduous oak woods of Sardinia (ltaly). Fitosociologia 41 (1), 53–65.
Farris, E., Secchi, Z., Filigheddu, R., 2007a. Phytosociological study of the shrub and pre-
forest communities of the effusive substrata of NW Sardinia. Fitosociologia 44 (2),
55–81.
Bacchetta, G., Farris, E., Fenu, G., Filigheddu, R., Mattana, E., Mulè, P., 2007. Contributo
alla conoscenza dei boschi a Laurus nobilis L. della Sardegna, habitat prioritario ai
sensi della Direttiva 92/43/CEE. Fitosociologia 44 (2), 239–244 suppl. 1.
Bacchetta, G., Bagella, S., Biondi, E., Farris, E., Filigheddu, R., Mossa, L., 2009.
Vegetazione forestale e serie di vegetazione della Sardegna (con rappresentazione
cartografica alla scala 1:350.000). Fitosociologia 46, 3–82.
Barredo, J.I., Caudullo, G., Dosio, A., 2016. Mediterranean habitat loss under future
climate conditions: assessing impacts on the Natura 2000 protected area network.
Appl. Geogr. 75, 83–92. https://doi.org/10.1016/j.apgeog.2016.08.003.
Bedia, J., Herrera, S., Gutiérrez, J.M., 2013. Dangers of using global bioclimatic datasets
for ecological niche modeling. Limitations for future climate projections. Glob.
Planet Change 107, 1–12. https://doi.org/10.1016/j.gloplacha.2013.04.005.
Bellard, C., Bertelsmeier, C., Leadley, P., Thuiller, W., Courchamp, F., 2012. Impacts of
climate change on the future of biodiversity. Ecol. Lett. 15, 365–377. https://doi.
org/10.1111/j.1461-0248.2011.01736.x.
Biondi, E., Filigheddu, R., Farris, E., 2001. Il paesaggio vegetale della Nurra.
Fitosociologia 38 (2, Suppl. 2), 3–105.
Booth, T.H., Nix, H.A., Busby, J.R., Hutchinson, M.F., 2014. bioclim: the first species
distribution modelling package, its early applications and relevance to most current
MaxEnt studies. Div. Distrib. 20, 1–9. https://doi.org/10.1111/ddi.12144.
Booth, T.H., Nix, H.A., Hutchinson, M.F., Busby, J.R., 1987. Grid matching: a new
method for homoclime analysis. Agric. For. Meteorol. 39, 241–255. https://doi.
org/10.1016/0168-1923(87)90041-4.
Vega, C., G., Pertierra, L.R., Olalla-Tárraga, M.Á., 2017. MERRAclim, a high-resolution
global dataset of remotely sensed bioclimatic variables for ecological modelling. Sci.
Data 4, 170078. https://doi.org/10.1038/sdata.2017.78.
Canu, S., Rosati, L., Fiori, M., Motroni, A., Filigheddu, R., Farris, E., 2015. Bioclimate
map of Sardinia (Italy). J. Maps 11, 711–718. https://doi.org/10.1080
/17445647.2014.988187.
Carmignani, L., Oggiano, G., Funedda, A., Conti, P., Pasci, S., 2016. The geological map
of Sardinia (Italy) at 1:250,000 scale. J. Maps 12, 826–835. https://doi.org/10.1
080/17445647.2015.1084544.
Casazza, G., Giordani, P., Benesperi, R., Foggi, B., Viciani, D., Filigheddu, R., Farris, E.,
Bagella, S., Pisanu, S., Mariotti, M.G., 2014. Climate change hastens the urgency of
Fig. 7. Spatial distribution of the differences between all pairs of the new high conservation for range-restricted plant species in the central-northern Mediterranean
spatial resolution dataset and WorldClim 2 bioclimatic variables related to both region. Biol. Conserv. 179, 129–138. https://doi.org/10.1016/j.biocon.2014.09.0
temperature and precipitation (BIO08-BIO09 and BIO18-BIO19). Red colours 15.
indicate overestimated areas by WorldClim 2, blue colours the under­ Chust, G., Pretus, J.Ll., Ducrot, D., Ventura, D., 2004. Scale dependency of insect
assemblages in response to landscape pattern. Landscape Ecol. 19, 41–57.
estimated ones. https://doi.org/10.1023/B:LAND.0000018368.99833.f2.
Conrad, O., Bechtel, B., Bock, M., Dietrich, H., Fischer, E., Gerlitz, L., Wehberg, J.,
Wichmann, V., Böhner, J., 2015. System for Automated Geoscientific Analyses
biologically meaningful variables, and drafted a first version of the
(SAGA) v. 2.1.4. Geosci. Model Dev. 8, 1991–2007. https://doi.org/10.5194/gmd-8-
manuscript. MM, LR, MF, EF and SC designed the study and helped draft 1991-2015.
the manuscript. All authors revised the manuscript and approved the Di Febbraro, M., D’Amen, M., Raia, P., De Rosa, D., Loy, A., Guisan, A., 2018. Using
final version of the manuscript for publication. macroecological constraints on spatial biodiversity predictions under climate
change: the modelling method matters. Ecol. Modell. 390, 79–87. https://doi.org/10
.1016/j.ecolmodel.2018.10.023.
CRediT authorship contribution statement Dobrowski, S.Z., 2011. A climatic basis for microrefugia: the influence of terrain on
climate. Glob. Chang. Biol. 17, 1022–1035. https://doi.org/10.1111/j.1365-2486.
2010.02263.x.
Erika Bazzato: Conceptualization, Writing - original draft, Formal Elith, J., Leathwick, J.R., 2009. Species distribution models: ecological explanation and
analysis. Leonardo Rosati: Methodology, Validation, Writing - review prediction across space and time. Annu. Rev. Ecol. Evol. Syst. 40, 677–697. https://
& editing. Simona Canu: Methodology, Validation, Writing - review & doi.org/10.1146/annurev.ecolsys.110308.120159.
Erschbamer, B., Kiebacher, T., Mallaun, M., Unterluggauer, P., 2009. Short-term signals
editing. Michele Fiori: Methodology, Validation, Writing - review & of climate change along an altitudinal gradient in the South Alps. Plant Ecol. 202,
editing. Emmanuele Farris: Methodology, Validation, Writing - review 79–89. https://doi.org/10.1007/s11258-008-9556-1.
& editing. Michela Marignani: Methodology, Validation, Writing - re­ Farris, E., Carta, M., Circosta, S., Falchi, S., Papuga, G., Lange, P.de, 2018. The
indigenous vascular flora of the forest domain of Anela (Sardinia, Italy). PhytoKeys
view & editing.
113, 97–143. https://doi.org/10.3897/phytokeys.113.28681.
Farris, E., Fenu, G., Bacchetta, G., 2012. Mediterranean Taxus baccata woodlands in
Declaration of Competing Interest Sardinia: a characterization of the EU priority habitat 9580. Phytocoenologia
231–246. https://doi.org/10.1127/0340-269X/2011/0041-0501.
Farris, E., Secchi, Z., Rosati, L., Filigheddu, R., 2013. Are all pastures eligible for
The authors declare that they have no conflict of interest. conservation? A phytosociological survey of the Sardinian–Corsican Province as a
basic tool for the Habitats Directive. Plant Biosyst. - An Int. J. Deal. Asp. Plant Biol.
Supplementary materials 147, 931–946. https://doi.org/10.1080/11263504.2013.778911.
Fick, S.E., Hijmans, R.J., 2017. WorldClim 2: new 1-km spatial resolution climate
surfaces for global land areas. Int. J. Climatol. 37, 4302–4315. https://doi.
Supplementary material associated with this article can be found, in org/10.1002/joc.5086.
the online version, at doi:10.1016/j.ecolmodel.2020.109354. Fois, M., Bacchetta, G., Cogoni, D., Fenu, G., 2018a. Current and future effectiveness of
the Natura 2000 network for protecting plant species in Sardinia: a nice and complex
strategy in its raw state? J. Environ. Plan. Manage. 61, 332–347. https://doi.org/1
0.1080/09640568.2017.1306496.
Fois, M., Cuena-Lombraña, A., Fenu, G., Bacchetta, G., 2018b. Using species distribution
models at local scale to guide the search of poorly known species: review,

9
E. Bazzato et al. Ecological Modelling xxx (xxxx) xxx

methodological issues and future directions. Ecol. Modell. 385, 124–132. https:// Médail, F., 2017. The specific vulnerability of plant biodiversity and vegetation on
doi.org/10.1016/j.ecolmodel.2018.07.018. Mediterranean islands in the face of global change. Reg. Environ. Change 17,
Fois, M., Fenu, G., Cañadas, E.M., Bacchetta, G., 2017. Disentangling the influence of 1775–1790. https://doi.org/10.1007/s10113-017-1123-7.
environmental and anthropogenic factors on the distribution of endemic vascular Meineri, E., Hylander, K., 2017. Fine-grain, large-domain climate models based on
plants in Sardinia. PLoS ONE 12, e0182539. https://doi.org/10.1371/journal.pon climate station and comprehensive topographic information improve microrefugia
e.0182539. detection. Ecography 40, 1003–1013. https://doi.org/10.1111/ecog.02494.
Franklin, J., Davis, F.W., Ikegami, M., Syphard, A.D., Flint, L.E., Flint, A.L., Hannah, L., Nezer, O., Bar-David, S., Gueta, T., Carmel, Y., 2017. High-resolution species-distribution
2013. Modeling plant species distributions under future climates: how fine scale do model based on systematic sampling and indirect observations. Biodivers Conserv
climate projections need to be? Glob. Chang. Biol. 19, 473–483. https://doi.org/ 26, 421–437. https://doi.org/10.1007/s10531-016-1251-2.
10.1111/gcb.12051. Nix, H., 1986. A biogeographic analysis of Australian elapid snakes. Atlas Elapid Snakes
Fréjaville, T., Benito Garzón, M., 2018. The EuMedClim database: yearly climate data Aust. 7, 4–15.
(1901–2014) of 1 km resolution grids for europe and the mediterranean basin. Front. Normand, S., Svenning, J.-C., Skov, F., 2007. National and European perspectives on
Ecol. Evol. 6. https://doi.org/10.3389/fevo.2018.00031. climate change sensitivity of the habitats directive characteristic plant species.
Giorgi, F., 2006. Climate change hot-spots. Geophys. Res. Lett. 33. https://doi.org/10.10 J. Nat. Conserv. 15, 41–53. https://doi.org/10.1016/j.jnc.2006.09.001.
29/2006GL025734. Ongaro, S., Martellos, S., Bacaro, G., Agostini, A.D., Cogoni, A., Cortis, P., 2018.
Giorgi, F., Lionello, P., 2008. Climate change projections for the Mediterranean region. Distributional pattern of Sardinian orchids under a climate change scenario.
Global and Planetary Change. Mediterr. Clim. 63, 90–104. https://doi.org/10.1016/ Commun. Ecol. 19, 223–232. https://doi.org/10.1556/168.2018.19.3.3.
j.gloplacha.2007.09.005. Papuga, G., Gauthier, P., Pons, V., Farris, E., Thompson, J.D., 2018. Ecological niche
González-Martínez, S.C., Dubreuil, M., Riba, M., Vendramin, G.G., Sebastiani, F., differentiation in peripheral populations: a comparative analysis of eleven
Mayol, M., 2010. Spatial genetic structure of Taxus baccata L. in the western Mediterranean plant species. Ecography 41, 1650–1664. https://doi.org/10.1111/e
Mediterranean Basin: past and present limits to gene movement over a broad cog.03331.
geographic scale. Mol. Phylogenet. Evol. 55, 805–815. https://doi.org/10.1016/j. Parmesan, C., 2006. Ecological and evolutionary responses to recent climate change.
ympev.2010.03.001. Annu. Rev. Ecol. Evol. Syst. 37, 637–669. https://doi.org/10.1146/annurev.ecolsys
Greenberg, J.A., Mattiuzzi, M., 2020. gdalUtils: wrappers for the geospatial data .37.091305.110100.
abstraction library (GDAL) utilities. R package version 2.0.3.2. https://CRAN. Pecchi, M., Marchi, M., Burton, V., Giannetti, F., Moriondo, M., Bernetti, I., Bindi, M.,
R-project.org/package=gdalUtils. Chirici, G., 2019. Species distribution modelling to support forest management. A
Guisan, A., Graham, C.H., Elith, J., Huettmann, F., 2007. Sensitivity of predictive species literature review. Ecol. Modell. 411, 108817. https://doi.org/10.1016/j.ec
distribution models to change in grain size. Divers. Distrib. 13, 332–340. https://doi. olmodel.2019.108817.
org/10.1111/j.1472-4642.2007.00342.x. Pesaresi, S., Biondi, E., Casavecchia, S., 2017. Bioclimates of Italy. J. Maps 13, 955–960.
Guisan, A., Thuiller, W., 2005. Predicting species distribution: offering more than simple https://doi.org/10.1080/17445647.2017.1413017.
habitat models. Ecol. Lett. 8, 993–1009. https://doi.org/10.1111/j.1461-0248.200 Pesaresi, S., Galdenzi, D., Biondi, E., Casavecchia, S., 2014. Bioclimate of Italy:
5.00792.x. application of the worldwide bioclimatic classification system. J. Maps 10, 538–553.
Hampe, A., Petit, R.J., 2005. Conserving biodiversity under climate change: the rear edge https://doi.org/10.1080/17445647.2014.891472.
matters. Ecol. Lett. 8, 461–467. https://doi.org/10.1111/j.1461-0248.2005.00739. Pironon, S., Papuga, G., Villellas, J., Angert, A.L., García, M.B., Thompson, J.D., 2017.
x. Geographic variation in genetic and demographic performance: new insights from an
Hannah, L., Flint, L., Syphard, A.D., Moritz, M.A., Buckley, L.B., McCullough, I.M., 2014. old biogeographical paradigm. Biol. Rev. 92, 1877–1909. https://doi.org/10.1111/
Fine-grain modeling of species’ response to climate change: holdouts, stepping- brv.12313.
stones, and microrefugia. Trends Ecol. Evol. (Amst.) 29, 390–397. https://doi. R core Team, 2020. A language and environment for statistical computing. R Found. Stat.
org/10.1016/j.tree.2014.04.006. Comput. https://www.R-project.org/.
Hess, G.R., Bartel, R.A., Leidner, A.K., Rosenfeld, K.M., Rubino, M.J., Snider, S.B., Rivas-Martínez, S., Rivas-Sáenz, S., Penas, A., 2011. Worldwide Bioclimatic classification
Ricketts, T.H., 2006. Effectiveness of biodiversity indicators varies with extent, system. Glob. Geobot. 1, 1–634.
grain, and region. Biol. Conserv. 132, 448–457. https://doi.org/10.1016/j. Rosati, L., Fascetti, S., Romano, V.A., Potenza, G., Lapenna, M.R., Capano, A.,
biocon.2006.04.037. Nicoletti, P., Farris, E., de Lange, P.J., Del Vico, E., Facioni, L., Fanfarillo, E.,
Hijmans, R.J., 2020. raster: geographic Data Analysis and Modeling. R package version 3, Lattanzi, E., Cano-Ortiz, A., Marignani, M., Fogu, M.C., Bazzato, E., Lallai, E.,
1–5. https://CRAN.R-project.org/package=raster. Laface, V.L.A., Musarella, C.M., Spampinato, G., Mei, G., Misano, G., Salerno, G.,
Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G., Jarvis, A., 2005. Very high Esposito, A., Stinca, A., 2020. New chorological data for the Italian vascular flora.
resolution interpolated climate surfaces for global land areas. Int. J. Climatol. 25, Diversity (Basel) 12, 22. https://doi.org/10.3390/d12010022.
1965–1978. https://doi.org/10.1002/joc.1276. Russo, D., Febbraro, M.D., Rebelo, H., Mucedda, M., Cistrone, L., Agnelli, P., Pasquale, P.
Hutchinson, M.F., 1995. Interpolating mean rainfall using thin plate smoothing splines. P.D., Martinoli, A., Scaravelli, D., Spilinga, C., Bosso, L., 2014. What story does
Int. J. Geogr. Inf. Syst. 9, 385–403. https://doi.org/10.1080/02693799508902045. geographic separation of insular bats tell? A case study on Sardinian Rhinolophids.
Hutchinson, M.F., Gessler, P.E., 1994. Splines — More than just a smooth interpolator. PLoS ONE 9, e110894. https://doi.org/10.1371/journal.pone.0110894.
Geoderma 62, 45–67. https://doi.org/10.1016/0016-7061(94)90027-2. Sandoval, S., López-González, C., Escobar-Flores, J.G., Martínez-Rincón, R.O., 2020.
Iannella, M., D’Alessandro, P., Longo, S., Biondi, M., 2019. New records and potential Effect of spatial resolution, algorithm and variable set on the estimated distribution
distribution by ecological niche modeling of Monoxia obesula in the Mediterranean of a mammal of concern: the squirrel Sciurus aberti. Écoscience 27 (3), 195–207.
area. Bull. Insectol. 72, 135–142. https://doi.org/10.1080/11956860.2020.1772609.
Jennings, M.D., Harris, G.M., 2017. Climate change and ecosystem composition across Sutherst, R.W., Maywald, G.F., 1985. A computerised system for matching climates in
large landscapes. Landscape Ecol. 32, 195–207. https://doi.org/10.1007/s10980 ecology. Agric. Ecosyst. Environ. 13, 281–299. https://doi.org/10.1016/0167-8809
-016-0435-1. (85)90016-7.
Karger, D.N., Conrad, O., Böhner, J., Kawohl, T., Kreft, H., Soria-Auza, R.W., Svenning, J.-C., Skov, F., 2005. The relative roles of environment and history as controls
Zimmermann, N.E., Linder, H.P., Kessler, M., 2017. Climatologies at high resolution of tree species composition and richness in Europe. J. Biogeogr. 32, 1019–1033.
for the earth’s land surface areas. Sci. Data 4, 170122. https://doi.org/10.1038/sda https://doi.org/10.1111/j.1365-2699.2005.01219.x.
ta.2017.122. Sýkora, V., García-Vázquez, D., Sánchez-Fernández, D., Ribera, I., 2017. Range
Kriticos, D.J., Webber, B.L., Leriche, A., Ota, N., Macadam, I., Bathols, J., Scott, J.K., expansion and ancestral niche reconstruction in the Mediterranean diving beetle
2012. CliMond: global high-resolution historical and future scenario climate surfaces genus Meladema (Coleoptera, Dytiscidae). Zool. Scr. 46, 445–458. https://doi.org/
for bioclimatic modelling. Methods Ecol. Evol. 3, 53–64. https://doi.org/10.1111/j. 10.1111/zsc.12229.
2041-210X.2011.00134.x. Title, P.O., Bemmels, J.B., 2018. ENVIREM: an expanded set of bioclimatic and
Lenoir, J., Svenning, J.-.C., 2015. Climate-related range shifts – a global topographic variables increases flexibility and improves performance of ecological
multidimensional synthesis and new research directions. Ecography 38, 15–28. https niche modeling. Ecography 41, 291–307. https://doi.org/10.1111/ecog.02880.
://doi.org/10.1111/ecog.00967. Thompson John, D., 2020. Plant Evolution in the Mediterranean, Insights for
Marchi, M., Sinjur, I., Bozzano, M., Westergren, M., 2019. Evaluating WorldClim version Conservation, 2nd Edition. Oxford University Press, p. 352. ISBN: 9780198835158.
1 (1961–1990) as the baseline for sustainable use of forest and environmental Xu, T., Hutchinson, M.F., 2013. New developments and applications in the ANUCLIM
resources in a changing climate. Sustainability 11, 3043. https://doi.org/10.3390/s spatial climatic and bioclimatic modelling package. Environ. Modell. Softw. 40,
u11113043. 267–279. https://doi.org/10.1016/j.envsoft.2012.10.003.

10