Inferring probable distributional gaps and climate change impacts on the medically important viper Echis leucogaster in the western Sahara- Sahel: An ecological niche modeling approach

Abstract

Knowledge of biodiversity distribution and how climate change may affect species across the Sahara-Sahel is scarce despite it harboring both high biodiversity and a high rate of endemism. As ectotherms, snakes are particularly vulnerable to climate change and susceptible to range shifts and demographic changes driven by climate change. Ecological niche models are a common method for predicting the probability of the occurrence of species and future range shifts induced by climate change. This study examines the probable gaps in the distribution of the white-bellied saw-scaled viper, Echis leucogaster, and the potential influence of climate change on its future geographic range in the western Sahara-Sahel. The currently predicted environmentally suitable areas fitted well with the known geographical range of the species showed relative congruence with the Sahara-Sahel ecoregion delineations and identified areas without known occurrences. In the future, the environmental conditions for the occurrence of E. leucogaster are predicted to increase, as the environmentally suitable areas will potentially experience an increase in their proportion. Future projections also showed that the potentially suitable areas might undergo moderate southward shifts during the late twenty-first century. The results of the present study significantly expand our knowledge on the potential distribution of E. leucogaster and provide valuable insights to guide future sampling efforts and conservation planning for the species.

Full text

B I O D I V E R S I T AS
ISSN: 1412-033X
Volume 23, Number 10, October 2022 E-ISSN: 2085-4722
Pages: 5175-5183 DOI: 10.13057/biodiv/d231025
Inferring probable distributional gaps and climate change impacts on
the medically important viper Echis leucogaster in the western Sahara-
Sahel: An ecological niche modeling approach
IDRISS BOUAM1,, FAROUK KHELFAOUI2, MESSAOUD SAOUDI1
1Department of Ecology and Environment, Faculty of Natural and Life Sciences, University of Batna 2. 05078 Fesdis, Batna, Algeria.
Tel./fax.: +213-670-262329, email: idriss.bouam@univ-batna2.dz
2Department of Biology, Faculty of Sciences, Badji Mokhtar University. 23000 Annaba, Algeria
Manuscript received: 30 August 2022. Revision accepted: 6 October 2022.
Abstract. Bouam I, Khelfaoui F, Saoudi M. 2022. Inferring probable distributional gaps and climate change impacts on the medically
important viper Echis leucogaster in the western Sahara-Sahel: An ecological niche modeling approach. Biodiversitas 23: 5175-5183.
Knowledge of biodiversity distribution and how climate change may affect species across the Sahara-Sahel is scarce despite it harboring
both high biodiversity and a high rate of endemism. As ectotherms, snakes are particularly vulnerable to climate change and susceptible
to range shifts and demographic changes driven by climate change. Ecological niche models are a common method for predicting the
probability of the occurrence of species and future range shifts induced by climate change. This study examines the probable gaps in the
distribution of the white-bellied saw-scaled viper, Echis leucogaster, and the potential influence of climate change on its future
geographic range in the western Sahara-Sahel. The currently predicted environmentally suitable areas fitted well with the known
geographical range of the species showed relative congruence with the Sahara-Sahel ecoregion delineations and identified areas without
known occurrences. In the future, the environmental conditions for the occurrence of E. leucogaster are predicted to increase, as the
environmentally suitable areas will potentially experience an increase in their proportion. Future projections also showed that the
potentially suitable areas might undergo moderate southward shifts during the late twenty-first century. The results of the present study
significantly expand our knowledge on the potential distribution of E. leucogaster and provide valuable insights to guide future
sampling efforts and conservation planning for the species.
Keywords: Climate warming, Echis leucogaster, ecological niche models, geographic distribution, white-bellied saw-scaled viper
INTRODUCTION
The Sahara Desert and the contiguous Sahel constitute
two major biogeographic regions of the African continent
(Linder et al. 2012), and exhibit several features that
distinguish them from other deserts and arid regions of the
world (Brito et al. 2014). The lack of distributional
information for species (Wallacean shortfall; Hortal et al.
2015) is particularly acute across much of the Sahara-Sahel
and has been a serious concern to biodiversity management
and conservation in this region (Brito et al. 2014). This
knowledge gap is due in large part to the expenditure and
the impracticality of sampling across the entire region,
exacerbated by political turmoil and civil unrest and the
resulting insecurity (Bauer et al. 2017; Brito et al. 2018).
At the same time, the velocity and magnitude of climate
change in deserts are expected to be fast and strong
(Williams 2014), causing increasing awareness for desert
biodiversity (Durant et al. 2012; Vale and Brito 2015), and
calls for accurate forecasting of its effects on biodiversity
to support the development of proactive conservation
strategies (Liz et al. 2022).
As ectotherms, snakes are highly dependent on climatic
conditions and susceptible to demographic changes and
range shifts are driven by climate change (Deutsch et al.
2008; Needleman et al. 2018). This susceptibility is
reinforced by slow life history traits and particular
ecological requirements, such as delayed sexual
maturation, feeding specialization, and low dispersal rate,
especially in viperid snakes (Maritz et al. 2016). While
numerous studies have assessed the potential impacts of
climate change on the distributional ranges of medically
significant snakes (Nori et al. 2014; Zacarias and Loyola
2019), data from the Sahara-Sahel are very scarce (Vale
and Brito 2015; El-Gabbass et al. 2016), despite the region
harboring high biodiversity and high rate of endemism
(Brito et al. 2016).
The genus Echis comprises Old World true vipers,
commonly known as saw-scaled or carpet vipers, and
arguably the most medically significant snakes in the world
(Spawls and Branch 2020). The genus is found throughout
the semi-arid/xeric regions of western Africa, thence
eastwards to southern Asia, and currently consists of 12
recognized species (Uetz et al. 2022). The white-bellied
saw-scaled viper, Echis leucogaster (part of the
taxonomically complex group E. pyramidum; Arnold et al.
2009; Pook et al. 2009), occurs continuously throughout
the western half of the Sahel and further north into the
Sahara Desert, exhibiting in the northern edges several
isolated populations in Algeria, Morocco, Tunisia, and
Western Sahara. However, it is still unclear whether this
disjunct distributional pattern is an artifact of sampling

B I O DI V E R S I TA S
23 (10): 5175-5183, October 2022
5176
effort or isolation (Spawls and Branch 2020). Nevertheless,
recent records of the species from Morocco (Koleska et al.
2018; Kane et al. 2019), which is considered one of the
best-sampled countries in North Africa in terms of
herpetofauna (Martínez del Mármol et al. 2019), suggests that
there are still knowledge gaps about the species distribution.
Ecological niche models (ENMs), often referred to as
species distribution models (SDMs) (Peterson and Soberón
2012), are increasingly used for a wide variety of
applications (Sillero et al. 2021), including the assessment
of potential climate change impacts on species distributions
(Anderson 2013), and guiding efforts to locate new
populations (Fois et al. 2018). Maximum entropy modeling
(Maxent) is widely recognized as the most used method for
modeling species niches and distributions, particularly for
small sample sizes, since it only requires species presence-
only data (Phillips et al. 2017), and is considered one of the
best-performing approaches in terms of predictive ability
(Elith et al. 2006).
In the present study, we used ENMs to (i) identify
environmentally suitable areas for E. leucogaster in the
western half of the Sahara-Sahel and (ii) forecast the
potential impacts of climate change for the mid and late
twenty-first century on the potential geographic distribution
and range shifts of the species. The results of this study are
intended to increase the knowledge about the distribution
of the white-bellied saw-scaled viper and guide future
sampling efforts and conservation planning.
MATERIALS AND METHODS
Study area and occurrence data
The vast majority of the western half of the Sahara-
Sahel (Naia and Brito 2021), as well as the Maghreb sensu
Freitas et al. (2018) were defined as our region of interest
(ROI), ranging from 17.0° to 37.5° latitude and -17.3° to
12.5° longitude, hence encompassing most of the range of
E. leucogaster (Spawls and Branch 2020). Occurrences
south of our ROI were not considered since almost all of
the available records exhibit data quality issues related to
spatial uncertainties (e.g., data with no or imprecise locality
information; GBIF 2022) and taxonomic biases (i.e.
extensive range overlap with three morphologically similar
congeners, namely E. ocellatus, E. jogeri and E. romani;
Trape 2018), which undermine comprehensive and
inclusive modeling (Anderson 2012). Besides, El-Gabbas
and Dormann (2018) showed that including whole
distribution data to inform ENMs for regional predictions
does not necessarily improve model accuracy.
Distributional data of E. leucogaster from the defined
ROI were obtained from the available bibliography, the
Global Biodiversity Information Facility (GBIF 2022), and
museum and institutional collections that have not yet been
digitized in the aforementioned database. Additional
records collected from Algeria between the years 2018-
2020 were added (Supplement). Only data with confident
taxonomic identification, precise locality description and/or
GPS coordinates with at least two decimal places were
included while avoiding the inclusion of records whose
coordinates refer to the centroid of large geographic areas
(e.g., counties). In total, 68 non-duplicate occurrence points
were compiled (Figure 1A).
The occurrence dataset was subjected to spatiotemporal
filtering to reduce autocorrelation resulting from sampling
biases (Kramer-Schadt et al. 2013). First, we removed
records documented before 1970 to ensure congruence with
the temporal range of current climate conditions (see
below). Then, we employed a graduated spatial filtering
method (Brown 2014) via the Python-based GIS toolkit
SDMtoolbox (version 2.5) (Brown et al. 2017), by
rarefying occurrence localities using increasing radii from
zero in increments of 5 km, until an optimal balance is
achieved between sample size and spatial independency of
localities. Average nearest neighbor analysis in ArcGIS
Desktop (version 10.5) (ESRI 2016) was used to assess the
spatial pattern of the species distribution. The final dataset
included 37 records at a minimum distance of 30 km apart,
with a low clustered distribution pattern (z-score = -1.16;
nearest neighbor ratio = 0.89). This sample size is well
above the minimum number of occurrences required to
develop accurate ENMs (van Proosdij et al. 2016).
Following Phillips and Dudík (2008), we sampled the
whole ROI for background contrast by allocating 10,000
pseudo-absences spatially at random. This choice is
furthermore justified by the fact that limiting background
extent should be relevant to the dispersal dynamics of the
species (Merow et al. 2013), which are virtually unknown
for E. leucogaster.
Environmental predictors
Current and future environmental data included the
standard 19 bioclimatic variables and elevation, all
retrieved from WorldClim (version 2.1) (Fick and Hijmans
2017) at a spatial resolution of 30 arc-seconds (~1 km²) to
match that of the occurrence dataset. Bioclimatic variables
are temperature- and precipitation-related data (see
worldclim.org for a detailed description), and both mean,
and variation of these variables has a momentous influence
on the performance of ectotherms and hence affect their
ranges (Clusella-Trullas et al. 2011). Elevation was
included as a predictor since E. leucogaster appears to
prefer relatively low elevations (Geniez 2015). These
variables have proved effective in predicting suitable areas
for many ophidian taxa (Mizsei et al. 2016; Freitas et al.
2018; Bouam et al. 2019), inter alia, arid-adapted viperid
snakes (Brito et al. 2011; Kane et al. 2019).
Current climatic conditions correspond to historical
averages for the period 1970-2000, while future climate
projections represent downscaled and calibrated data from
the Coupled Model Inter-comparison Project Phase 6
(CMIP6; Eyring et al. 2016) for the mid (averages for
2041-2060) and late (averages for 2071-2100) twenty-first
century. Two Shared Socioeconomic Pathways (SSPs)
were considered for each model, namely a moderate
(SSP2) and an extreme (SSP5) scenario. The SSPs are a set
of alternative future scenarios of societal development that
provide concrete assumptions on the future changes of
elements such as greenhouse gas emissions, which serve as
inputs to model global climate change (Riahi et al. 2017).

BOUAM et al. – Current and future distribution of Echis leucogaster
5177
Since topography barely changes within short periods of
time (Merow et al. 2014), we kept elevation consistent for
both the current and future conditions.
In order to account for uncertainties originating from
variation among climate change projections of different
Global Circulation Models (GCMs) (Thuiller et al. 2019),
GCMs that represent a “near ensemble” model adapted to
our ROI for each SSP were selected based on results
obtained via the R-based web application “GCM
compareR” (Fajardo et al. 2020). This left us with three
GCMs, namely BCC.CSM2.MR, MIROC6 and
MIROC.ES2L. All environmental data were cropped to our
ROI using ArcGIS Desktop (version 10.5) (ESRI 2016).
Given that multicollinearity does not affect Maxent
model performance (Feng et al. 2019), and that the fitting
process leverages of the existing collinearity in finding the
best set of parameters (De Marco and Nóbrega 2018), we
refrained from excluding highly correlated predictors. In
addition, in the absence of thorough knowledge about the
ecology of E. leucogaster, the selection of which collinear
predictors to omit is not straightforward and can be a
source of additional error (Dormann et al. 2013).
Nonetheless, since our models are projected through time,
we quantified collinearity shifts to better infer model
uncertainties as recommended by Feng et al. (2019). To do
so, we compared the Pearson correlation coefficient (r)
matrices of both current and future environmental predictor
variables, averaged across all future scenarios and periods,
by computing the Similarity of Matrix Index (SMI) via the
R package “MatrixCorrelation” (Indahl et al. 2018).
Model tuning and evaluation
The default parameter settings of Maxent (i.e., auto-
features) could ultimately produce over-simplistic or -
complex models, particularly when dealing with small
samples (Morales et al. 2017). The recommendation is to
evaluate the best potential combination of regularization
multipliers and feature classes, which affect model
accuracy by determining the complexity and type of
dependencies on the environment Maxent tries to fit
(Merow et al. 2013). The optimal model parameters were
tuned using the R package “ENMeval” (Kass et al. 2021).
We jackknifed each occurrence record (Shcheglovitova and
Anderson 2013) and tested all possible combinations of the
five feature classes (L: linear; Q: quadratic; P: product; T:
threshold; H: hinge) (n: 31) across a range of regularization
multiplier values from 0.5 to 5 in increments of 0.5, which
resulted in 310 candidate models.
Selecting which evaluation metric to adopt in order to
determine the optimal model may be a challenging task
(Radosavljevic and Anderson 2014, Velasco and González-
Salazar 2019). A common metric used is the area under the
receiver-operating characteristic curve (AUC). However, it
can be representative of performance when true absences
exist and is thus considered a poor metric for presence-
background models (Jiménez‐Valverde 2012); we
nevertheless report AUC for comparison purposes. In our
case, we applied the Symmetric Extremal Dependence
Index (SEDI) (Wunderlich et al. 2019) as the main metric
for model selection, calculated using the R package
“maxentools” (Scavetta 2019). This index is analogous to
the widely used True Skill Statistic (TSS) (Allouche et al.
2006) but better behaved in presence-background models
as its error weighting reflects the low confidence in the
pseudo-absence data, particularly in models with a high
number of background points and low prevalence as in this
study. SEDI makes use of confusion matrix components
(i.e., rates of true and false positives and negatives) and
ranges from -1 to 1, where values ≤ 0 suggest a
performance equal or worse than random, and better
predictions are associated to higher SEDI. Moreover, we
inspected the omission rate for testing points using a
threshold set by the 10% training omission rate (OR10) to
determine whether the selected model is also high
performing in terms of overfitting (Radosavljevic and
Anderson 2014). The continuous environmental suitability
score, which ranges from 0 to 1, of the optimal model was
reclassified into four classes, namely unsuitable (<0.25),
low suitability (0.25-0.50), moderate suitability (0.50-
0.75), and high suitability (>0.75). Following Drake and
Richards (2018), we considered environmental suitability
as a proxy for the probability of occurrence of the species.
Predicted changes under climate change
The potential range of changes in environmentally
suitable areas for E. leucogaster under the different SSPs
and periods was quantified using SDMtoolbox (version
2.5) (Brown et al. 2017). We first converted the models for
current and future environmental conditions from a
continuous logistic output to a binary classification using
the maximum sum of sensitivity and specificity threshold
(maxSSS), which is recommended when only presence data
are available (Liu et al. 2016). Then, the predicted areas of
expansion, contraction, stability, and unsuitability were
computed. To further investigate the trends of changes in
suitable areas, we determined the core distributional shifts
of the species by reducing its distribution to a centroid and
creating a vector that depicts the direction and magnitude of
change across all future scenarios and periods (Brown 2014).
RESULTS AND DISCUSSION
Model assessment
The optimal model (Figure 1B) had a regularization
multiplier of 0.5, allowed for linear and product features,
had the highest AUC value (0.882) among all models, and
a satisfactorily high SEDI value of 0.817, which indicates a
very good model performance and a balance of commission
and omission errors. The model also had a low degree of
overfitting, with an OR10 value of 0.081. Based on these
performance estimates, we considered our model to have a
high discrimination capacity in recovering environmentally
suitable areas for E. leucogaster.
The comparison between the Pearson correlation
coefficient (r) matrices of current and future environmental
predictors (Figure 2) yielded an SMI value of 0.989, and
the largest absolute change of (r) was 0.23 for BIO3 and
BIO5, demonstrating the highly similar collinearity
structure between training and projection environmental

B I O DI V E R S I TA S
23 (10): 5175-5183, October 2022
5178
data, and thus minimal shifts in collinearity. Besides,
transferring models led to no significant change in model performance, indicated by convergent values of SEDI and
AUC compared to the current optimal model (Table 1).
Table 1. Summary information for projected models under the two Shared Socioeconomic Pathways (SSPs) for the mid (2041-2060)
and late (2071-2100) twenty-first century
Time period
SSP
Model metrics
Changes in suitable areas (km²)
Centroid changes
SEDI
AUC
Expansion
Contraction
Stable
Unsuitable
Shift direction
Shift distance (km)
2041-2060
SSP2
0.819
0.884
67,130
65,207
50,6555
4,724,522
East
49.55
SSP5
0.803
0.872
280,984
52,225
51,9537
4,510,668
South-south-east
56.92
2071-2100
SSP2
0.797
0.882
497,028
3310
56,8452
4,294,624
South-south-east
105.82
SSP5
0.78
0.883
297,551
20,607
55,1155
4,494,101
South-south-east
79.8
Figure 1. A. Geographic distribution of Echis leucogaster based on known (red dots) and new (blue dots) records; B. Potential
environmentally suitable areas for the species; C-F. Projected future range changes for the species under the two Shared Socioeconomic
Pathways (SSPs) for the mid (2041-2060) and late (2071-2100) twenty-first century. Species image used with permission from Daniel Kane

BOUAM et al. – Current and future distribution of Echis leucogaster
5179
Figure 2. Pearson pairwise correlation matrix between current (lower-left matrix) and future (upper-right matrix) environmental predictors
Figure 3. Centroid change in the potential distribution of Echis leucogaster under the two Shared Socioeconomic Pathways (SSPs) for
the mid (2041-2060) and late (2071-2100) twenty-first century
Environmentally suitable areas
The current environmentally suitable areas for E.
leucogaster were estimated to account for 32.8% of the
global ROI land extent. The low suitability class accounted
for most of the overall suitability areas with 71.9%,
whereas moderate and high suitability areas represented
23.9% and 4.2%, respectively (Figure 1B). The latter was
characterized by a markedly disjunct and patchy
distribution and were mainly found in (i) Mauritania: large
areas restricted to the Sahel region of the country; (ii)
Algeria: continuously along the southern limits of the
Aurès-Nemencha massif extending westwards into the

B I O DI V E R S I TA S
23 (10): 5175-5183, October 2022
5180
plain of Hodna and Bou Saâda region, as well as in the
vicinity of the country’s southern massifs, particularly
Tassili n’Ajjer and Tassili n’Immidir; (iii) Morocco: along
the Anti-Atlas range and north of the Moulouya basin (i.e.,
Taourirt-Guercif region); Tunisia: north of the Dahar-
Matmama plateau and northwest of the low steppes. The
optimal model identified central Algeria as being
unsuitable despite the availability of a newly confirmed
record of E. leucogaster from south of the township of
Hassi Gara, located nearly equidistantly between the
northern and the southern known records from the country
(Figure 1A).
Climate change impacts
The future projections predicted positive differences
between the expansions and contractions of the
environmentally suitable areas for the species under all
SSPs and periods, thus reflecting an overall increase in
suitable areas over time, although the magnitude of the
potential expansion varied depending on the periods and
scenarios (Figure 1C-F; Table 1). The gain in the extent of
suitable areas ranged from moderate (11.67%; SSP2 2041-
2060) to substantial (85.54%; SSP2 2071-2100), while
areas of contraction were of a small scale, particularly in
the late twenty-first century under the SSP2 scenario. Most
of the currently suitable areas (>88%) were predicted to
remain stable under all future scenarios and periods. On the
contrary, the centroid change analysis demonstrated that
both the direction and distance of the core distribution
varied somewhat according to scenarios and periods
(Figure 3). The centroid of the environmentally suitable
areas was predicted to shift south-south-east across all
scenarios and periods, except for the SSP2 of the mid-
twenty-first century, which displayed an eastward shift.
The predicted shift in suitability distance ranged from
49.55 km for the SSP2 during 2041-2060 and 105.82 km
for the SSP2 during 2071-2100 (Table 1).
Discussion
The current environmentally suitable areas predicted by
the optimal model fitted well with the known geographical
range of E. leucogaster (Spawls and Branch 2020), and
their spatial patterns showed relative congruence with the
Sahara-Sahel ecoregion delineations (Naia and Brito 2021).
Suitable areas were mainly located in the Sahelian Acacia
savanna, North Saharan xeric steppe and woodlands,
particularly along their northern edges, and the West
Saharan montane xeric woodlands. By contrast, the Sahara
Desert ecoregion types and the Maghreb were largely
unsuitable, except for the Taourirt-Guercif region in
Morocco. However, the lack of known records from this
area despite intensive surveys (Mediani et al. 2015) may
suggest that factors other than bioclimatic variables and
elevation (e.g., biotic interactions) may limit its occurrence
there. Central Algeria appears as mostly unsuitable, despite
coinciding with a newly reported occurrence of the species.
This is presumably because this area was underrepresented
in the training data, suggesting that the new record either
may represent an isolated population or is part of large and
continuous distribution. Improving sampling efforts across
this region would help to make this distinction.
An important aspect revealed by the optimal model is
the predicted geographical isolation of central Saharan
populations of southern Algeria. Similar relict distribution
patterns within the region were also observed for many
taxa, including the snakes Macroprotodon cucullatus
(Carranza et al. 2004), Hemorrhois algirus (Trape and
Mané 2006) and Malpolon insignitus (Bakhouche et al.
2019). Several lines of evidence indicate that the moist
habitats restricted to the vicinity of the region’s main
massifs likely serve as climate refugia sensu Hampe et al.
(2013), and maintain favorable conditions for many species
compared to the surrounding plains where hyper-aridity
prevails.
The distribution of E. leucogaster in our ROI could be
much less disjunct than previously thought. For instances
and as already suggested by Martínez et al. (2012), the
model predicted a continuous distribution between known
records from north-eastern Algeria and central Tunisia.
Therefore, we emphasize that environmentally suitable
areas without known occurrences, as well as unsuitable
areas with known occurrences (e.g., Central Algeria),
should both be considered priority target areas for future
surveys to empirically test the results of our model through
systematic and habitat-specific surveys. Given that saw-
scaled vipers are primarily nocturnal foragers (Tsairi and
Bouskila 2004), surveys should be conducted during the
warmer months of the year between dusk and dawn when
the species is most likely active and at multiple times and
dates since the detection of infrequently encountered snake
species cannot be achieved within a few visits (Kery 2002).
Sampling efforts should also be directed to physiologically
convenient habitats and those most frequently associated
with the species occurrence in previous studies (Martínez
and Fernandez 2012; Koleska et al. 2018; Kane et al. 2019).
Climate change will potentially expand the
environmentally suitable areas of E. leucogaster under
moderate (SSP2) and extreme (SSP5) scenarios and during
both the mid (2041-2060) and late (2071-2100) twenty-first
century. El-Gabbas et al. (2016) projected similar results
for the congener species E. coloratus but not for E.
pyramidum. These results corroborate a recent review by
Needleman et al. (2018) who pointed out that the response
of venomous snakes to climate change varied between taxa
and among regions. Such variation may, in part, be
explained by the species-specific divergence in biological
traits (e.g., physiology, behavior, ecology, life history)
(Foden et al. 2009), hence the importance of further
investigations into the biological and life history traits of E.
leucogaster, which are hitherto understudied but vital for
the understanding on how the species buffers against
climate change.
All future projections showed that most of the currently
predicted environmentally suitable areas for E. leucogaster
would remain stable under climate change. These stable
areas likely represent environmental refugia that are
particularly vital for species with limited dispersal abilities
and slow life history (Ashcroft 2010), such as viperids
(Maritz et al. 2016), and therefore can potentially promote

BOUAM et al. – Current and future distribution of Echis leucogaster
5181
the long-term persistence and conservation of the species
under climate change (Keppel et al. 2015).
The centroid change analysis revealed that the
environmentally suitable areas for E. leucogaster will
potentially experience moderate shifts under all SSPs and
periods. Our results, particularly those for the late twenty-
first century, are in agreement with the general theory that
species are expected to shift their distributions poleward in
latitude as a response to shifting climate envelopes
(Walther et al. 2002). Similar climate-related range shifts
have been predicted for many venomous snake species in
different parts of the world (Nori et al. 2014; Yañez-Arenas
et al. 2016; Zacarias and Loyola 2019). It should be
emphasized, however, that the predicted range shifts for E.
leucogaster assume that the species instantaneously adapts
its range to any change in the distribution of
environmentally suitable areas (i.e., unlimited dispersal
scenario). Though we do not know how E. leucogaster will
respond to the potential environmental suitability
expansions, we assume that the capability of the species to
track suitable conditions and reach newly suitable areas
should ultimately depend on its dispersal and colonizing
abilities, anthropogenic pressure, and other ecological
aspects (e.g., landscape connectivity, competition, prey
availability, predation).
ACKNOWLEDGEMENTS
The authors are grateful to Carlos José Brito
(Universidade do Porto, Portugal) for his valuable
comments on an earlier version of the manuscript and
Stephen Spawls (Herpetologist, UK) for proofreading and
correcting the English.
REFERENCES
Ashcroft MB. 2010. Identifying refugia from climate change. J Biogeogr
37 (8): 1407-1413. DOI: 10.1111/j.1365-2699.2010.02300.x.
Allouche O, Tsoar A, Kadmon R. 2006. Assessing the accuracy of species
distribution models: Prevalence, kappa and the true skill statistic
(TSS). J Appl Ecol 43 (6): 1223-1232. DOI: 10.1111/j.1365-
2664.2006.01214.x.
Anderson RP. 2012. Harnessing the world’s biodiversity data: Promise
and peril in ecological niche modeling of species distributions. Ann
NY Acad Sci 1260 (1): 66-80. DOI: 10.1111/j.1749-
6632.2011.06440.x.
Anderson RP. 2013. A framework for using niche models to estimate
impacts of climate change on species distributions. Ann NY Acad Sci
1297 (1): 8-28. DOI: 10.1111/nyas.12264.
Arnold N, Robinson M, Carranza S. 2009. A preliminary analysis of
phylogenetic relationships and biogeography of the dangerously
venomous Carpet Vipers, Echis (Squamata, Serpentes, Viperidae)
based on mitochondrial DNA sequences. Amphibia-Reptilia 30 (2):
273-282. DOI: 10.1163/156853809788201090.
Bakhouche B, Escoriza D, Sánchez-Vialas A, Khalil D, Calvo-Revuelta
M. 2019. Malpolon Insignitus (Geoffroy de Saint-Hilaire, 1809) in the
Hoggar mountains (Southeast Algeria). Afr Herp News 71: 29-35.
Bauer AM, DeBoer JC, Taylor DJ. 2017. Atlas of the reptiles of Libya.
Proc Cal Acad Sci 64 (8): 155-318.
Bouam I, Benmokhtar E, Guechi R. 2019. A fortuitous encounter with the
Vulnerable Vipera latastei: A new locality record from Algeria and
distributional range extension. Herpetol Notes 12: 809-812.
Brito JC, Durant SM, Pettorelli N. 2018. Armed conflicts and wildlife
decline: Challenges and recommendations for effective conservation
policy in the Sahara‐Sahel. Conserv Lett 11 (5): e12446. DOI:
10.1111/conl.12446.
Brito JC, Fahd S, Geniez P, Martínez-Freiría F, Pleguezuelos JM, Trape
JF. 2011. Biogeography and conservation of viperids from North-
West Africa: An application of ecological niche-based models and
GIS. J Arid Environ 75 (11): 1029-1037. DOI:
10.1016/j.jaridenv.2011.06.006.
Brito JC, Godinho R, Martínez‐Freiría F. 2014. Unravelling biodiversity,
evolution and threats to conservation in the Sahara‐Sahel. Biol Rev 89
(1): 215-231. DOI: 10.1111/brv.12049.
Brito JC, Tarroso P, Vale CG. 2016. Conservation biogeography of the
Sahara‐Sahel: Additional protected areas are needed to secure unique
biodiversity. Divers Distrib 22 (4): 371-384. DOI: 10.1111/ddi.12416.
Brown JL. 2014. SDMtoolbox: A python-based GIS toolkit for landscape
genetic, biogeographic and species distribution model analyses.
Methods Ecol Evol 5 (7): 694-700. DOI: 10.1111/2041-210X.12200.
Brown JL, Bennett JR, French CM. 2017. SDMtoolbox 2.0: The next
generation Python-based GIS toolkit for landscape genetic,
biogeographic and species distribution model analyses. Peer J 5:
e4095. DOI: 10.7717/peerj.4095.
Carranza S, Arnold EN, Wade E, Fahd S. 2004. Phylogeography of the
false smooth snakes, Macroprotodon (Serpentes, Colubridae):
mitochondrial DNA sequences show European populations arrived
recently from Northwest Africa. Mol Phylogenet Evol 33 (3): 523-
532. DOI: 10.1016/j.ympev.2004.07.009.
Clusella-Trullas S, Blackburn TM, Chown SL. 2011. Climatic predictors
of temperature performance curve parameters in ectotherms imply
complex responses to climate change. Am Nat 177 (6): 738-751. DOI:
10.1086/660021.
De Marco Jr P, Nóbrega CC. 2018. Evaluating collinearity effects on
species distribution models: An approach based on virtual species
simulation. Plos One 13 (9): e0202403. DOI:
10.1371/journal.pone.0202403.
Deutsch CA, Tewksbury JJ, Huey RB, Sheldon KS, Ghalambor CK, Haak
DC, Martin PR. 2008. Impacts of climate warming on terrestrial
ectotherms across latitude. P Natl Acad Sci-Biol 105 (18): 6668-6672.
DOI: 10.1073/pnas.0709472105.
Dormann CF, Elith J, Bacher S. 2013. Collinearity: A review of methods
to deal with it and a simulation study evaluating their performance.
Ecography 36 (1): 27-46. DOI: 10.1111/j.1600-0587.2012.07348.x.
Drake JM, Richards RL. 2018. Estimating environmental suitability.
Ecosphere 9 (9): e02373. DOI: 10.1002/ecs2.2373.
Durant SM, Pettorelli N, Bashir S. 2012. Forgotten biodiversity in desert
ecosystems. Science 336 (6087): 1379-1380. DOI:
10.1126/science.336.6087.1379.
El-Gabbas A, Baha ES, Zalat S, Gilbert F. 2016. Conserving Egypt's
reptiles under climate change. J Arid Environ 127: 211-221. DOI:
10.1016/j.jaridenv.2015.12.007.
El-Gabbas A, Dormann CF. 2018. Wrong, but useful: Regional species
distribution models may not be improved by range-wide data under
biased sampling. Ecol Evol 8 (4): 2196-2206. DOI:
10.1002/ece3.3834.
Elith J, Graham CH, Anderson RP. 2006. Novel methods improve
prediction of species’ distributions from occurrence data. Ecography
29 (2): 129-151. DOI: 10.1111/j.2006.0906-7590.04596.x.
ESRI ESRI 2016. ArcGIS, version 10.5. Environmental Systems Research
Institute, California.
Eyring V, Bony S, Meehl GA, Senior CA, Stevens B, Stouffer RJ, Taylor
KE. 2016. Overview of the Coupled Model Intercomparison Project
Phase 6 (CMIP6) experimental design and organization. Geosci
Model Dev 9 (5): 1937-1958. DOI: 10.5194/gmd-9-1937-2016.
Fajardo J, Corcoran D, Roehrdanz PR, Hannah L, Marquet PA. 2020.
GCM compareR: A web application to assess differences and assist in
the selection of general circulation models for climate change
research. Methods Ecol Evol 11 (5): 656-663. DOI: 10.1111/2041-
210X.13360.
Feng X, Park DS, Liang Y, Pandey R, Papeş M. 2019. Collinearity in
ecological niche modeling: Confusions and challenges. Ecol Evol 9
(18): 10365-10376. DOI: 10.1002/ece3.5555.
Fick SE, Hijmans RJ. 2017. WorldClim 2: New 1-km spatial resolution
climate surfaces for global land areas. Intl J Climatol 37 (12): 4302-
4315. DOI: 10.1002/joc.5086.
Foden W, Mace GM, Vié JC. 2009. Species susceptibility to climate
change impacts. In: Vié JC, Hilton-Taylor C, Stuart SN (eds).
Wildlife in a Changing World. An Analysis of the 2008 IUCN Red
List of Threatened Species. IUCN, Gland.

B I O DI V E R S I TA S
23 (10): 5175-5183, October 2022
5182
Fois M, Cuena-Lombraña A, Fenu G, Bacchetta G. 2018. Using species
distribution models at local scale to guide the search of poorly known
species: Review, methodological issues and future directions. Ecol
Model 385: 124-132. DOI: 10.1016/j.ecolmodel.2018.07.018.
Freitas I, Fahd S, Velo-Antón G, Martínez-Freiría F. 2018. Chasing the
phantom: Biogeography and conservation of Vipera latastei-
monticola in the Maghreb (North Africa). Amphibia-Reptilia 39 (2):
145-161. DOI: 10.1163/15685381-17000197.
GBIF 2022. GBIF Occurrence Download. DOI: 10.15468/dl.zmbusk
Geniez P. 2015. Serpents d’Europe, d’Afrique du Nord et du Moyen-
Orient. Delachaux et Niestlé, Paris. [French]
Hampe A, Rodríguez-Sánchez F, Dobrowski S, Hu FS, Gavin DG. 2013.
Climate refugia: From the Last Glacial Maximum to the twenty-first
century. New Phytol 197 (1): 16-18. DOI: 10.1111/nph.12059.
Hortal J, de Bello F, Diniz-Filho JAF, Lewinsohn TM, Lobo JM, Ladle
RJ. 2015. Seven shortfalls that beset large-scale knowledge of
biodiversity. Ann Rev Ecol Evol S 46: 523-549. DOI:
10.1146/annurev-ecolsys-112414-054400.
Indahl UG, Næs T, Liland KH. 2018. A similarity index for comparing
coupled matrices. J Chemometr 32 (10): e3049. DOI:
10.1002/cem.3049.
Jiménez-Valverde A. 2012. Insights into the area under the receiver
operating characteristic curve (AUC) as a discrimination measure in
species distribution modelling. Glob Ecol Biogeogr 21 (4): 498-507.
DOI: 10.1111/j.1466-8238.2011.00683.x.
Kane D, Goodwin S, Verspui GJ, Tump A, Martínez G. 2019. Reptile
diversity of southern Morocco: Range extensions and the role of the
Djebel Ouarkziz as a biogeographical barrier. Herpetol Notes 12: 787-
793.
Kass JM, Muscarella R, Galante PJ, Bohl CL, Pinilla-Buitrago GE, Boria
RA, Soley-Guardia M, Anderson RP. 2021. ENMeval 2.0:
Redesigned for customizable and reproducible modeling of species’
niches and distributions. Methods Ecol Evol 12 (9): 1602-1608. DOI:
10.1111/2041-210X.13628.
Keppel G, Mokany K, Wardell-Johnson GW, Phillips BL, Welbergen JA,
Reside AE. 2015. The capacity of refugia for conservation planning
under climate change. Front Ecol Environ 13 (2): 106-112. DOI:
10.1890/140055.
Kery M. 2002. Inferring the absence of a species: A case study of snakes.
J Wildl Manag 66 (2): 330-338. DOI: 10.2307/3803165.
Koleska D, Karhánek J, Martinez MMG, Sassoè-Pognetto M. 2018. New
records of Echis leucogaster in Morocco. Herpetol Notes 11: 655-
657.
Kramer-Schadt S, Niedballa J, Pilgrim JD. 2013. The importance of
correcting for sampling bias in MaxEnt species distribution models.
Divers Distrib 19 (11): 1366-1379. DOI: 10.1111/ddi.12096.
Linder HP, de Klerk HM, Born J, Burgess ND, Fjeldså J, Rahbek C. 2012.
The partitioning of Africa: Statistically defined biogeographical
regions in sub‐Saharan Africa. J Biogeogr 39 (7): 1189-1205. DOI:
10.1111/j.1365-2699.2012.02728.x.
Liu C, White M, Newell G. 2016. On the selection of thresholds for
predicting species occurrence with presence-only data. Ecol Evol 6
(1): 337-348. DOI: 10.1002/ece3.1878.
Liz AV, Gonçalves DV, Velo-Antón G, Brito JC, Crochet PA, Rödder D.
2022. Adapt biodiversity targets to climate change. Science 376
(6593): 589-590. DOI: 10.1126/science.abo7381.
Maritz B, Penner J, Martins M. 2016. Identifying global priorities for the
conservation of vipers. Biol Conserv 204: 94-102. DOI:
10.1016/j.biocon.2016.05.004.
Martínez MMG, Fernandez RB. 2012. An important new record of Echis
leucogaster Roman, 1972 from Morocco. Herpetol Notes 5: 229-231.
Martínez MMG, Fernandez RB, Mazuch T. 2012. Reflexions about
potential distribution of Echis leucogaster in North Africa.
http://moroccoherps.com/en/reflexions-about-potential-distribution-
of-echis-leucogaster-in-north-africa/.
Martínez MMG, Harris DJ, Geniez P, de Pous P, Salvi D. 2019.
Amphibians and Reptiles of Morocco. Edition Chimaira, Frankfurt.
Mediani M, Brito JC, Fahd S. 2015. Atlas of the amphibians and reptiles
of northern Morocco: Updated distribution and patterns of habitat
selection. Basic Appl Herpetol 29: 81-107. DOI: 10.11160/bah.14009.
Merow C, Smith MJ, Edwards Jr TC. 2014. What do we gain from
simplicity versus complexity in species distribution models?
Ecography 37 (12): 1267-1281. DOI: 10.1111/ecog.00845.
Merow C, Smith MJ, Silander Jr JA. 2013. A practical guide to MaxEnt
for modeling species' distributions: what it does, and why inputs and
settings matter. Ecography 36 (10): 1058-1069. DOI: 10.1111/j.1600-
0587.2013.07872.x.
Mizsei E, Üveges B, Vági B, Szabolcs M, Lengyel S, Pfliegler WP, Nagy
ZT, Tóth JP. 2016. Species distribution modelling leads to the
discovery of new populations of one of the least known European
snakes, Vipera ursinii graeca, in Albania. Amphibia-Reptilia 37 (1):
55-68. DOI: 10.1163/15685381-00003031.
Morales NS, Fernández IC, Baca-González V. 2017. MaxEnt’s parameter
configuration and s mall samples: are we paying attention to
recommendations? A systematic review. Peer J 5: e3093. DOI:
10.7717/peerj.3093.
Naia M, Brito JC. 2021. Ecoregions of the Sahara-Sahel: characteristics
and Conservation Status. CIBIO Research Center in Biodiversity and
Genetic Resources, Portugal.
Needleman RK, Neylan IP, Erickson T. 2018. Potential environmental and
ecological effects of global climate change on venomous terrestrial
species in the wilderness. Wildl Environ Med 29 (2): 226-238. DOI:
10.1016/j.wem.2017.11.004.
Nori J, Carrasco PA, Leynaud GC. 2014. Venomous snakes and climate
change: ophidism as a dynamic problem. Clim Change 122: 67-80.
DOI: 10.1007/s10584-013-1019-6.
Peterson AT, Soberón J. 2012. Species distribution modeling and
ecological niche modeling: Getting the concepts right. Nat
Conservação 10 (2): 102-107. DOI: 10.4322/natcon.2012.019.
Phillips SJ, Anderson RP, Dudík M, Schapire RE, Blair ME. 2017.
Opening the black box: An open‐source release of Maxent.
Ecography 40 (7): 887-893. DOI: 10.1111/ecog.03049.
Phillips SJ, Dudík M. 2008. Modeling of species distributions with
Maxent: New extensions and a comprehensive evaluation. Ecography
31 (2): 161-175. DOI: 10.1111/j.0906-7590.2008.5203.x.
Pook CE, Joger U, Stümpel N, Wüster W. 2009. When continents collide:
Phylogeny, historical biogeography and systematics of the medically
important viper genus Echis (Squamata: Serpentes: Viperidae). Mol
Phylogenet Evol 53 (3): 792-807. DOI:
10.1016/j.ympev.2009.08.002.
Radosavljevic A, Anderson RP. 2014. Making better Maxent models of
species distributions: Complexity, overfitting and evaluation. J
Biogeogr 41 (4): 629-643. DOI: 10.1111/jbi.12227.
Riahi K, van Vuuren DP, Kriegler E. 2017. The shared socioeconomic
pathways and their energy, land use, and greenhouse gas emissions
implications: An overview. Glob Environ Chang 42: 153-168. DOI:
10.1016/j.gloenvcha.2016.05.009.
Scavetta J. 2019. Maxentools, version 1.1.0. DOI:
10.5281/zenodo.3381023.svg.
Shcheglovitova M, Anderson RP. 2013. Estimating optimal complexity
for ecological niche models: A jackknife approach for species with
small sample sizes. Ecol Model 269: 9-17. DOI:
10.1016/j.ecolmodel.2013.08.011.
Sillero N, Arenas-Castro S, Enriquez‐Urzelai U. 2021. Want to model a
species niche? A step-by-step guideline on correlative ecological
niche modelling. Ecol Model 456: 109671. DOI:
10.1016/j.ecolmodel.2021.109671.
Spawls S, Branch B. 2020. The Dangerous Snakes of Africa. Bloomsbury
Publishing, London.
Thuiller W, Guéguen M, Renaud J, Karger DN, Zimmermann NE. 2019.
Uncertainty in ensembles of global biodiversity scenarios. Nat
Commun 10: 1446. DOI: 10.1038/s41467-019-09519-w.
Trape JF. 2018. Partition d’Echis ocellatus Stemmler, 1970 (Squamata,
Viperidae), avec la description d’une espèce nouvelle. B Soc Herpétol
Fr 167: 13-34. [French]
Trape JF, Mané Y. 2006. Guide des Serpents d'Afrique Occidentale:
Savane et Désert. IRD Éditions, Paris. [French]
Tsairi H, Bouskila A. 2004. Ambush site selection of a desert snake (Echis
coloratus) at an oasis. Herpetologica 60 (1): 13-23. DOI: 10.1655/20-47.
Uetz P, Freed P, Aguilar R, Hošek J. 2022. The Reptile Database.
http://www.reptile-database.org.
Vale CG, Brito JC. 2015. Desert-adapted species are vulnerable to climate
change: Insights from the warmest region on Earth. Glob Ecol
Conserv 4: 369-379. DOI: 10.1016/j.gecco.2015.07.012.
van Proosdij ASJ, Sosef MSM, Wieringa JJ, Raes N. 2016. Minimum
required number of specimen records to develop accurate species
distribution models. Ecography 39 (6): 542-552. DOI:
10.1111/ecog.01509.
Velasco JA, González-Salazar C. 2019. Akaike information criterion
should not be a “test” of geographical prediction accuracy in

BOUAM et al. – Current and future distribution of Echis leucogaster
5183
ecological niche modelling. Ecol Inform 51: 25-32. DOI:
10.1016/j.ecoinf.2019.02.005.
Walther GR, Post E, Convey P. 2002. Ecological responses to recent
climate change. Nature 416: 389-395. DOI: 10.1038/416389a.
Williams M. 2014. Climate Change in Deserts. Past, Present and Future.
Cambridge University Press, New York.
Wunderlich RF, Lin YP, Anthony J, Petway JR. 2019. Two alternative
evaluation metrics to replace the true skill statistic in the assessment
of species distribution models. Nat Conserv 35: 97-116. DOI:
10.3897/natureconservation.35.33918.
Yañez-Arenas C, Townsend PA, Rodríguez-Medina K, Barve N. 2016.
Mapping current and future potential snakebite risk in the new world.
Clim Change 134: 697-711. DOI: 10.1007/s10584-015-1544-6.
Zacarias D, Loyola R. 2019. Climate change impacts on the distribution of
venomous snakes and snakebite risk in Mozambique. Clim Change
152: 195-207. DOI: 10.1007/s10584-018-2338-4.