Journal of Arid Environments - elgabbas.github.io · e Faculty of Sciences & Arts, Taibah...

11
Conserving Egypt's reptiles under climate change Ahmed El-Gabbas a, b, c, * , Sherif Baha El Din d , Samy Zalat e, f , Francis Gilbert b a Gebel Elba National Park, Nature Conservation Sector, Egypt b School of Biology, University of Nottingham, Nottingham NG7 2RD, UK c Department of Biometry and Environmental System Analysis, University of Freiburg, D-79106 Freiburg, Germany d Nature Conservation Egypt, 3 Abdala El Katib St, Dokki, Cairo, Egypt e Faculty of Sciences & Arts, Taibah University, Al-Ula Campus, Saudi Arabia f Department of Zoology, Suez Canal University, Ismailia, Egypt article info Article history: Received 5 May 2015 Received in revised form 4 December 2015 Accepted 27 December 2015 Available online 7 January 2016 Keywords: Climate change Conservation prioritization Egyptian herpetofauna Egyptian reptiles Maxent Species distribution modelling Species extinctions Zonation abstract Climate change has caused range shifts and extinctions of many species in the recent past. In this study, the effects of climate change on Egyptian reptiles were investigated for the rst time using species distribution models. Maxent was used to model the current and future distributions of suitable habitats for 75 terrestrial reptile species from Egypt. The modelled distribution for current suitable conditions for each species was projected into the future at three time slices using two emission scenarios from four global circulation models and under two assumptions of dispersal ability. Climate change is expected to vary in its effects spatially, with some areas characterized by increased species richness while others show declines. Future range changes vary among species and different future projections, from the entire loss to large gains in range. Two species were expected to become extinct in at least one future pro- jection, and eight species were expected to lose >80% of their current distribution. Although Protected Areas have greater conservation value, on average, compared to unprotected areas, they appear inade- quate to conserve Egyptian reptiles under expected climate change. © 2016 Elsevier Ltd. All rights reserved. 1. Introduction Over the last century, relatively rapid changes of the Earth's climate have been recorded with warmer temperatures accompa- nied by altered geographical and seasonal distributions of precip- itation (Araújo and Rahbek, 2006; Thuiller, 2007). There is widespread agreement that climate change will have a large impact on the survival of populations, species and communities (Suarez and Tsutsui, 2004), and that biodiversity is continually being transformed in response to it (Hannah et al., 2005). Over the last 40 years climate change has been implicated as the main cause for distributional shifts and extinctions (Thomas et al., 2004), with a particularly strong impact on butteries, birds and species at high altitude (Hannah, 2011). Although the recorded effects of climate change on biodiversity seem in general to materialize rather slowly, its effects are expected to become increasingly prominent over the next 50 years and beyond (Thuiller, 2007). Some climate change model forecasts suggest that 15e37% of current species will be committed to extinction by 2050 (even without taking into consideration the biological factors of competition and evolu- tionary history) (Thuiller, 2007), making it essential to involve measures for mitigating its potential impacts in future conservation plans. Detailed information on the ecological and geographical distri- butions of species is essential for conservation planning and fore- casting (Elith et al., 2006) especially where species face multiple conservation problems. Species Distribution Models (SDMs) quantify the relationship between species' occurrence and envi- ronmental variables and allow users to extrapolate this relationship to new areas or time periods. SDMs have been widely used to es- timate the potential impacts of climate change on species distri- butions and ecosystems (Franklin, 2009) and estimate potential future extinction risks. Once a model has been calibrated for cur- rent climate conditions, it can be used to estimate potential dis- tributions at different time periods (in the future or the past) by using information on expected climates, or to different study areas in order to assess the potential locations where invasions are more * Corresponding author.Current address: Department of Biometry and Environ- mental System Analysis, University of Freiburg, Tennenbacherstr. 4, D-79106 Frei- burg, Germany. E-mail address: [email protected] (A. El-Gabbas). Contents lists available at ScienceDirect Journal of Arid Environments journal homepage: www.elsevier.com/locate/jaridenv http://dx.doi.org/10.1016/j.jaridenv.2015.12.007 0140-1963/© 2016 Elsevier Ltd. All rights reserved. Journal of Arid Environments 127 (2016) 211e221

Transcript of Journal of Arid Environments - elgabbas.github.io · e Faculty of Sciences & Arts, Taibah...

Page 1: Journal of Arid Environments - elgabbas.github.io · e Faculty of Sciences & Arts, Taibah University, Al-Ula Campus, Saudi Arabia f Department of Zoology, Suez Canal University, Ismailia,

lable at ScienceDirect

Journal of Arid Environments 127 (2016) 211e221

Contents lists avai

Journal of Arid Environments

journal homepage wwwelsevier comlocate jar idenv

Conserving Egypts reptiles under climate change

Ahmed El-Gabbas a b c Sherif Baha El Din d Samy Zalat e f Francis Gilbert b

a Gebel Elba National Park Nature Conservation Sector Egyptb School of Biology University of Nottingham Nottingham NG7 2RD UKc Department of Biometry and Environmental System Analysis University of Freiburg D-79106 Freiburg Germanyd Nature Conservation Egypt 3 Abdala El Katib St Dokki Cairo Egypte Faculty of Sciences amp Arts Taibah University Al-Ula Campus Saudi Arabiaf Department of Zoology Suez Canal University Ismailia Egypt

a r t i c l e i n f o

Article historyReceived 5 May 2015Received in revised form4 December 2015Accepted 27 December 2015Available online 7 January 2016

KeywordsClimate changeConservation prioritizationEgyptian herpetofaunaEgyptian reptilesMaxentSpecies distribution modellingSpecies extinctionsZonation

Corresponding authorCurrent address Departmemental System Analysis University of Freiburg Tennburg Germany

E-mail address elgabbasoutlookcom (A El-Gabb

httpdxdoiorg101016jjaridenv2015120070140-1963copy 2016 Elsevier Ltd All rights reserved

a b s t r a c t

Climate change has caused range shifts and extinctions of many species in the recent past In this studythe effects of climate change on Egyptian reptiles were investigated for the first time using speciesdistribution models Maxent was used to model the current and future distributions of suitable habitatsfor 75 terrestrial reptile species from Egypt The modelled distribution for current suitable conditions foreach species was projected into the future at three time slices using two emission scenarios from fourglobal circulation models and under two assumptions of dispersal ability Climate change is expected tovary in its effects spatially with some areas characterized by increased species richness while othersshow declines Future range changes vary among species and different future projections from the entireloss to large gains in range Two species were expected to become extinct in at least one future pro-jection and eight species were expected to lose gt80 of their current distribution Although ProtectedAreas have greater conservation value on average compared to unprotected areas they appear inade-quate to conserve Egyptian reptiles under expected climate change

copy 2016 Elsevier Ltd All rights reserved

1 Introduction

Over the last century relatively rapid changes of the Earthsclimate have been recorded with warmer temperatures accompa-nied by altered geographical and seasonal distributions of precip-itation (Arauacutejo and Rahbek 2006 Thuiller 2007) There iswidespread agreement that climate changewill have a large impacton the survival of populations species and communities (Suarezand Tsutsui 2004) and that biodiversity is continually beingtransformed in response to it (Hannah et al 2005) Over the last 40years climate change has been implicated as the main cause fordistributional shifts and extinctions (Thomas et al 2004) with aparticularly strong impact on butterflies birds and species at highaltitude (Hannah 2011) Although the recorded effects of climatechange on biodiversity seem in general tomaterialize rather slowlyits effects are expected to become increasingly prominent over the

nt of Biometry and Environ-enbacherstr 4 D-79106 Frei-

as)

next 50 years and beyond (Thuiller 2007) Some climate changemodel forecasts suggest that 15e37 of current species will becommitted to extinction by 2050 (even without taking intoconsideration the biological factors of competition and evolu-tionary history) (Thuiller 2007) making it essential to involvemeasures for mitigating its potential impacts in future conservationplans

Detailed information on the ecological and geographical distri-butions of species is essential for conservation planning and fore-casting (Elith et al 2006) especially where species face multipleconservation problems Species Distribution Models (SDMs)quantify the relationship between species occurrence and envi-ronmental variables and allow users to extrapolate this relationshipto new areas or time periods SDMs have been widely used to es-timate the potential impacts of climate change on species distri-butions and ecosystems (Franklin 2009) and estimate potentialfuture extinction risks Once a model has been calibrated for cur-rent climate conditions it can be used to estimate potential dis-tributions at different time periods (in the future or the past) byusing information on expected climates or to different study areasin order to assess the potential locations where invasions are more

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221212

likely to establish (Franklin 2009) This helps to manage speciesfacing possible future threats by identifying biological corridors fordispersal determining sites for re-introduction and areas requiremore protection measures (Thuiller 2007)

Climate change will potentially affect the biodiversity and spe-cies composition of Egyptian ecosystems (Tolba and Saab 2009)but only a handful of quantitative studies of the local faunaflorahave been conducted using SDMs This may be because the modelsare relatively new and because until very recently reliablecomprehensive biodiversity records for the Egyptian fauna andflora were not available As a developing country Egypt lacks arecording scheme to collate species sightings (either at national oreven at local Protected Area level) Between 2004 and 2008 allavailable biodiversity records were collated by the BioMAP project(see httpwwwbiomapegyptorg) and studies of butterflies andmammals using SDMs became possible (Gilbert and Zalat 2008Basuony et al 2010 Newbold 2009 Newbold et al 2009ab) Afew other SDM studies have been published El Alqamy et al (2010)to estimate the potential distribution of the Nubian Ibex (Capranubiana) in South Sinai Soultan (2011) to test the potential impactof climate change on the distribution of four Egyptian antelopespecies and Leach et al (2013) who assessed the effect of climatechange on the future effectiveness of the Protected Area networkusing data on Egyptian butterflies and mammals The only otherstudy of the effects of climate change in Egypt by Hoyle and James(2005) on the worlds smallest butterfly the Sinai Baton Blue(Pseudophilotes sinaicus) used an occupancy-based PopulationViability Analysis

The global current Protected Areas seem not to overlap wellwith areas of high biodiversity value and are traditionally deter-mined spatially and environmentally under the assumption ofrelatively low changes in species distribution in the future (Arauacutejo2009 Leach et al 2013) As climate change is expected to affect thefuture distributionrange of many species globally (potentiallymoving some species out from Protected Areas Hannah et al2007) it challenges the future effectiveness of current ProtectedAreas Future conservation investments should be effectivelyprioritized due to the limited resources available (Wilson et al2009 especially in the developing countries) and early actionsmay be more effective and less costly than delayed or no actions(Hannah et al 2007) Conservation prioritization can be at taxo-nomic (for the protection of some rare or endangered species) orspatial (conserving a particular habitat type or species rich areaseg potential Protected Areas) scale Spatial conservation prioriti-zation uses spatial quantitative data to identify areas of high con-servation priority (Wilson et al 2009) and some techniques havebeen available recently eg Zonation (Moilanen et al 2012) andMarxan (Ball et al 2009) Some of those techniques can use spatialoutputs from SDMs to prioritize areas for conservation under cur-rent and future climates

To date there are 30 Protected Areas in Egypt covering gt15 ofits total area Their distribution shows good spatial coveragealthough some areas of high diversity importance (especially atthe Nile Valley and Delta) are not yet protected (Newbold et al2009a) They were declared relatively recently (first in 1983)and were determinedmainly based on experts known knowledgeof the country biodiversity (Newbold et al 2009a) The capacity ofcurrent Protected Areas in Egypt to mitigate for potential impactsof climate change on different species groups is not well-investigated yet and a qualitative assessment of their futureeffectiveness is highly required Also spatial prioritization of theEgypts landscape (inside and outside the Protected Areas) isrequired to identify potential locations for future Protected Areasand identify current Protected Areas need more conservationeffort in the future

As a developing country Egypt lacks enough good-quality datato be used directly for spatial conservation prioritization but SDMscan provide valuable estimates for species suitabilities in the spaceIn this study we use data for the Egyptian reptiles for the first timeto run SDMs (as a representation group for the Egyptian fauna)Baha El Din (2001 2006) presented a geographic interpretation ofEgypts herpetofaunal distribution qualitatively identifying prior-ity conservation areas but very little has been published on howthe Egyptian herpetofauna may respond to climate change Weused Maxent to model the distribution of Egyptian reptiles undercurrent climate conditions then models are projected into thefuture to show how collectively and individual species are expectedto respond to future climate change under different assumptionsFor each species future expected range change ( loss or gain ofcurrently suitable habitats) is estimated aiming to shed light onsome species require more strict conservation actions Expectedreptiles species richness (under current and future climates) isestimated to identify areas of current high reptile suitability andareas expected to undergomuch changes in suitability in the futureModel predictions were used for prioritizing the Egyptian land-scape under current and future climates We used Zonation soft-ware (Moilanen et al 2012) to assess the likely effectiveness ofEgypts Protected Area network under current and future climateOutputs from Zonation represent hierarchic ranking of the land-scape for conservation and can be easily visualized as maps Wehope the results of the current study to be useful for biodiversityconservation in Egypt and (along with the results of relativestudies) to be taken into account by the decision makers in futurenational conservation plans

2 Methods

21 Study area and presence records

According to Baha El Din (2006) Egypt has at least 109 species ofreptiles 61 lizards 39 snakes a crocodile seven turtles and a tor-toise and Acanthodactylus aegyptius has been added as a distinctspecies since (Baha El Din 2007) The main source of specieslocation records was the BioMAP (wwwbiomapegyptorg) data-base of Egyptian biodiversity records with our own records thedata are comprehensive for the herpetofauna Records for marineor aquatic species from the Nile were excluded because of the lackof GIS predictor layers for aquatic environments

The taxonomic and georeferencing accuracy of these recordswere exhaustively checked assessed and revised We limited ourscope to species with at least eight unique pixels (at a resolution of25 arc minutes) to avoid over-fitting (Baldwin 2009) this meantthat a total of 75 species were processed (49 lizards 25 snakes anda tortoise Table S1) The coverage of the records is good (Fig S1a)they include most of Egypts landscape and habitats There was aninevitable bias in recording effort (represented by the number ofvalid records) towards the main cities and populated areas(Fig S1b) unsurprisingly the highest collection effort was foundaround the greater Cairo district followed by South Sinai (the StKatherine area) the Alexandria area some areas around Fayoumand Wadi El-Natrun and small patches near El-Arish and MersaMatruh (Figs S2eS3 show a map of Egypt overlaid withgeographical locations of areas mentioned in this study and thelocations of currently established Protected Areas respectively)

22 Environmental predictor variables

Climate data for the near past (1950e2000) were downloadedfrom WorldClim Global Climate Data v14 (release 3 e see httpwwwworldclimorgbioclim) (Hijmans et al 2005) These data

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 213

represent current climate conditions and comprise 19 lsquobioclimaticrsquovariables derived from precipitation and temperature records(Table 1) Four (Bio-7 -14 -17 amp -18) were rejected a priori as notsufficiently variable across Egypt reflecting the countrys status asthe driest country in the world (FAO 2012) A resolution of 25 arc-minutes (~5 5 km) best matched the level of positional uncer-tainty that accompanies the records (most of which derive frommuseum records georeferenced by site name) and because theclimate data for Egypt were interpolated using relatively fewweather stations largely concentrated in the Nile Valley and Delta(see Fig 23 in Newbold 2009) This relatively coarse resolution isalso useful tominimise the effect of ignoring species interactions onmodelled distributions (Pearson and Dawson 2003)

Elevation data were obtained from the SRTM Digital ElevationDatabase v41 [available at httpwwwcgiar-csiorgdataelevation] The tiles that cover Egypt were downloaded at 90-mresolution and rescaled to 25 arc minutes We also used an Egyp-tian habitat map from the BioMAP project (A A Hassan unpub-lished data) In this map the habitat of Egypt is classified into 11classes (sea littoral coastal land cultivated land sand dune wadimetamorphic rock igneous rock gravels serir sand sheets sabkhasand sedimentary rocks e see Newbold et al 2009a) NormalizedDifference Vegetation Index (NDVI) data for seven years (from Jan2004 to Dec 2010) were downloaded from the SPOT Vegetationwebsite (see httpfreevgtvitobe) Two layers were derived fromthese maps and used as predictors in the models maximum NDVIvalue (indicating how much vegetation there is per pixel) and thedifference between maximum and minimum NDVI value (indi-cating variability in vegetation per pixel)

To reduce colinearity among predictors and over-fitting theVariance Inflation Factor (VIF) was calculated for the continuouspredictors using R v2150 (R Development Core Team 2012) andused to prune predictors before use (eg Bombi et al 2012) until allremaining variables yielded VIF values below 10 (Table 1) thesewere then used in the Maxent modelling

23 Climate models used

To estimate the potential impact of climate change currentdistribution models were projected into the future for three timeslices (2020s 2050s and 2080s each the median of a 30-yearwindow) Future climate data (IPCC 2007 downloaded fromhttpwwwccafs-climateorg) derived from four independentGlobal Circulation Models (GCMs) were used to obtain consensus

Table 1List of variables used to calculate VIF values rows shaded wiso used to run the models Four (Bio-7 -14 -17 amp -18) were

Altitude AltitudeNDVI_Max NDVI maximNDVI_Difference Absolute diffBio1 Annual meanBio2 Mean diurna

Bio3 IsothermalityBio4 TemperatureBio5 Maximum te

Bio6 Minimum teBio8 Mean tempeBio9 Mean tempe

Bio10 Mean tempeBio11 Mean tempeBio12 Annual preci

Bio13 PrecipitationBio15 Precipitation

Bio16 PrecipitationBio19 Precipitation

estimations about the fate of the herpetofauna These GCMs werethe Hadley Centre Coupled Model version 3 (HadCM3) the Cana-dian second-generation coupled global climate model (CGCM2-CCCma) the Australian model (CSIRO Mk2) and a Japanese model(NIES99) all frequently used in similar studies (eg Arauacutejo et al2006 Holt et al 2009) Our results assume that species try totrack their suitable habitat (dependent on dispersal ability) ratherthan accommodate or adapt to the new conditions

For each GCMmodel and time slice two emission scenarios (A2aand B2a) were used the most commonly used scenarios in climate-change assessments (Hannah 2011) They reflect two differentassumptions about the levels of CO2 emissions linked to assumeddemographic changes and socio-economic and technological de-velopments (Marini et al 2009) The A2a scenario assumes a lot ofclimate change caused by medium to high CO2 emission rates (theldquobusiness as usualrdquo scenario) while the B2a scenario assumesrelatively little change because of mitigation measures (theldquomoderaterdquo scenario) (Sauer et al 2011 Taubmann et al 2011)There are no data on how NDVI and habitat variables might changein Egypt in the future and so we assume they (and altitude) remainconstant Apart from the A2a scenario for 2080 nearly all the ex-pected climatic conditions are similar to current values (see Fig S4)

24 Species distribution modelling

We used Maxent v333k (Phillips et al 2006 e see httpwwwcsprincetonedu~schapiremaxent) to run the species dis-tribution models Maxent performs well in comparison withalternative algorithms and shows equal or higher predictive accu-racy on cross-validation (Elith et al 2006 Hernandez et al 2006)Crucial to our decision not to use ensemble modelling Maxent isvery robust to small numbers of training records outperformsother algorithms when using a relatively small number of occur-rence records (Franklin 2009) and is robust against moderatelocation errors (Graham et al 2008) We used lsquopermutationimportancersquo (see Phillips and ATampT Research 2011) to detect whichpredictor had the greatest influence on the model because it doesnot depend on the path used by the algorithm (Songer et al 2012)

Ten replicated runs with cross-validation and default settings(except iterations frac14 1000) were made for each species by itera-tively partitioning the records to use 90 for training and 10 fortesting (as recommended by Phillips and ATampT Research 2011)this gives a more stable model performance (avoiding overfitting)and minimises the effects of errors and biases The default logistic

th grey show variables with VIF values less than 10 andrejected a priori (see text)

um valueerence between the highest and lowest NDVI valuestemperature

l range (mean of monthly (max temp min temp))

(Bio2Bio7) (100)seasonality (standard deviation100)mperature of the warmest month

mperature of the coldest monthrature of the wettest quarterrature of the driest quarter

rature of the warmest quarterrature of the coldest quarterpitation

of the wettest monthseasonality (coefficient of variation)

of the wettest quarterof the coldest quarter

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221214

output format was chosen ie related to probability of suitableconditions ranging from 0 to 1 Logistic output has been criticizedrecently (eg Merow et al 2013) becausemany users assume that itrepresents the probability of occurrence to do that however thecalculation must assume that overall prevalence is fixed at 05 forall species which may not be accurate estimating true prevalenceis very difficult (Elith et al 2011) Both logistic and raw outputs aremonotonically related and rank locations identically (Elith et al2011) making rank-based measures such as AUC (for model eval-uation Elith et al 2011) and thresholding (using values frompredictions of the training presence locations eg 10th-percentilethreshold) identical whichever is used We used logistic output fortwomain reasons a) it standardises the expected habitat suitabilityvalues across species permitting the adding together of species tocreate the lsquoprobability richnessrsquo maps b) it allows Maxent auto-matically to convert the probabilities into expected presenceabsence (or more precisely suitablenon-suitable) habitat For thelatter we used the 10th-percentile training presence threshold(following Pearson et al 2007) For each species (and for each ofthe logistic and thresholded distributions) the ten replicate dis-tribution maps were converted to a single consensus map byMaxent for logistic values and manually for thresholded values(where each pixel was a consensus presence if it had presencevalues in more than half of the model runs ie gt5)

Distributions were then projected into three future time slices(2020 2050 and 2080) in each of four GCM models (HadCM3CCCma CSIRO and NIES99) and two emission scenarios (A2a andB2a) giving a total of 24 future projections Thresholded futuredistributions involved two extreme assumptions about dispersalability unlimited and no dispersal For each of the 24 projectionsthree consensus maps were produced one using the probability(logistic) output and two using the thresholded distributions(assuming unlimited and no-dispersal) The maps were then aver-aged across the different GCMs for thresholded distributions thisinvolved assigning a pixel as potentially suitable if it had a value ofpresence in more than two GCMs Thus there were 18 (three timeslices two scenarios three consensus) maps per species

25 Comparisons across species

Current and future maps of estimated species richness werecalculated in two different ways first by simply adding togetherthe average probability of suitable conditions for each pixel acrossthe distribution maps of all species (this assumes unlimiteddispersal eg Gilbert and Zalat 2008) and second by addingtogether the thresholded distribution maps (under unlimited- andno-dispersal assumptions eg Trotta-Moreu and Lobo 2010) Mapsof future species richness were obtained for each of the 24 futureprojections and then averaged across the four GCMs Future po-tential changes in species richness (but not composition) werecalculated by subtracting current species richness at each pixel

At the species level gains and losses in future distribution werecalculated using current and future consensus thresholded distri-butions Then the losses across all the 24 future potential distri-butions across all species were added together into a species lossmap showing which areas are expected to suffer most in losingspecies (for each dispersal assumption) The same was carried outfor gains in suitability to produce a species gain map showingwhich areas are expected to gain species (only for the unlimiteddispersal assumption) Species turnover an index of dissimilaritybetween current and future species composition (Thuiller 2004)was calculated for each future projection and dispersal assumption(following Thuiller et al 2005)

In order to assess future extinction risk and to determine whichspecies may require more protection in the future species range

changes were calculated as the percentage loss (or gain) in suitablehabitats carried out for each dispersal assumption Each specieswas classified into one of the following categories Extinct (ex-pected loss of the entire suitable habitat - 100) Critically Endan-gered (loss gt80) Endangered (loss 50e80) Vulnerable (loss30e50) Least Concern (loss lt30) gain 1 (gain lt30) gain 2(gain 30e50) gain 3 (gain 50e80) gain 4 (gain 80e100) andgain 5 (gain gt100) (modified from Thuiller et al 2005) (seeTables S2eS3)

26 Area prioritization for conservation

Several algorithms are now available to prioritize areas forconservation and conservation planning Using the SDM pre-dictions Zonation v31 (Moilanen et al 2012) created a nestedspatial conservation prioritization network to evaluate the effec-tiveness and performance of current Protected Area network inEgypt and identify new sites for conservation Zonation prioritizesthe conservation value of the landscape hierarchically based on theconservation value of sites (cells) (Moilanen et al 2012) by itera-tively removing the least valuable cells from the edges of thelandscape according to set rules The last to be removed are themost important areas (Moilanen et al 2012) We used two removalrules lsquobasic core-area functionrsquo to identify important (or poor)locations where a single or a few species have important occur-rences and lsquoadditive benefit functionrsquo to give more weight to lo-cations with high species richness (Moilanen et al 2012)Fragmentation of the chosen areas is minimized by an aggregationrule and we chose to use lsquodistribution smoothingrsquo for which anestimate of dispersal distance is required The dispersal ability ofalmost all terrestrial reptile species is very limited (Cadby et al2010 Edgar et al 2010) To be conservative in this study thedispersal distance of all reptiles was set to be equal to 1 km

Because some species are more important to conserve thanothers each species was given a weight Weights were developedfrom the global and national IUCN assessments the world distri-bution and distribution patterns within Egypt Following Gilbertand Zalat (2008) and Basuony et al (2010) we classified Egyptianreptiles according to their world distribution and their distributionpattern within Egypt We assessed their IUCN Red List status inEgypt according to IUCN guideline and categories (IUCN Standardsand Petitions Subcommittee 2010) although inevitably incom-plete We gave each element a score (modified from Leach et al2013) to indicate relative importance and then the sum of thescores gave the relative conservation weight of each species (seeTables S1 and S4) We used these values to weight the Zonationprioritization process

For current and future climates Zonation was run using themaps of habitat suitabilities for all species The number of cellsremoved at each iteration was set to 10 (frac14lsquowarp factorrsquo) Zonationproduces ASCII raster files with ranked pixel values scaled to befrom 0 to 1 values gt 07 were considered to be of high conservationimportance These important areas were then overlaid with thecurrent Protected Area network in Egypt to show if the network inEgypt is adequate to conserve the species in the face of climatechange and if there are areas outside the current Protected Areasboundaries that require special protection measures

3 Results

31 Model performance amp variable importance

In terms of mean AUC values for the 10 cross-validation runs theperformance of current-distribution models was good (seeTable S1) ranging from 078 to 099 with an overall mean of

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 215

093 plusmn 005 Only one species (Ptyodactylus siphonorhina) had amean AUC of less than 08 while in 36 species it was greater than095 As almost all species had a mean AUC gt08 all models wereaccepted and processed for further analyses More widespreadspecies tended to have lower AUC values (correlations [n frac14 75]with number of suitable pixels rs frac14 085 p lt 0005 extent ofoccurrence rs frac14 077 p lt 0005 number of pixels with recordsrs frac14 0433 p lt 0005)

Variables with highest mean permutation importance across allmodelled species were altitude (highest for 34 species) Bio4(temperature seasonality 10 species) and Bio13 (precipitation ofwettest month 9 species) Those with the lowest mean permuta-tion importance and thus not influencing the final models verymuch were the difference betweenmaximum andminimumNDVIBio15 (precipitation seasonality) habitat and Bio9 (mean temper-ature of driest quarter) the first and last of these were never themost influential variable for any species

32 Species richness

The patterns of current species richness (Fig 1) were similarusing either logistic or thresholded distributions both emphasizingthe lower Nile Valley the Suez Canal area and the coasts of theMediterranean and Suez Gulf (Fig S2 for locations map) The

Fig 1 Current estimated reptile species richness of Egypt using the sum of (a)Maxent logistic predictions and (b) thresholded predicted distributions

pattern is concordant with the species richness map in Baha El Din(2006) produced using records and qualitative criteria (seeFig S5a) Under climate change regardless of emission scenario ordispersal assumptions several areas are expected to have increasedspecies richness (Fig 2) with generally greater magnitudes underthe A2a scenario while others are expected to decline (eg the SuezCanal area) with greater declines under the B2a scenario by 2050and 2080

Areas with the highest expected loss of species include betweengreater Cairo Ismailia and Suez south of Suez on both sides of theSuez Gulf and Wadi El-Natrun much greater under the B2a sce-nario (Fig 3a) If species can disperse well then the highest gains inspecies are expected to include a large proportion of the northernhalf of the Western Desert again greater under the B2a scenario(Fig 3b) In terms of percentage change of the reptile assemblage ifdispersal is unlimited then the Western Desert is expected to un-dergo high turnover (Fig S6a) but this is because there are so fewspecies there currently Under no dispersal turnover is very limited(Fig S6b) again with highest values again in the Western desertbecause of the species poor assemblages there

33 Range and status changes

With unlimited dispersal no species is expected to becomeldquoextinctrdquo (100 loss of suitable habitat) under all or the average ofall GCMs There are a couple of species expected to become extinctby losing their entire area of suitable habitat in at least one of thefuture projections Tarentola mindiae and Hemidactylus robustusUsing the average gain or loss of suitable habitat across the fourdifferent GCMs eight species are expected to be classified as Crit-ically Endangered (loss gt 80) twenty as Endangered (loss50e80) and twenty-two as Vulnerable (loss 30e50) in at leastone mean future projection (details see Table S2)

When assuming no dispersal at all none of the species are ex-pected to lose all suitable habitats under all or the average of allGCMs but T mindiae and H robustus again experience completeloss of habitat under at least one of the projections Using theaverage loss of suitable habitat across the four different GCMs tenspecies are expected to be classified as Critically Endangeredtwenty-seven as Endangered and thirty-three as Vulnerable in atleast one mean future projection (details see Table S3)

34 Area prioritization for conservation

The areas with the highest current prioritization value for con-servation were similar for both cell-removal methods mainlyemphasizing the north coast Suez Canal area South Sinai theGebel Elba region Qattara Depression Wadi El-Natrun and aroundCairo amp Fayoum with the lsquoadditive benefitrsquo function putting greateremphasis on the Nile Delta and its fringing areas (Fig 4) Thisoverall pattern does not change much under future climate pro-jections though a few areas decline in priority and much greaterarea of the northern Western Desert increases These effects aremore pronounced with the lsquocore-area functionrsquo used as the cellremoval rule

The mean prioritization value in all models was higher in Pro-tected Areas than outside them (Fig 5) with both slightlyincreasing in future projections The difference in prioritizationvalue between inside and outside Protected Areas declined withtime especially by 2050 and 2080 under the B2a scenario

4 Discussion

The results of this study show for first time the potential impactsof climate change on the distribution of Egyptian reptiles Some

Fig 2 Estimated future changes in species richness as a result of anthropogenic climate change under two IPCC scenarios (A2a B2a) and at three times in the future (2020 20502080) Each map is the difference from the current species richness (shown in Fig 1a) and is averaged across four GCMmodels Maps made (a) using Maxent logistic output and (b)using thresholded distributions (see Methods)

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221216

reptiles are expected to lose much of their currently suitabilityareas (up to 80 in some cases) and some areas will lose or changea large proportion of their species Thus there will be a need forgreater conservation efforts in the future Some of these areas arealready inside Protected Areas and so hypothetically they are pro-tected while others are either just outside or very far from Pro-tected Areas Although overall Protected Areas contain better(higher priority) values compared to unprotected sites the currentProtected-Area network appears inadequate for future conserva-tion Our results together with others (eg Leach et al 2013) willhelp the Egyptian Environmental Affairs Agency and otherdecision-making parties in Egypt to direct conservation efforts andlimited conservation resources in the right direction

41 Model performance amp variable importance

The negative correlations found between AUC and the areaoccupied and with the extent of occurrence concur with findingsof many other studies (eg Brotons et al 2004 Elith et al 2006Hernandez et al 2006) Rare species are usually habitat special-ists with narrow environmental tolerances and are environmen-tally or geographically restricted compared to widespread species

(Elith et al 2006 Pandit et al 2009) Widespread species are likelyto be generalists occupying a wide range of habitats and climatesmaking it difficult to distinguish between suitable and unsuitablehabitats (Franklin et al 2009) However as pointed out by Elithet al (2006) this may be an artefact an inevitable result for spe-cies with small relative occurrence areas and perhaps a reason notto use AUC to compare model performances (Lobo et al 2008)Most (about 60) of the Egyptian herpetofauna are narrowlydistributed occupying less than 10 of Egypts area (Baha El Din2006) There is no evidence of a strong correlation betweenmodel accuracy and the number of records used (see Elith et al2006 Newbold et al 2009b) and the weak negative correlationfound here is unusual since most (eg Stockwell and Peterson2002 Hernandez et al 2006) find that models with larger sam-ple sizes have higher accuracy (but see de Pous et al 2011)

We found altitude to be the most effective predictor for manyspecies In contrast the NDVI predictors were not useful unlike inmany other cases (eg Egbert et al 2002 Anderson et al 2006Kgosiesele 2010 Taheri 2010) but perhaps typical for arid envi-ronments (eg Soultans (2011) study on Egyptian antelopes) or forherpetofauna (eg Teodoro et al 2013) Habitat categories likewisemade only a low contribution to the reptile models

Fig 3 Expected patterns of cumulative (a) losses and (b) gains of species as a result of climate change The losses of each species in each pixel (under both dispersal assumptions)and gains (under unlimited dispersal only) were averaged across GCM models accumulated across species and then plotted

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 217

42 Species richness

Species richness by itself is not adequate as an indicator of bioticchange since species composition can change independently Ourresults suggest that in the future some areas with currently highexpected species richness may lose many species or undergo largechanges in species composition Some of these areas are alreadyunder a certain degree of protection but even if they can bemaintained or enhanced (eg Siwa oasis Gebel Elba and the coastalareas over the Aqaba gulf Fig S2) current Protected Areas are notsufficient to prevent expected loss of species Other areas (eg WadiEl-Natrun the Suez Canal area Red Sea inland wadis the poorlyknown Qattara Depression and the areas around Gebel El-Gallalaand Gebel El-Hallal) are unprotected and measures are needed toconserve them There are similar situations elsewhere for examplehigh species loss and turnover in the reptiles of a priority conser-vation hotspot in Europe (the Iberian Peninsula) are expected (dePous 2011) with great variation in expected patterns of loss andchange with different projections and dispersal assumptions Otherstudies have also concluded that large changes are likely in chelo-nians worldwide (Ihlow et al 2012) Mediterranean mammals(Maiorano et al 2011) US trees and vertebrates (Currie 2001) andMesoAmerican trees (Golicher et al 2012)

43 Range and status changes

Two species are expected to lose their suitable habitats entirelyin at least one future projection 1) T mindiae is a near-endemicspecies recorded just from northwest Egypt and northern Cyrena-ica (eastern Libya) its distribution in Egypt is restricted to Siwaoasis the Qattara Depression and their periphery (Baha El Din2006) Future range changes suggest it will be Endangered (loss50e80) by 2020 and as Critically Endangered (loss gt80) by 2050and 2080 (under all emission scenarios and dispersal assumptions)2) H robustus is localized to the East African coast from Zanzibar tosouthern Egypt Arabia east to Pakistan its distribution in Egypt ison the Red Sea coast from El-Quseir southwards (Baha El Din2006) The A2a scenario suggests it will be Critically Endangeredby 2020 and 2050 while under the B2a scenario it will be Endan-gered by 2020 and Critically Endangered by 2050 and 2080 Morethan half of its current distribution in Egypt is located within theWadi El-Gemal and Gebel Elba Protected Areas so hypothetically itis protected However it is found mainly on buildings and henceassociated with man it may in fact be an invasive to Egypt

Another eight species are expected to be classified as CriticallyEndangered (loss gt80) in at least one averaged future projection(Acanthodactylus longipes Cerastes vipera Eryx jaculus Eumeces

Fig 4 Current conservation prioritizationl of areas according to the Zonation algo-rithm under two cell removal rules (a) core-area function (identifies areas where oneor more species have important occurrences) and (b) additive benefit function (whichgives greater emphasis to areas with high species richness) Darker colours representhigher conservation value (For interpretation of the references to colour in this figurelegend the reader is referred to the web version of this article)

Fig 5 Mean prioritization value for conservation (plusmn95 confidence limits) derivedfrom Zonation for Protected Areas (PAs) and outside Protected Areas (non-PAs) forcurrent and expected future conditions using the cell-removal rule of (a) core-areafunction and (b) additive benefit function

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221218

schneiderii Malpolon moilensis Ptyodactylus guttatusP siphonorhina and Trapelus mutabilis) According to the currentassessment twenty-seven species are classified as Endangered(loss 50e80) in at least one projection (mostly by 2080 in the A2scenario) two of these are classified assuming no dispersal asEndangered in all future projections (Uromastyx aegyptia and Pris-turus flavipunctatus for details see Tables S2eS3)

Reptiles are also likely to gain from global warming especially inmore northern areas of the globe if they can disperse and expandtheir distributions but under perhaps the more realistic assump-tion of no dispersal most species were expected to lose habitatsuitability by 2050 (Arauacutejo et al 2006) In Egypt areas of clearfuture expected gain (assuming unlimited dispersal) are the areabetween the Siwa oasis and Qattara Depression northwards almostto the northern coastal strip and some sparse locations in centralSinai There were greater gains under the B2a scenario in theeastern part of the Western Desert and some patchy areas aroundthe Nile (Fig 3b) This pattern is similar to that reported by Araujoet al (2006)

Some of the species identified here as being most at risk due tofuture climate change are in fact resilient species that are ecolog-ically very adaptable and are currently experiencing range expan-sion and population growth at least in Egypt (eg H robustus) This

disparity is probably because modelling is based solely on theprobable impacts of climate change on the spatial distribution ofsuitable habitats without taking evolution or ecological adapt-ability and resilience into account Species vary greatly in theirresponse to ecological change some adapting very well even tovery rapid anthropogenically induced changes (eg invasive spe-cies) while others are much less adaptable and tend to go extinctfirst

There are no published studies that discuss the effect of climatechange on Egyptian reptiles obviating any comparisons Informa-tion on the reptiles of the adjacent countries of Sudan and Libya arevery limited making it impossible to assess any movements thatmight compensate for Egyptian losses The patterns of expectedchanges in Leach et als (2013) similar study on mammals andbutterflies do not concur with ours Mammal species richness wasexpected to decline in the Mediterranean and Red Sea areas (by40e60) and increase elsewhere (by 80e100) while for butter-flies declines are expected over almost the whole of Egypt (by40e60) except the south which is expected to increase (by40e60) In the models of European reptiles (de Pous 2011)

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 219

between eight (unlimited dispersal) and 21 (no dispersal) specieswere expected to become extinct in at least one future projectionwith two more losing more than 80 of their currently suitablehabitats This is a much more damaging scenario than the expectedresponses of Egyptian reptiles in our study There are no othercomparable studies from North Africa or the Middle East

44 Area prioritization for conservation

In our study Protected Areas (Fig S3) were more valuable inconserving reptiles than areas outside them both currently and inthe future although the difference is expected to get smaller withtime (Fig 5) Although this is encouraging the Protected Areanetwork appears inadequate to conserve Egyptian reptiles NewProtected Areas that have already been proposed would go someway towards reducing this deficiency in coverage although otherswill be needed in the middle to north Sinai (especially aroundGebel El-Hallal) the Suez Canal area both sides of the Suez Gulf thenorthern Mediterranean coast coastal wadis of the Red Sea be-tween Hurghada and Mersa Alam and the Gebel El-Gallala areaMore effective protection is required in the established ProtectedAreas including those in South Sinai Gebel Elba and Siwa oasis dueto their relatively high expected future suitability loss The Pro-tected Areas of Siwa oasis and El-Gilf El-Kebir plus the areas ofWadiEl-Natrun the Gebel El-Hallal area and wadis of the Red Sea areexpected to have particularly high turnover in species composition(Fig S6) and hence probably need more conservation efforts Somecomparable information is available for Egypts butterflies andmammals (Gilbert and Zalat 2008 Basuony et al 2010 Leach et al2013) Egypts network of Protected Areas appears adequate toconserve Egyptian butterfly hotspots except for the Mediterraneancoast between Alexandria and Sallum (Gilbert and Zalat 2008) butinadequate to conserve important mammal areas with new Pro-tected Areas needed to be constructed in the lower Nile Valleyalong the north coast between Alexandria and Sallum top part ofthe Suez Gulf and perhaps the Qattara Depression (Basuony et al2010) Suggestions for conserving mammals are consistent to someextent with our results which also suggest new Protected Areasalong the Suez Gulf Mediterranean coast from Mersa Matruh toSallum and the Qattara Depression Leach et al (2013) concur withour results in suggesting priority areas along the western Medi-terranean coast around Greater Cairo central and north Sinai andthe Red Sea coast

45 Limitations

There are some limitations of the results shown in this study 1)Egyptian climate data in WorldClim are interpolated from ratherfew non-randomly placed weather stations in Egypt and adjacentcountries Although making the climate data less reliable thanelsewhere there are no alternative climate data available 2) Anensemble approach (eg using BIOMOD Thuiller et al 2009) isoften suggested for SDM modelling to avoid dependence on oneparticular algorithm This is not possible without a greater numberof records since other algorithms are not as flexible or robust asMaxent Despite all the caveats with using SDMs and criticisms ofMaxent in particular (Royle et al 2012 but see Hastie and Fithian2013 Elith et al 2006 Hernandez et al 2006) in practice coun-tries such as Egypt have little else to guide them in attempts atclimate change adaptation 3) There were signs of sampling bias inthe available records Although this violates the assumptions of allSDMs it is a problem common to nearly all datasets worldwideNew techniques may allow for the correction of sampling bias egusing target-group approach (Phillips et al 2009) although thedata requirements are currently impossible for Egypt and most

other developing countries) 4) Using Maxents default settings iscriticized by some authors they need to be modified dependent onthe data (Warren and Seifert 2011 Radosavljevic and Anderson2014 Merow et al 2013) but we think using the default settingsis enough in this study

5 Conclusion

Although Egyptian datasets exist knowledge of Egyptian spe-cies and conservation priorities are limited (with consequencesexplored in Leach et al 2013) nevertheless our results and those ofLeach et al (2013) encourage us to believe that useful insights canbe gained for conservation priorities by using the combination ofspecies distribution modelling plus spatial conservation prioriti-zation techniques (Hannah 2010 Klorvuttimontara et al 2011) Itis crucial for the Egyptian Environmental Affairs Agency to establishan accurate and reliable database of records assessed by experts tohelp improve the basic information for future conservation as-sessments and analyses

Although many SDM studies have been done on a variety oforganisms in recent decades aiming to solve conservation prob-lems evidence of their practical utility in real-world conservation issparse with very few studies clearly targeting conservation de-cisions (Guisan et al 2013) If implemented correctly SDMs canplay a critical role in supporting spatial conservation decision-making (Addison et al 2013) For reptiles there is a shortage ofstudies estimating their response to expected climate changes(Arauacutejo et al 2006) Although some suggest that reptiles (and to agreater extent amphibians) will be strongly affected by climatechange (eg Gibbons et al 2000 Arauacutejo et al 2006 Carvalho et al2010) very few of themwere concerned with arid environments orthe Middle East We hope that this and other similar studies thatestimate the fate of reptiles under climate change help decision-making authorities to take the measures necessary to conservethem

Acknowledgements

We thank the Italian Cooperation (Debt Swap) in Egypt forfunding the BioMAP project that allowed the collation of the da-tabases used in this work Dr Tim Newbold for allowing the use ofthe data from his field trips Dr Mustafa Fouda (Former Director ofthe Nature Conservation Sector EEAA) for facilities offered duringBioMAP project and all the BioMAP staff This work was supportedby the award of a Chevening Scholarship Masters grant to AEl-GWe are especially grateful to two anonymous referees whogreatly improved an earlier manuscript

Appendix A Supplementary data

Supplementary data related to this article can be found at httpdxdoiorg101016jjaridenv201512007

References

Addison PFE Rumpff L Bau SS Carey JM Chee YE Jarrad FC McBride MFBurgman MA 2013 Practical solutions for making models indispensable inconservation decision-making Divers Distrib 19 (5e6) 490e502 httpdxdoiorg101111ddi12054

Anderson RP Peterson AT Egbert SL 2006 Vegetation-index models predictareas vulnerable to purple loosestrife (Lythrum salicaria) invasion in KansasSouthwest Nat 51 471e480 httpdxdoiorg1018940038-4909(2006)51[471VMPAVT]20CO2

Arauacutejo MB 2009 Climate change and spatial conservation planning InMoilanen A Wilson KA Possingham HP (Eds) Spatial Conservation Prior-itization Quantitative Methods and Computational Tools Oxford UniversityPress pp 172e184 ISBN 9780199547760

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221220

Arauacutejo MB Rahbek C 2006 How does climate change affect biodiversity Sci-ence 313 1396e1397 httpdxdoiorg101126science1131758

Arauacutejo MB Thuiller W Pearson RG 2006 Climate warming and the decline ofamphibians and reptiles in Europe J Biogeogr 33 1712e1728 httpdxdoiorg101111j1365-2699200601482x

Baha El Din SM 2001 The Herpetofauna of Egypt Species Communities andAssemblages University of Nottingham p 435 Unpublished PhD thesis

Baha El Din SM 2006 A Guide to the Reptiles and Amphibians of Egypt TheAmerican University in Cairo Press ISBN 9789774249792

Baha El Din SM 2007 A new lizard of the Acanthodactylus scutellatus group(Squamata Lacertidae) from Egypt Zool Middle East 40 21e32 httpdxdoiorg10108009397140200710638200

Baldwin RA 2009 Use of maximum entropy modeling in wildlife research En-tropy 11 854e866 httpdxdoiorg103390e11040854

Ball IR Possingham HP Watts M 2009 Marxan and relatives software forspatial conservation prioritisation In Moilanen A Wilson KAPossingham HP (Eds) Spatial Conservation Prioritization QuantitativeMethods and Computational Tools Oxford University Press pp 185e195 ISBN9780199547760

Basuony MI Gilbert F Zalat S 2010 The Mammals of Egypt Atlas Red DataListing amp Conservation BioMAP amp CultNat EEAA amp Bibliotheca AlexandrinaCairo available at httpecologynottinghamacuk~plzfg

Bombi P Salvi D Bologna MA Pettorelli N 2012 Cross-scale predictions allowthe identification of local conservation priorities from atlas data Anim Conserv15 378e387 httpdxdoiorg101111j1469-1795201200526x

Brotons L Thuiller W Arauacutejo MB Hirzel AH 2004 Presence-absence versuspresence-only modelling methods for predicting bird habitat suitability Ecog-raphy 27 437e448 httpdxdoiorg101111j0906-7590200403764x

Cadby CD While GM Hobday AJ Uller T Wapstra E 2010 Multi-scaleapproach to understanding climate effects on offspring size at birth and date ofbirth in a reptile Integr Zool 5 164e175 httpdxdoiorg101111j1749-4877201000201x

Carvalho SB Brito JC Crespo EJ Possingham HP 2010 From climate changepredictions to actions e conserving vulnerable animal groups in hotspots at aregional scale Glob Change Biol 16 (12) 3257e3270 httpdxdoiorg101111j1365-2486201002212x

Currie DJ 2001 Projected effects of climate change on patterns of vertebrate andtree species richness in the conterminous United States Ecosystems 4216e225 httpdxdoiorg101007s10021-001-0005-4

de Pous P 2011 Climate Change and European Reptiles from Large-scale SpatialRange Shift Projections towards a More Realistic Assessment of Species Re-sponses University of East Anglia UK p 64 MSc thesis

de Pous P Beukema W Weterings M Duumlmmer I Geniez P 2011 Area priori-tization and performance evaluation of the conservation area network for theMoroccan herpetofauna a preliminary assessment Biodivers Conserv 2089e118 httpdxdoiorg101007s10531-010-9948-0

Edgar P Foster J Baker J 2010 Reptile Habitat Management HandbookAmphibian and Reptile Conservation Bournemouth

Egbert SL Martiacutenez-Meyer E Ortega-Huerta M Peterson AT 2002 Use ofdatasets derived from time-series AVHRR imagery as surrogates for land covermaps in predicting species distributions In Proceedings IEEE 2002 Interna-tional Geoscience and Remote Sensing Symposium (IGARSS) vol 4pp 2337e2339 httpdxdoiorg101109IGARSS20021026537

El Alqamy H Ismael A Abdelhameed A Nagy A Hamada A et al 2010 Pre-dicting the status and distribution of the Nubian Ibex (Capra nubiana) in thehigh-altitude mountains of south Sinai (Egypt) In Galemys Newsletter of theSpanish Society for the Preservation and Study of Mammals vol 22pp 517e530

Elith J Graham CH Anderson RP Dudiacutek M Ferrier S et al 2006 Novelmethods improve prediction of species distributions from occurrence dataEcography 29 129e151 httpdxdoiorg101111j20060906-759004596x

Elith J Phillips SJ Hastie T Dudiacutek M Chee YE Yates CJ 2011 A statisticalexplanation of MaxEnt for ecologists Divers Distrib 17 (1) 43e57 httpdxdoiorg101111j1472-4642201000725x

FAO 2012 Food and Agriculture Organisation World Development IndicatorsAverage Precipitation in Depth (mm per year) httpdataworldbankorgindicatorAGLNDPRCPMM

Franklin J 2009 Mapping Species Distributions Spatial Inference and PredictionCambridge University Press Cambridge UK ISBN 9780521700023

Franklin J Wejnert KE Hathaway SA Rochester CJ Fisher RN 2009 Effect ofspecies rarity on the accuracy of species distribution models for reptiles andamphibians in southern California Divers Distrib 15 167e177 httpdxdoiorg101111j1472-4642200800536x

Gibbons JW Scott DE Ryan TJ Buhlmann KA Tuberville TD Metts BSGreene JL Mills T et al 2000 The global decline of reptiles Deja Vu Am-phibians BioScience 50 (8) 653e666 httpdxdoiorg1016410006-3568(2000)050[0653TGDORD]20CO2

Gilbert F Zalat S 2008 The Butterflies of Egypt Atlas Red Data Listing amp Con-servation BioMAP EEAA Cairo available at httpecologynottinghamacuk~plzfg

Golicher DJ Cayuela L Newton AC 2012 Effects of climate change on the po-tential species richness of Mesoamerican forests Biotropica 44 284e293httpdxdoiorg101111j1744-7429201100815x

Graham CH Elith J Hijmans RJ Guisan A Peterson AT et al 2008 Theinfluence of spatial errors in species occurrence data used in distribution

models J Appl Ecol 45 239e247 httpdxdoiorg101111j1365-2664200701408x

Guisan A Tingley R Baumgartner JB Naujokaitis-Lewis I Sutcliffe PRTulloch AIT Regan TJ Brotons L et al 2013 Predicting species distributionsfor conservation decisions Ecol Lett 16 (12) 1424e1435 httpdxdoiorg101111ele12189

Hannah L 2010 A global conservation system for climate-change adaptationConserv Biol 24 70e77 httpdxdoiorg101111j1523-1739200901405x

Hannah L 2011 Climate Change Biology Elsevier Ltd ISBN 9780123741820Hannah L Lovejoy TE Schneider SH 2005 Biodiversity and climate change in

context In Lovejoy TE Hannah L (Eds) Climate Change and BiodiversityYale University Press New Haven CT USA

Hannah L Midgley G Andelman S Arauacutejo M Hughes G et al 2007 Protectedarea needs in a changing climate Front Ecol Environ 5 131e138 httpdxdoiorg1018901540-9295(2007)5[131PANIAC]20CO2

Hastie T Fithian W 2013 Inference from presence-only data the ongoing con-troversy Ecography 36 864e867 httpdxdoiorg101111j1600-0587201300321x

Hernandez PA Graham CH Master LL Albert DL 2006 The effect of samplesize and species characteristics on performance of different species distributionmodeling methods Ecography 29 773e785 httpdxdoiorg101111j0906-7590200604700x

Hijmans RJ Cameron SE Parra JL Jones PG Jarvis A 2005 Very high reso-lution interpolated climate surfaces for global land areas Int J Climatol 251965e1978 httpdxdoiorg101002Joc1276

Holt AC Salkeld DJ Fritz CL Tucker JR Gong P 2009 Spatial analysis ofplague in California niche modeling predictions of the current distribution andpotential response to climate change Int J Health Geogr 8 38 httpdxdoiorg1011861476-072X-8-38

Hoyle M James M 2005 Global warming human population pressure andviability of the worlds smallest butterfly Conserv Biol 19 1113e1124 httpdxdoiorg101111j1523-1739200500166x

Ihlow F Dambach J Engler JO Flecks M Hartmann T et al 2012 On the brinkof extinction How climate change may affect global chelonian species richnessand distribution Glob Change Biol 18 1520e1530 httpdxdoiorg101111j1365-2486201102623x

IPCC 2007 Climate Change 2007 Synthesis Report Contribution of WorkingGroups I II and III to the Fourth Assessment Report of the IntergovernmentalPanel on Climate Change IPCC Geneva Switzerland

Kgosiesele E 2010 Predictive Distribution Modelling of Timon Lepida in SpainUniversity of Twente (ITC) the Netherlands p 87 MSc thesis

Klorvuttimontara S McClean CJ Hill JK 2011 Evaluating the effectiveness ofprotected areas for conserving tropical forest butterflies of Thailand BiolConserv 144 2534e2540 httpdxdoiorg101016jbiocon201107012

Leach K Zalat S Gilbert F 2013 Egypts protected area network under futureclimate change Biol Conserv 159 490e500 httpdxdoiorg101016jbiocon201211025

Lobo JM Jimenez-Valverde A Real R 2008 AUC a misleading measure of theperformance of predictive distribution models Glob Ecol Biogeogr 17145e151 httpdxdoiorg101111j1466-8238200700358x

Maiorano L Falcucci A Zimmermann NE Psomas A Pottier J et al 2011 Thefuture of terrestrial mammals in the Mediterranean basin under climatechange Philosophical Trans R Soc Lond Ser B Biol Sci 366 2681e2692httpdxdoiorg101098rstb20110121

Marini MA Barbet-Massin M Lopes LE Jiguet F 2009 Predicted climate-driven bird distribution changes and forecasted conservation conflicts in aneotropical savanna Conserv Biol 23 1558e1567 httpdxdoiorg101111j1523-1739200901258x

Merow C Smith MJ Silander JA 2013 A practical guide to MaxEnt for modelingspecies distributions what it does and why inputs and settings matterEcography 36 (10) 1058e1069 httpdxdoiorg101111j1600-0587201307872x

Moilanen A Meller L Leppeuroanen J Pouzols FM Arponen A et al 2012 Zona-tion Spatial Conservation Planning Framework and Software v31 User Manualwwwhelsinkifibioscienceconsplan

Newbold T 2009 The Value of Species Distribution Models as a Tool for Conser-vation and Ecology in Egypt and Britain University of Nottingham UK p 312PhD thesis

Newbold T Gilbert F Zalat S El-Gabbas A Reader T 2009a Climate-basedmodels of spatial patterns of species richness in Egypts butterfly and mammalfauna J Biogeogr 36 2085e2095 httpdxdoiorg101111j1365-2699200902140x

Newbold T Reader T Zalat S El-Gabbas A Gilbert F 2009b Effect of charac-teristics of butterfly species on the accuracy of distribution models in an aridenvironment Biodivers Conserv 18 3629e3641 httpdxdoiorg101007s10531-009-9668-5

Pandit SN Kolasa J Cottenie K 2009 Contrasts between habitat generalists andspecialists an empirical extension to the basic metacommunity frameworkEcology 90 2253e2262 httpdxdoiorg10189008-08511

Pearson RG Dawson TP 2003 Predicting the impacts of climate change on thedistribution of species are bioclimate envelope models useful Glob EcolBiogeogr 12 361e371 httpdxdoiorg101046j1466-822X200300042x

Pearson RG Raxworthy CJ Nakamura M Peterson AT 2007 Predicting speciesdistributions from small numbers of occurrence records a test case using

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 221

cryptic geckos in Madagascar J Biogeogr 34 102e117 httpdxdoiorg101111j1365-2699200601594x

Phillips SJ ATampT Research 2011 A Brief Tutorial on Maxent Available at httpwwwcsprincetonedu~schapiremaxenttutorialtutorialdoc (Last accessed290512)

Phillips SJ Anderson RP Schapire RE 2006 Maximum entropy modeling ofspecies geographic distributions Ecol Model 190 231e259 httpdxdoiorg101016jecolmodel200503026

Phillips SJ Dudiacutek M Elith J Graham CH Lehmann A Leathwick J Ferrier S2009 Sample selection bias and presence-only distribution models implica-tions for background and pseudo-absence data Ecol Appl 19 181e197 httpdxdoiorg10189007-21531

R Development Core Team 2012 R a Language and Environment for StatisticalComputing R Foundation for Statistical Computing Vienna Austria ISBN 3-900051-07-0 httpwwwR-projectorg

Radosavljevic A Anderson RP 2014 Making better Maxent models of speciesdistributions complexity overfitting and evaluation J Biogeogr 41 (4)629e643 httpdxdoiorg101111jbi12227

Royle JA Chandler RB Yackulic C Nichols JD 2012 Likelihood analysis ofspecies occurrence probability from presence-only data for modelling speciesdistributions Methods Ecol Evol 3 545e554 httpdxdoiorg101111j2041-210X201100182x

Sauer J Domisch S Nowak C Haase P 2011 Lowmountain ranges summit trapsfor montane freshwater species under climate change Biodivers Conserv 203133e3146 httpdxdoiorg101007s10531-011-0140-y

Songer M Delion M Biggs A Huang Q 2012 Modeling impacts of climatechange on Giant Panda Habitat Int J Ecol 2012 1e12 httpdxdoiorg1011552012108752

Soultan AE 2011 Antelope distribution in Egypt Modeling the Impact of ClimateChange on Antelopes Distribution in Egypt Lambert Academic Publishing

Standards IUCN Subcommittee Petitions 2010 Guidelines for Using the IUCNRed List Categories and Criteria Version 81

Stockwell DRB Peterson AT 2002 Effects of sample size on accuracy of speciesdistribution models Ecol Model 148 1e13 httpdxdoiorg101016S0304-3800(01)00388-X

Suarez AV Tsutsui ND 2004 The value of museum collections for research andsociety Bioscience 54 66 httpdxdoiorg1016410006-3568(2004)054[0066tvomcf]20co2

Taheri S 2010 Hyper Temporal NDVI Images for Modelling and Prediction theHabitat Distribution of Balkan Green Lizard (Lacerta Trilineata) e Case Study

Crete (Greece) University of Twente (ITC) the Netherlands p 54 MScthesis

Taubmann J Theissinger K Feldheim KA Laube I Graf W et al 2011Modelling range shifts and assessing genetic diversity distribution of themontane aquatic mayfly Ameletus inopinatus in Europe under climate changescenarios Conserv Genet 12 503e515 httpdxdoiorg101007s10592-010-0157-x

Teodoro AC Sillero N Alves S Duarte L 2013 Correlation between the habitatsproductivity and species richness (amphibians and reptiles) in Portugal throughremote sensed data In Proc SPIE 8887 Remote Sensing for Agriculture Eco-systems and Hydrology httpdxdoiorg101117122028502

Thomas CD Cameron A Green RE Bakkenes M Beaumont LJ et al 2004Extinction risk from climate change Nature 427 145e148 httpdxdoiorg101038nature02121

Thuiller W 2004 Patterns and uncertainties of species range shifts under climatechange Glob Change Biol 10 2020e2027 httpdxdoiorg101111j1365-2486200400859x

Thuiller W 2007 Biodiversity climate change and the ecologist Nature 448550e552 httpdxdoiorg101038448550a

Thuiller W Lavorel S Arauacutejo MB Sykes MT Prentice IC 2005 Climate changethreats to plant diversity in Europe Proc Natl Acad Sci U S A 1028245e8250 httpdxdoiorg101073pnas0409902102

Thuiller W Lafourcade B Engler R Arauacutejo MB 2009 BIOMOD e a platform forensemble forecasting of species distributions Ecography 32 (3) 369e373httpdxdoiorg101111j1600-0587200805742x

Tolba MK Saab NW 2009 Arab environment Climate Change - Impact ofClimate Change on Arab Countries Arab Forum for Environment andDvelopment

Trotta-Moreu N Lobo JM 2010 Deriving the species richness distribution ofGeotrupinae (Coleoptera Scarabaeoidea) in Mexico from the overlap of indi-vidual model predictions Environ Entomol 39 42e49 httpdxdoiorg101603EN08179

Warren DL Seifert SN 2011 Ecological niche modeling in Maxent the impor-tance of model complexity and the performance of model selection criteriaEcol Appl 21 (2) 335e342 httpdxdoiorg10189010-11711

Wilson KA Cabeza M Klein CJ 2009 Fundamental concepts of spatial con-servation prioritisation In Moilanen A Wilson KA Possingham HP (Eds)Spatial Conservation Prioritization Quantitative Methods and ComputationalTools Oxford University Press pp 16e27 ISBN 9780199547760

Page 2: Journal of Arid Environments - elgabbas.github.io · e Faculty of Sciences & Arts, Taibah University, Al-Ula Campus, Saudi Arabia f Department of Zoology, Suez Canal University, Ismailia,

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221212

likely to establish (Franklin 2009) This helps to manage speciesfacing possible future threats by identifying biological corridors fordispersal determining sites for re-introduction and areas requiremore protection measures (Thuiller 2007)

Climate change will potentially affect the biodiversity and spe-cies composition of Egyptian ecosystems (Tolba and Saab 2009)but only a handful of quantitative studies of the local faunaflorahave been conducted using SDMs This may be because the modelsare relatively new and because until very recently reliablecomprehensive biodiversity records for the Egyptian fauna andflora were not available As a developing country Egypt lacks arecording scheme to collate species sightings (either at national oreven at local Protected Area level) Between 2004 and 2008 allavailable biodiversity records were collated by the BioMAP project(see httpwwwbiomapegyptorg) and studies of butterflies andmammals using SDMs became possible (Gilbert and Zalat 2008Basuony et al 2010 Newbold 2009 Newbold et al 2009ab) Afew other SDM studies have been published El Alqamy et al (2010)to estimate the potential distribution of the Nubian Ibex (Capranubiana) in South Sinai Soultan (2011) to test the potential impactof climate change on the distribution of four Egyptian antelopespecies and Leach et al (2013) who assessed the effect of climatechange on the future effectiveness of the Protected Area networkusing data on Egyptian butterflies and mammals The only otherstudy of the effects of climate change in Egypt by Hoyle and James(2005) on the worlds smallest butterfly the Sinai Baton Blue(Pseudophilotes sinaicus) used an occupancy-based PopulationViability Analysis

The global current Protected Areas seem not to overlap wellwith areas of high biodiversity value and are traditionally deter-mined spatially and environmentally under the assumption ofrelatively low changes in species distribution in the future (Arauacutejo2009 Leach et al 2013) As climate change is expected to affect thefuture distributionrange of many species globally (potentiallymoving some species out from Protected Areas Hannah et al2007) it challenges the future effectiveness of current ProtectedAreas Future conservation investments should be effectivelyprioritized due to the limited resources available (Wilson et al2009 especially in the developing countries) and early actionsmay be more effective and less costly than delayed or no actions(Hannah et al 2007) Conservation prioritization can be at taxo-nomic (for the protection of some rare or endangered species) orspatial (conserving a particular habitat type or species rich areaseg potential Protected Areas) scale Spatial conservation prioriti-zation uses spatial quantitative data to identify areas of high con-servation priority (Wilson et al 2009) and some techniques havebeen available recently eg Zonation (Moilanen et al 2012) andMarxan (Ball et al 2009) Some of those techniques can use spatialoutputs from SDMs to prioritize areas for conservation under cur-rent and future climates

To date there are 30 Protected Areas in Egypt covering gt15 ofits total area Their distribution shows good spatial coveragealthough some areas of high diversity importance (especially atthe Nile Valley and Delta) are not yet protected (Newbold et al2009a) They were declared relatively recently (first in 1983)and were determinedmainly based on experts known knowledgeof the country biodiversity (Newbold et al 2009a) The capacity ofcurrent Protected Areas in Egypt to mitigate for potential impactsof climate change on different species groups is not well-investigated yet and a qualitative assessment of their futureeffectiveness is highly required Also spatial prioritization of theEgypts landscape (inside and outside the Protected Areas) isrequired to identify potential locations for future Protected Areasand identify current Protected Areas need more conservationeffort in the future

As a developing country Egypt lacks enough good-quality datato be used directly for spatial conservation prioritization but SDMscan provide valuable estimates for species suitabilities in the spaceIn this study we use data for the Egyptian reptiles for the first timeto run SDMs (as a representation group for the Egyptian fauna)Baha El Din (2001 2006) presented a geographic interpretation ofEgypts herpetofaunal distribution qualitatively identifying prior-ity conservation areas but very little has been published on howthe Egyptian herpetofauna may respond to climate change Weused Maxent to model the distribution of Egyptian reptiles undercurrent climate conditions then models are projected into thefuture to show how collectively and individual species are expectedto respond to future climate change under different assumptionsFor each species future expected range change ( loss or gain ofcurrently suitable habitats) is estimated aiming to shed light onsome species require more strict conservation actions Expectedreptiles species richness (under current and future climates) isestimated to identify areas of current high reptile suitability andareas expected to undergomuch changes in suitability in the futureModel predictions were used for prioritizing the Egyptian land-scape under current and future climates We used Zonation soft-ware (Moilanen et al 2012) to assess the likely effectiveness ofEgypts Protected Area network under current and future climateOutputs from Zonation represent hierarchic ranking of the land-scape for conservation and can be easily visualized as maps Wehope the results of the current study to be useful for biodiversityconservation in Egypt and (along with the results of relativestudies) to be taken into account by the decision makers in futurenational conservation plans

2 Methods

21 Study area and presence records

According to Baha El Din (2006) Egypt has at least 109 species ofreptiles 61 lizards 39 snakes a crocodile seven turtles and a tor-toise and Acanthodactylus aegyptius has been added as a distinctspecies since (Baha El Din 2007) The main source of specieslocation records was the BioMAP (wwwbiomapegyptorg) data-base of Egyptian biodiversity records with our own records thedata are comprehensive for the herpetofauna Records for marineor aquatic species from the Nile were excluded because of the lackof GIS predictor layers for aquatic environments

The taxonomic and georeferencing accuracy of these recordswere exhaustively checked assessed and revised We limited ourscope to species with at least eight unique pixels (at a resolution of25 arc minutes) to avoid over-fitting (Baldwin 2009) this meantthat a total of 75 species were processed (49 lizards 25 snakes anda tortoise Table S1) The coverage of the records is good (Fig S1a)they include most of Egypts landscape and habitats There was aninevitable bias in recording effort (represented by the number ofvalid records) towards the main cities and populated areas(Fig S1b) unsurprisingly the highest collection effort was foundaround the greater Cairo district followed by South Sinai (the StKatherine area) the Alexandria area some areas around Fayoumand Wadi El-Natrun and small patches near El-Arish and MersaMatruh (Figs S2eS3 show a map of Egypt overlaid withgeographical locations of areas mentioned in this study and thelocations of currently established Protected Areas respectively)

22 Environmental predictor variables

Climate data for the near past (1950e2000) were downloadedfrom WorldClim Global Climate Data v14 (release 3 e see httpwwwworldclimorgbioclim) (Hijmans et al 2005) These data

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 213

represent current climate conditions and comprise 19 lsquobioclimaticrsquovariables derived from precipitation and temperature records(Table 1) Four (Bio-7 -14 -17 amp -18) were rejected a priori as notsufficiently variable across Egypt reflecting the countrys status asthe driest country in the world (FAO 2012) A resolution of 25 arc-minutes (~5 5 km) best matched the level of positional uncer-tainty that accompanies the records (most of which derive frommuseum records georeferenced by site name) and because theclimate data for Egypt were interpolated using relatively fewweather stations largely concentrated in the Nile Valley and Delta(see Fig 23 in Newbold 2009) This relatively coarse resolution isalso useful tominimise the effect of ignoring species interactions onmodelled distributions (Pearson and Dawson 2003)

Elevation data were obtained from the SRTM Digital ElevationDatabase v41 [available at httpwwwcgiar-csiorgdataelevation] The tiles that cover Egypt were downloaded at 90-mresolution and rescaled to 25 arc minutes We also used an Egyp-tian habitat map from the BioMAP project (A A Hassan unpub-lished data) In this map the habitat of Egypt is classified into 11classes (sea littoral coastal land cultivated land sand dune wadimetamorphic rock igneous rock gravels serir sand sheets sabkhasand sedimentary rocks e see Newbold et al 2009a) NormalizedDifference Vegetation Index (NDVI) data for seven years (from Jan2004 to Dec 2010) were downloaded from the SPOT Vegetationwebsite (see httpfreevgtvitobe) Two layers were derived fromthese maps and used as predictors in the models maximum NDVIvalue (indicating how much vegetation there is per pixel) and thedifference between maximum and minimum NDVI value (indi-cating variability in vegetation per pixel)

To reduce colinearity among predictors and over-fitting theVariance Inflation Factor (VIF) was calculated for the continuouspredictors using R v2150 (R Development Core Team 2012) andused to prune predictors before use (eg Bombi et al 2012) until allremaining variables yielded VIF values below 10 (Table 1) thesewere then used in the Maxent modelling

23 Climate models used

To estimate the potential impact of climate change currentdistribution models were projected into the future for three timeslices (2020s 2050s and 2080s each the median of a 30-yearwindow) Future climate data (IPCC 2007 downloaded fromhttpwwwccafs-climateorg) derived from four independentGlobal Circulation Models (GCMs) were used to obtain consensus

Table 1List of variables used to calculate VIF values rows shaded wiso used to run the models Four (Bio-7 -14 -17 amp -18) were

Altitude AltitudeNDVI_Max NDVI maximNDVI_Difference Absolute diffBio1 Annual meanBio2 Mean diurna

Bio3 IsothermalityBio4 TemperatureBio5 Maximum te

Bio6 Minimum teBio8 Mean tempeBio9 Mean tempe

Bio10 Mean tempeBio11 Mean tempeBio12 Annual preci

Bio13 PrecipitationBio15 Precipitation

Bio16 PrecipitationBio19 Precipitation

estimations about the fate of the herpetofauna These GCMs werethe Hadley Centre Coupled Model version 3 (HadCM3) the Cana-dian second-generation coupled global climate model (CGCM2-CCCma) the Australian model (CSIRO Mk2) and a Japanese model(NIES99) all frequently used in similar studies (eg Arauacutejo et al2006 Holt et al 2009) Our results assume that species try totrack their suitable habitat (dependent on dispersal ability) ratherthan accommodate or adapt to the new conditions

For each GCMmodel and time slice two emission scenarios (A2aand B2a) were used the most commonly used scenarios in climate-change assessments (Hannah 2011) They reflect two differentassumptions about the levels of CO2 emissions linked to assumeddemographic changes and socio-economic and technological de-velopments (Marini et al 2009) The A2a scenario assumes a lot ofclimate change caused by medium to high CO2 emission rates (theldquobusiness as usualrdquo scenario) while the B2a scenario assumesrelatively little change because of mitigation measures (theldquomoderaterdquo scenario) (Sauer et al 2011 Taubmann et al 2011)There are no data on how NDVI and habitat variables might changein Egypt in the future and so we assume they (and altitude) remainconstant Apart from the A2a scenario for 2080 nearly all the ex-pected climatic conditions are similar to current values (see Fig S4)

24 Species distribution modelling

We used Maxent v333k (Phillips et al 2006 e see httpwwwcsprincetonedu~schapiremaxent) to run the species dis-tribution models Maxent performs well in comparison withalternative algorithms and shows equal or higher predictive accu-racy on cross-validation (Elith et al 2006 Hernandez et al 2006)Crucial to our decision not to use ensemble modelling Maxent isvery robust to small numbers of training records outperformsother algorithms when using a relatively small number of occur-rence records (Franklin 2009) and is robust against moderatelocation errors (Graham et al 2008) We used lsquopermutationimportancersquo (see Phillips and ATampT Research 2011) to detect whichpredictor had the greatest influence on the model because it doesnot depend on the path used by the algorithm (Songer et al 2012)

Ten replicated runs with cross-validation and default settings(except iterations frac14 1000) were made for each species by itera-tively partitioning the records to use 90 for training and 10 fortesting (as recommended by Phillips and ATampT Research 2011)this gives a more stable model performance (avoiding overfitting)and minimises the effects of errors and biases The default logistic

th grey show variables with VIF values less than 10 andrejected a priori (see text)

um valueerence between the highest and lowest NDVI valuestemperature

l range (mean of monthly (max temp min temp))

(Bio2Bio7) (100)seasonality (standard deviation100)mperature of the warmest month

mperature of the coldest monthrature of the wettest quarterrature of the driest quarter

rature of the warmest quarterrature of the coldest quarterpitation

of the wettest monthseasonality (coefficient of variation)

of the wettest quarterof the coldest quarter

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221214

output format was chosen ie related to probability of suitableconditions ranging from 0 to 1 Logistic output has been criticizedrecently (eg Merow et al 2013) becausemany users assume that itrepresents the probability of occurrence to do that however thecalculation must assume that overall prevalence is fixed at 05 forall species which may not be accurate estimating true prevalenceis very difficult (Elith et al 2011) Both logistic and raw outputs aremonotonically related and rank locations identically (Elith et al2011) making rank-based measures such as AUC (for model eval-uation Elith et al 2011) and thresholding (using values frompredictions of the training presence locations eg 10th-percentilethreshold) identical whichever is used We used logistic output fortwomain reasons a) it standardises the expected habitat suitabilityvalues across species permitting the adding together of species tocreate the lsquoprobability richnessrsquo maps b) it allows Maxent auto-matically to convert the probabilities into expected presenceabsence (or more precisely suitablenon-suitable) habitat For thelatter we used the 10th-percentile training presence threshold(following Pearson et al 2007) For each species (and for each ofthe logistic and thresholded distributions) the ten replicate dis-tribution maps were converted to a single consensus map byMaxent for logistic values and manually for thresholded values(where each pixel was a consensus presence if it had presencevalues in more than half of the model runs ie gt5)

Distributions were then projected into three future time slices(2020 2050 and 2080) in each of four GCM models (HadCM3CCCma CSIRO and NIES99) and two emission scenarios (A2a andB2a) giving a total of 24 future projections Thresholded futuredistributions involved two extreme assumptions about dispersalability unlimited and no dispersal For each of the 24 projectionsthree consensus maps were produced one using the probability(logistic) output and two using the thresholded distributions(assuming unlimited and no-dispersal) The maps were then aver-aged across the different GCMs for thresholded distributions thisinvolved assigning a pixel as potentially suitable if it had a value ofpresence in more than two GCMs Thus there were 18 (three timeslices two scenarios three consensus) maps per species

25 Comparisons across species

Current and future maps of estimated species richness werecalculated in two different ways first by simply adding togetherthe average probability of suitable conditions for each pixel acrossthe distribution maps of all species (this assumes unlimiteddispersal eg Gilbert and Zalat 2008) and second by addingtogether the thresholded distribution maps (under unlimited- andno-dispersal assumptions eg Trotta-Moreu and Lobo 2010) Mapsof future species richness were obtained for each of the 24 futureprojections and then averaged across the four GCMs Future po-tential changes in species richness (but not composition) werecalculated by subtracting current species richness at each pixel

At the species level gains and losses in future distribution werecalculated using current and future consensus thresholded distri-butions Then the losses across all the 24 future potential distri-butions across all species were added together into a species lossmap showing which areas are expected to suffer most in losingspecies (for each dispersal assumption) The same was carried outfor gains in suitability to produce a species gain map showingwhich areas are expected to gain species (only for the unlimiteddispersal assumption) Species turnover an index of dissimilaritybetween current and future species composition (Thuiller 2004)was calculated for each future projection and dispersal assumption(following Thuiller et al 2005)

In order to assess future extinction risk and to determine whichspecies may require more protection in the future species range

changes were calculated as the percentage loss (or gain) in suitablehabitats carried out for each dispersal assumption Each specieswas classified into one of the following categories Extinct (ex-pected loss of the entire suitable habitat - 100) Critically Endan-gered (loss gt80) Endangered (loss 50e80) Vulnerable (loss30e50) Least Concern (loss lt30) gain 1 (gain lt30) gain 2(gain 30e50) gain 3 (gain 50e80) gain 4 (gain 80e100) andgain 5 (gain gt100) (modified from Thuiller et al 2005) (seeTables S2eS3)

26 Area prioritization for conservation

Several algorithms are now available to prioritize areas forconservation and conservation planning Using the SDM pre-dictions Zonation v31 (Moilanen et al 2012) created a nestedspatial conservation prioritization network to evaluate the effec-tiveness and performance of current Protected Area network inEgypt and identify new sites for conservation Zonation prioritizesthe conservation value of the landscape hierarchically based on theconservation value of sites (cells) (Moilanen et al 2012) by itera-tively removing the least valuable cells from the edges of thelandscape according to set rules The last to be removed are themost important areas (Moilanen et al 2012) We used two removalrules lsquobasic core-area functionrsquo to identify important (or poor)locations where a single or a few species have important occur-rences and lsquoadditive benefit functionrsquo to give more weight to lo-cations with high species richness (Moilanen et al 2012)Fragmentation of the chosen areas is minimized by an aggregationrule and we chose to use lsquodistribution smoothingrsquo for which anestimate of dispersal distance is required The dispersal ability ofalmost all terrestrial reptile species is very limited (Cadby et al2010 Edgar et al 2010) To be conservative in this study thedispersal distance of all reptiles was set to be equal to 1 km

Because some species are more important to conserve thanothers each species was given a weight Weights were developedfrom the global and national IUCN assessments the world distri-bution and distribution patterns within Egypt Following Gilbertand Zalat (2008) and Basuony et al (2010) we classified Egyptianreptiles according to their world distribution and their distributionpattern within Egypt We assessed their IUCN Red List status inEgypt according to IUCN guideline and categories (IUCN Standardsand Petitions Subcommittee 2010) although inevitably incom-plete We gave each element a score (modified from Leach et al2013) to indicate relative importance and then the sum of thescores gave the relative conservation weight of each species (seeTables S1 and S4) We used these values to weight the Zonationprioritization process

For current and future climates Zonation was run using themaps of habitat suitabilities for all species The number of cellsremoved at each iteration was set to 10 (frac14lsquowarp factorrsquo) Zonationproduces ASCII raster files with ranked pixel values scaled to befrom 0 to 1 values gt 07 were considered to be of high conservationimportance These important areas were then overlaid with thecurrent Protected Area network in Egypt to show if the network inEgypt is adequate to conserve the species in the face of climatechange and if there are areas outside the current Protected Areasboundaries that require special protection measures

3 Results

31 Model performance amp variable importance

In terms of mean AUC values for the 10 cross-validation runs theperformance of current-distribution models was good (seeTable S1) ranging from 078 to 099 with an overall mean of

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 215

093 plusmn 005 Only one species (Ptyodactylus siphonorhina) had amean AUC of less than 08 while in 36 species it was greater than095 As almost all species had a mean AUC gt08 all models wereaccepted and processed for further analyses More widespreadspecies tended to have lower AUC values (correlations [n frac14 75]with number of suitable pixels rs frac14 085 p lt 0005 extent ofoccurrence rs frac14 077 p lt 0005 number of pixels with recordsrs frac14 0433 p lt 0005)

Variables with highest mean permutation importance across allmodelled species were altitude (highest for 34 species) Bio4(temperature seasonality 10 species) and Bio13 (precipitation ofwettest month 9 species) Those with the lowest mean permuta-tion importance and thus not influencing the final models verymuch were the difference betweenmaximum andminimumNDVIBio15 (precipitation seasonality) habitat and Bio9 (mean temper-ature of driest quarter) the first and last of these were never themost influential variable for any species

32 Species richness

The patterns of current species richness (Fig 1) were similarusing either logistic or thresholded distributions both emphasizingthe lower Nile Valley the Suez Canal area and the coasts of theMediterranean and Suez Gulf (Fig S2 for locations map) The

Fig 1 Current estimated reptile species richness of Egypt using the sum of (a)Maxent logistic predictions and (b) thresholded predicted distributions

pattern is concordant with the species richness map in Baha El Din(2006) produced using records and qualitative criteria (seeFig S5a) Under climate change regardless of emission scenario ordispersal assumptions several areas are expected to have increasedspecies richness (Fig 2) with generally greater magnitudes underthe A2a scenario while others are expected to decline (eg the SuezCanal area) with greater declines under the B2a scenario by 2050and 2080

Areas with the highest expected loss of species include betweengreater Cairo Ismailia and Suez south of Suez on both sides of theSuez Gulf and Wadi El-Natrun much greater under the B2a sce-nario (Fig 3a) If species can disperse well then the highest gains inspecies are expected to include a large proportion of the northernhalf of the Western Desert again greater under the B2a scenario(Fig 3b) In terms of percentage change of the reptile assemblage ifdispersal is unlimited then the Western Desert is expected to un-dergo high turnover (Fig S6a) but this is because there are so fewspecies there currently Under no dispersal turnover is very limited(Fig S6b) again with highest values again in the Western desertbecause of the species poor assemblages there

33 Range and status changes

With unlimited dispersal no species is expected to becomeldquoextinctrdquo (100 loss of suitable habitat) under all or the average ofall GCMs There are a couple of species expected to become extinctby losing their entire area of suitable habitat in at least one of thefuture projections Tarentola mindiae and Hemidactylus robustusUsing the average gain or loss of suitable habitat across the fourdifferent GCMs eight species are expected to be classified as Crit-ically Endangered (loss gt 80) twenty as Endangered (loss50e80) and twenty-two as Vulnerable (loss 30e50) in at leastone mean future projection (details see Table S2)

When assuming no dispersal at all none of the species are ex-pected to lose all suitable habitats under all or the average of allGCMs but T mindiae and H robustus again experience completeloss of habitat under at least one of the projections Using theaverage loss of suitable habitat across the four different GCMs tenspecies are expected to be classified as Critically Endangeredtwenty-seven as Endangered and thirty-three as Vulnerable in atleast one mean future projection (details see Table S3)

34 Area prioritization for conservation

The areas with the highest current prioritization value for con-servation were similar for both cell-removal methods mainlyemphasizing the north coast Suez Canal area South Sinai theGebel Elba region Qattara Depression Wadi El-Natrun and aroundCairo amp Fayoum with the lsquoadditive benefitrsquo function putting greateremphasis on the Nile Delta and its fringing areas (Fig 4) Thisoverall pattern does not change much under future climate pro-jections though a few areas decline in priority and much greaterarea of the northern Western Desert increases These effects aremore pronounced with the lsquocore-area functionrsquo used as the cellremoval rule

The mean prioritization value in all models was higher in Pro-tected Areas than outside them (Fig 5) with both slightlyincreasing in future projections The difference in prioritizationvalue between inside and outside Protected Areas declined withtime especially by 2050 and 2080 under the B2a scenario

4 Discussion

The results of this study show for first time the potential impactsof climate change on the distribution of Egyptian reptiles Some

Fig 2 Estimated future changes in species richness as a result of anthropogenic climate change under two IPCC scenarios (A2a B2a) and at three times in the future (2020 20502080) Each map is the difference from the current species richness (shown in Fig 1a) and is averaged across four GCMmodels Maps made (a) using Maxent logistic output and (b)using thresholded distributions (see Methods)

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221216

reptiles are expected to lose much of their currently suitabilityareas (up to 80 in some cases) and some areas will lose or changea large proportion of their species Thus there will be a need forgreater conservation efforts in the future Some of these areas arealready inside Protected Areas and so hypothetically they are pro-tected while others are either just outside or very far from Pro-tected Areas Although overall Protected Areas contain better(higher priority) values compared to unprotected sites the currentProtected-Area network appears inadequate for future conserva-tion Our results together with others (eg Leach et al 2013) willhelp the Egyptian Environmental Affairs Agency and otherdecision-making parties in Egypt to direct conservation efforts andlimited conservation resources in the right direction

41 Model performance amp variable importance

The negative correlations found between AUC and the areaoccupied and with the extent of occurrence concur with findingsof many other studies (eg Brotons et al 2004 Elith et al 2006Hernandez et al 2006) Rare species are usually habitat special-ists with narrow environmental tolerances and are environmen-tally or geographically restricted compared to widespread species

(Elith et al 2006 Pandit et al 2009) Widespread species are likelyto be generalists occupying a wide range of habitats and climatesmaking it difficult to distinguish between suitable and unsuitablehabitats (Franklin et al 2009) However as pointed out by Elithet al (2006) this may be an artefact an inevitable result for spe-cies with small relative occurrence areas and perhaps a reason notto use AUC to compare model performances (Lobo et al 2008)Most (about 60) of the Egyptian herpetofauna are narrowlydistributed occupying less than 10 of Egypts area (Baha El Din2006) There is no evidence of a strong correlation betweenmodel accuracy and the number of records used (see Elith et al2006 Newbold et al 2009b) and the weak negative correlationfound here is unusual since most (eg Stockwell and Peterson2002 Hernandez et al 2006) find that models with larger sam-ple sizes have higher accuracy (but see de Pous et al 2011)

We found altitude to be the most effective predictor for manyspecies In contrast the NDVI predictors were not useful unlike inmany other cases (eg Egbert et al 2002 Anderson et al 2006Kgosiesele 2010 Taheri 2010) but perhaps typical for arid envi-ronments (eg Soultans (2011) study on Egyptian antelopes) or forherpetofauna (eg Teodoro et al 2013) Habitat categories likewisemade only a low contribution to the reptile models

Fig 3 Expected patterns of cumulative (a) losses and (b) gains of species as a result of climate change The losses of each species in each pixel (under both dispersal assumptions)and gains (under unlimited dispersal only) were averaged across GCM models accumulated across species and then plotted

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 217

42 Species richness

Species richness by itself is not adequate as an indicator of bioticchange since species composition can change independently Ourresults suggest that in the future some areas with currently highexpected species richness may lose many species or undergo largechanges in species composition Some of these areas are alreadyunder a certain degree of protection but even if they can bemaintained or enhanced (eg Siwa oasis Gebel Elba and the coastalareas over the Aqaba gulf Fig S2) current Protected Areas are notsufficient to prevent expected loss of species Other areas (eg WadiEl-Natrun the Suez Canal area Red Sea inland wadis the poorlyknown Qattara Depression and the areas around Gebel El-Gallalaand Gebel El-Hallal) are unprotected and measures are needed toconserve them There are similar situations elsewhere for examplehigh species loss and turnover in the reptiles of a priority conser-vation hotspot in Europe (the Iberian Peninsula) are expected (dePous 2011) with great variation in expected patterns of loss andchange with different projections and dispersal assumptions Otherstudies have also concluded that large changes are likely in chelo-nians worldwide (Ihlow et al 2012) Mediterranean mammals(Maiorano et al 2011) US trees and vertebrates (Currie 2001) andMesoAmerican trees (Golicher et al 2012)

43 Range and status changes

Two species are expected to lose their suitable habitats entirelyin at least one future projection 1) T mindiae is a near-endemicspecies recorded just from northwest Egypt and northern Cyrena-ica (eastern Libya) its distribution in Egypt is restricted to Siwaoasis the Qattara Depression and their periphery (Baha El Din2006) Future range changes suggest it will be Endangered (loss50e80) by 2020 and as Critically Endangered (loss gt80) by 2050and 2080 (under all emission scenarios and dispersal assumptions)2) H robustus is localized to the East African coast from Zanzibar tosouthern Egypt Arabia east to Pakistan its distribution in Egypt ison the Red Sea coast from El-Quseir southwards (Baha El Din2006) The A2a scenario suggests it will be Critically Endangeredby 2020 and 2050 while under the B2a scenario it will be Endan-gered by 2020 and Critically Endangered by 2050 and 2080 Morethan half of its current distribution in Egypt is located within theWadi El-Gemal and Gebel Elba Protected Areas so hypothetically itis protected However it is found mainly on buildings and henceassociated with man it may in fact be an invasive to Egypt

Another eight species are expected to be classified as CriticallyEndangered (loss gt80) in at least one averaged future projection(Acanthodactylus longipes Cerastes vipera Eryx jaculus Eumeces

Fig 4 Current conservation prioritizationl of areas according to the Zonation algo-rithm under two cell removal rules (a) core-area function (identifies areas where oneor more species have important occurrences) and (b) additive benefit function (whichgives greater emphasis to areas with high species richness) Darker colours representhigher conservation value (For interpretation of the references to colour in this figurelegend the reader is referred to the web version of this article)

Fig 5 Mean prioritization value for conservation (plusmn95 confidence limits) derivedfrom Zonation for Protected Areas (PAs) and outside Protected Areas (non-PAs) forcurrent and expected future conditions using the cell-removal rule of (a) core-areafunction and (b) additive benefit function

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221218

schneiderii Malpolon moilensis Ptyodactylus guttatusP siphonorhina and Trapelus mutabilis) According to the currentassessment twenty-seven species are classified as Endangered(loss 50e80) in at least one projection (mostly by 2080 in the A2scenario) two of these are classified assuming no dispersal asEndangered in all future projections (Uromastyx aegyptia and Pris-turus flavipunctatus for details see Tables S2eS3)

Reptiles are also likely to gain from global warming especially inmore northern areas of the globe if they can disperse and expandtheir distributions but under perhaps the more realistic assump-tion of no dispersal most species were expected to lose habitatsuitability by 2050 (Arauacutejo et al 2006) In Egypt areas of clearfuture expected gain (assuming unlimited dispersal) are the areabetween the Siwa oasis and Qattara Depression northwards almostto the northern coastal strip and some sparse locations in centralSinai There were greater gains under the B2a scenario in theeastern part of the Western Desert and some patchy areas aroundthe Nile (Fig 3b) This pattern is similar to that reported by Araujoet al (2006)

Some of the species identified here as being most at risk due tofuture climate change are in fact resilient species that are ecolog-ically very adaptable and are currently experiencing range expan-sion and population growth at least in Egypt (eg H robustus) This

disparity is probably because modelling is based solely on theprobable impacts of climate change on the spatial distribution ofsuitable habitats without taking evolution or ecological adapt-ability and resilience into account Species vary greatly in theirresponse to ecological change some adapting very well even tovery rapid anthropogenically induced changes (eg invasive spe-cies) while others are much less adaptable and tend to go extinctfirst

There are no published studies that discuss the effect of climatechange on Egyptian reptiles obviating any comparisons Informa-tion on the reptiles of the adjacent countries of Sudan and Libya arevery limited making it impossible to assess any movements thatmight compensate for Egyptian losses The patterns of expectedchanges in Leach et als (2013) similar study on mammals andbutterflies do not concur with ours Mammal species richness wasexpected to decline in the Mediterranean and Red Sea areas (by40e60) and increase elsewhere (by 80e100) while for butter-flies declines are expected over almost the whole of Egypt (by40e60) except the south which is expected to increase (by40e60) In the models of European reptiles (de Pous 2011)

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 219

between eight (unlimited dispersal) and 21 (no dispersal) specieswere expected to become extinct in at least one future projectionwith two more losing more than 80 of their currently suitablehabitats This is a much more damaging scenario than the expectedresponses of Egyptian reptiles in our study There are no othercomparable studies from North Africa or the Middle East

44 Area prioritization for conservation

In our study Protected Areas (Fig S3) were more valuable inconserving reptiles than areas outside them both currently and inthe future although the difference is expected to get smaller withtime (Fig 5) Although this is encouraging the Protected Areanetwork appears inadequate to conserve Egyptian reptiles NewProtected Areas that have already been proposed would go someway towards reducing this deficiency in coverage although otherswill be needed in the middle to north Sinai (especially aroundGebel El-Hallal) the Suez Canal area both sides of the Suez Gulf thenorthern Mediterranean coast coastal wadis of the Red Sea be-tween Hurghada and Mersa Alam and the Gebel El-Gallala areaMore effective protection is required in the established ProtectedAreas including those in South Sinai Gebel Elba and Siwa oasis dueto their relatively high expected future suitability loss The Pro-tected Areas of Siwa oasis and El-Gilf El-Kebir plus the areas ofWadiEl-Natrun the Gebel El-Hallal area and wadis of the Red Sea areexpected to have particularly high turnover in species composition(Fig S6) and hence probably need more conservation efforts Somecomparable information is available for Egypts butterflies andmammals (Gilbert and Zalat 2008 Basuony et al 2010 Leach et al2013) Egypts network of Protected Areas appears adequate toconserve Egyptian butterfly hotspots except for the Mediterraneancoast between Alexandria and Sallum (Gilbert and Zalat 2008) butinadequate to conserve important mammal areas with new Pro-tected Areas needed to be constructed in the lower Nile Valleyalong the north coast between Alexandria and Sallum top part ofthe Suez Gulf and perhaps the Qattara Depression (Basuony et al2010) Suggestions for conserving mammals are consistent to someextent with our results which also suggest new Protected Areasalong the Suez Gulf Mediterranean coast from Mersa Matruh toSallum and the Qattara Depression Leach et al (2013) concur withour results in suggesting priority areas along the western Medi-terranean coast around Greater Cairo central and north Sinai andthe Red Sea coast

45 Limitations

There are some limitations of the results shown in this study 1)Egyptian climate data in WorldClim are interpolated from ratherfew non-randomly placed weather stations in Egypt and adjacentcountries Although making the climate data less reliable thanelsewhere there are no alternative climate data available 2) Anensemble approach (eg using BIOMOD Thuiller et al 2009) isoften suggested for SDM modelling to avoid dependence on oneparticular algorithm This is not possible without a greater numberof records since other algorithms are not as flexible or robust asMaxent Despite all the caveats with using SDMs and criticisms ofMaxent in particular (Royle et al 2012 but see Hastie and Fithian2013 Elith et al 2006 Hernandez et al 2006) in practice coun-tries such as Egypt have little else to guide them in attempts atclimate change adaptation 3) There were signs of sampling bias inthe available records Although this violates the assumptions of allSDMs it is a problem common to nearly all datasets worldwideNew techniques may allow for the correction of sampling bias egusing target-group approach (Phillips et al 2009) although thedata requirements are currently impossible for Egypt and most

other developing countries) 4) Using Maxents default settings iscriticized by some authors they need to be modified dependent onthe data (Warren and Seifert 2011 Radosavljevic and Anderson2014 Merow et al 2013) but we think using the default settingsis enough in this study

5 Conclusion

Although Egyptian datasets exist knowledge of Egyptian spe-cies and conservation priorities are limited (with consequencesexplored in Leach et al 2013) nevertheless our results and those ofLeach et al (2013) encourage us to believe that useful insights canbe gained for conservation priorities by using the combination ofspecies distribution modelling plus spatial conservation prioriti-zation techniques (Hannah 2010 Klorvuttimontara et al 2011) Itis crucial for the Egyptian Environmental Affairs Agency to establishan accurate and reliable database of records assessed by experts tohelp improve the basic information for future conservation as-sessments and analyses

Although many SDM studies have been done on a variety oforganisms in recent decades aiming to solve conservation prob-lems evidence of their practical utility in real-world conservation issparse with very few studies clearly targeting conservation de-cisions (Guisan et al 2013) If implemented correctly SDMs canplay a critical role in supporting spatial conservation decision-making (Addison et al 2013) For reptiles there is a shortage ofstudies estimating their response to expected climate changes(Arauacutejo et al 2006) Although some suggest that reptiles (and to agreater extent amphibians) will be strongly affected by climatechange (eg Gibbons et al 2000 Arauacutejo et al 2006 Carvalho et al2010) very few of themwere concerned with arid environments orthe Middle East We hope that this and other similar studies thatestimate the fate of reptiles under climate change help decision-making authorities to take the measures necessary to conservethem

Acknowledgements

We thank the Italian Cooperation (Debt Swap) in Egypt forfunding the BioMAP project that allowed the collation of the da-tabases used in this work Dr Tim Newbold for allowing the use ofthe data from his field trips Dr Mustafa Fouda (Former Director ofthe Nature Conservation Sector EEAA) for facilities offered duringBioMAP project and all the BioMAP staff This work was supportedby the award of a Chevening Scholarship Masters grant to AEl-GWe are especially grateful to two anonymous referees whogreatly improved an earlier manuscript

Appendix A Supplementary data

Supplementary data related to this article can be found at httpdxdoiorg101016jjaridenv201512007

References

Addison PFE Rumpff L Bau SS Carey JM Chee YE Jarrad FC McBride MFBurgman MA 2013 Practical solutions for making models indispensable inconservation decision-making Divers Distrib 19 (5e6) 490e502 httpdxdoiorg101111ddi12054

Anderson RP Peterson AT Egbert SL 2006 Vegetation-index models predictareas vulnerable to purple loosestrife (Lythrum salicaria) invasion in KansasSouthwest Nat 51 471e480 httpdxdoiorg1018940038-4909(2006)51[471VMPAVT]20CO2

Arauacutejo MB 2009 Climate change and spatial conservation planning InMoilanen A Wilson KA Possingham HP (Eds) Spatial Conservation Prior-itization Quantitative Methods and Computational Tools Oxford UniversityPress pp 172e184 ISBN 9780199547760

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221220

Arauacutejo MB Rahbek C 2006 How does climate change affect biodiversity Sci-ence 313 1396e1397 httpdxdoiorg101126science1131758

Arauacutejo MB Thuiller W Pearson RG 2006 Climate warming and the decline ofamphibians and reptiles in Europe J Biogeogr 33 1712e1728 httpdxdoiorg101111j1365-2699200601482x

Baha El Din SM 2001 The Herpetofauna of Egypt Species Communities andAssemblages University of Nottingham p 435 Unpublished PhD thesis

Baha El Din SM 2006 A Guide to the Reptiles and Amphibians of Egypt TheAmerican University in Cairo Press ISBN 9789774249792

Baha El Din SM 2007 A new lizard of the Acanthodactylus scutellatus group(Squamata Lacertidae) from Egypt Zool Middle East 40 21e32 httpdxdoiorg10108009397140200710638200

Baldwin RA 2009 Use of maximum entropy modeling in wildlife research En-tropy 11 854e866 httpdxdoiorg103390e11040854

Ball IR Possingham HP Watts M 2009 Marxan and relatives software forspatial conservation prioritisation In Moilanen A Wilson KAPossingham HP (Eds) Spatial Conservation Prioritization QuantitativeMethods and Computational Tools Oxford University Press pp 185e195 ISBN9780199547760

Basuony MI Gilbert F Zalat S 2010 The Mammals of Egypt Atlas Red DataListing amp Conservation BioMAP amp CultNat EEAA amp Bibliotheca AlexandrinaCairo available at httpecologynottinghamacuk~plzfg

Bombi P Salvi D Bologna MA Pettorelli N 2012 Cross-scale predictions allowthe identification of local conservation priorities from atlas data Anim Conserv15 378e387 httpdxdoiorg101111j1469-1795201200526x

Brotons L Thuiller W Arauacutejo MB Hirzel AH 2004 Presence-absence versuspresence-only modelling methods for predicting bird habitat suitability Ecog-raphy 27 437e448 httpdxdoiorg101111j0906-7590200403764x

Cadby CD While GM Hobday AJ Uller T Wapstra E 2010 Multi-scaleapproach to understanding climate effects on offspring size at birth and date ofbirth in a reptile Integr Zool 5 164e175 httpdxdoiorg101111j1749-4877201000201x

Carvalho SB Brito JC Crespo EJ Possingham HP 2010 From climate changepredictions to actions e conserving vulnerable animal groups in hotspots at aregional scale Glob Change Biol 16 (12) 3257e3270 httpdxdoiorg101111j1365-2486201002212x

Currie DJ 2001 Projected effects of climate change on patterns of vertebrate andtree species richness in the conterminous United States Ecosystems 4216e225 httpdxdoiorg101007s10021-001-0005-4

de Pous P 2011 Climate Change and European Reptiles from Large-scale SpatialRange Shift Projections towards a More Realistic Assessment of Species Re-sponses University of East Anglia UK p 64 MSc thesis

de Pous P Beukema W Weterings M Duumlmmer I Geniez P 2011 Area priori-tization and performance evaluation of the conservation area network for theMoroccan herpetofauna a preliminary assessment Biodivers Conserv 2089e118 httpdxdoiorg101007s10531-010-9948-0

Edgar P Foster J Baker J 2010 Reptile Habitat Management HandbookAmphibian and Reptile Conservation Bournemouth

Egbert SL Martiacutenez-Meyer E Ortega-Huerta M Peterson AT 2002 Use ofdatasets derived from time-series AVHRR imagery as surrogates for land covermaps in predicting species distributions In Proceedings IEEE 2002 Interna-tional Geoscience and Remote Sensing Symposium (IGARSS) vol 4pp 2337e2339 httpdxdoiorg101109IGARSS20021026537

El Alqamy H Ismael A Abdelhameed A Nagy A Hamada A et al 2010 Pre-dicting the status and distribution of the Nubian Ibex (Capra nubiana) in thehigh-altitude mountains of south Sinai (Egypt) In Galemys Newsletter of theSpanish Society for the Preservation and Study of Mammals vol 22pp 517e530

Elith J Graham CH Anderson RP Dudiacutek M Ferrier S et al 2006 Novelmethods improve prediction of species distributions from occurrence dataEcography 29 129e151 httpdxdoiorg101111j20060906-759004596x

Elith J Phillips SJ Hastie T Dudiacutek M Chee YE Yates CJ 2011 A statisticalexplanation of MaxEnt for ecologists Divers Distrib 17 (1) 43e57 httpdxdoiorg101111j1472-4642201000725x

FAO 2012 Food and Agriculture Organisation World Development IndicatorsAverage Precipitation in Depth (mm per year) httpdataworldbankorgindicatorAGLNDPRCPMM

Franklin J 2009 Mapping Species Distributions Spatial Inference and PredictionCambridge University Press Cambridge UK ISBN 9780521700023

Franklin J Wejnert KE Hathaway SA Rochester CJ Fisher RN 2009 Effect ofspecies rarity on the accuracy of species distribution models for reptiles andamphibians in southern California Divers Distrib 15 167e177 httpdxdoiorg101111j1472-4642200800536x

Gibbons JW Scott DE Ryan TJ Buhlmann KA Tuberville TD Metts BSGreene JL Mills T et al 2000 The global decline of reptiles Deja Vu Am-phibians BioScience 50 (8) 653e666 httpdxdoiorg1016410006-3568(2000)050[0653TGDORD]20CO2

Gilbert F Zalat S 2008 The Butterflies of Egypt Atlas Red Data Listing amp Con-servation BioMAP EEAA Cairo available at httpecologynottinghamacuk~plzfg

Golicher DJ Cayuela L Newton AC 2012 Effects of climate change on the po-tential species richness of Mesoamerican forests Biotropica 44 284e293httpdxdoiorg101111j1744-7429201100815x

Graham CH Elith J Hijmans RJ Guisan A Peterson AT et al 2008 Theinfluence of spatial errors in species occurrence data used in distribution

models J Appl Ecol 45 239e247 httpdxdoiorg101111j1365-2664200701408x

Guisan A Tingley R Baumgartner JB Naujokaitis-Lewis I Sutcliffe PRTulloch AIT Regan TJ Brotons L et al 2013 Predicting species distributionsfor conservation decisions Ecol Lett 16 (12) 1424e1435 httpdxdoiorg101111ele12189

Hannah L 2010 A global conservation system for climate-change adaptationConserv Biol 24 70e77 httpdxdoiorg101111j1523-1739200901405x

Hannah L 2011 Climate Change Biology Elsevier Ltd ISBN 9780123741820Hannah L Lovejoy TE Schneider SH 2005 Biodiversity and climate change in

context In Lovejoy TE Hannah L (Eds) Climate Change and BiodiversityYale University Press New Haven CT USA

Hannah L Midgley G Andelman S Arauacutejo M Hughes G et al 2007 Protectedarea needs in a changing climate Front Ecol Environ 5 131e138 httpdxdoiorg1018901540-9295(2007)5[131PANIAC]20CO2

Hastie T Fithian W 2013 Inference from presence-only data the ongoing con-troversy Ecography 36 864e867 httpdxdoiorg101111j1600-0587201300321x

Hernandez PA Graham CH Master LL Albert DL 2006 The effect of samplesize and species characteristics on performance of different species distributionmodeling methods Ecography 29 773e785 httpdxdoiorg101111j0906-7590200604700x

Hijmans RJ Cameron SE Parra JL Jones PG Jarvis A 2005 Very high reso-lution interpolated climate surfaces for global land areas Int J Climatol 251965e1978 httpdxdoiorg101002Joc1276

Holt AC Salkeld DJ Fritz CL Tucker JR Gong P 2009 Spatial analysis ofplague in California niche modeling predictions of the current distribution andpotential response to climate change Int J Health Geogr 8 38 httpdxdoiorg1011861476-072X-8-38

Hoyle M James M 2005 Global warming human population pressure andviability of the worlds smallest butterfly Conserv Biol 19 1113e1124 httpdxdoiorg101111j1523-1739200500166x

Ihlow F Dambach J Engler JO Flecks M Hartmann T et al 2012 On the brinkof extinction How climate change may affect global chelonian species richnessand distribution Glob Change Biol 18 1520e1530 httpdxdoiorg101111j1365-2486201102623x

IPCC 2007 Climate Change 2007 Synthesis Report Contribution of WorkingGroups I II and III to the Fourth Assessment Report of the IntergovernmentalPanel on Climate Change IPCC Geneva Switzerland

Kgosiesele E 2010 Predictive Distribution Modelling of Timon Lepida in SpainUniversity of Twente (ITC) the Netherlands p 87 MSc thesis

Klorvuttimontara S McClean CJ Hill JK 2011 Evaluating the effectiveness ofprotected areas for conserving tropical forest butterflies of Thailand BiolConserv 144 2534e2540 httpdxdoiorg101016jbiocon201107012

Leach K Zalat S Gilbert F 2013 Egypts protected area network under futureclimate change Biol Conserv 159 490e500 httpdxdoiorg101016jbiocon201211025

Lobo JM Jimenez-Valverde A Real R 2008 AUC a misleading measure of theperformance of predictive distribution models Glob Ecol Biogeogr 17145e151 httpdxdoiorg101111j1466-8238200700358x

Maiorano L Falcucci A Zimmermann NE Psomas A Pottier J et al 2011 Thefuture of terrestrial mammals in the Mediterranean basin under climatechange Philosophical Trans R Soc Lond Ser B Biol Sci 366 2681e2692httpdxdoiorg101098rstb20110121

Marini MA Barbet-Massin M Lopes LE Jiguet F 2009 Predicted climate-driven bird distribution changes and forecasted conservation conflicts in aneotropical savanna Conserv Biol 23 1558e1567 httpdxdoiorg101111j1523-1739200901258x

Merow C Smith MJ Silander JA 2013 A practical guide to MaxEnt for modelingspecies distributions what it does and why inputs and settings matterEcography 36 (10) 1058e1069 httpdxdoiorg101111j1600-0587201307872x

Moilanen A Meller L Leppeuroanen J Pouzols FM Arponen A et al 2012 Zona-tion Spatial Conservation Planning Framework and Software v31 User Manualwwwhelsinkifibioscienceconsplan

Newbold T 2009 The Value of Species Distribution Models as a Tool for Conser-vation and Ecology in Egypt and Britain University of Nottingham UK p 312PhD thesis

Newbold T Gilbert F Zalat S El-Gabbas A Reader T 2009a Climate-basedmodels of spatial patterns of species richness in Egypts butterfly and mammalfauna J Biogeogr 36 2085e2095 httpdxdoiorg101111j1365-2699200902140x

Newbold T Reader T Zalat S El-Gabbas A Gilbert F 2009b Effect of charac-teristics of butterfly species on the accuracy of distribution models in an aridenvironment Biodivers Conserv 18 3629e3641 httpdxdoiorg101007s10531-009-9668-5

Pandit SN Kolasa J Cottenie K 2009 Contrasts between habitat generalists andspecialists an empirical extension to the basic metacommunity frameworkEcology 90 2253e2262 httpdxdoiorg10189008-08511

Pearson RG Dawson TP 2003 Predicting the impacts of climate change on thedistribution of species are bioclimate envelope models useful Glob EcolBiogeogr 12 361e371 httpdxdoiorg101046j1466-822X200300042x

Pearson RG Raxworthy CJ Nakamura M Peterson AT 2007 Predicting speciesdistributions from small numbers of occurrence records a test case using

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 221

cryptic geckos in Madagascar J Biogeogr 34 102e117 httpdxdoiorg101111j1365-2699200601594x

Phillips SJ ATampT Research 2011 A Brief Tutorial on Maxent Available at httpwwwcsprincetonedu~schapiremaxenttutorialtutorialdoc (Last accessed290512)

Phillips SJ Anderson RP Schapire RE 2006 Maximum entropy modeling ofspecies geographic distributions Ecol Model 190 231e259 httpdxdoiorg101016jecolmodel200503026

Phillips SJ Dudiacutek M Elith J Graham CH Lehmann A Leathwick J Ferrier S2009 Sample selection bias and presence-only distribution models implica-tions for background and pseudo-absence data Ecol Appl 19 181e197 httpdxdoiorg10189007-21531

R Development Core Team 2012 R a Language and Environment for StatisticalComputing R Foundation for Statistical Computing Vienna Austria ISBN 3-900051-07-0 httpwwwR-projectorg

Radosavljevic A Anderson RP 2014 Making better Maxent models of speciesdistributions complexity overfitting and evaluation J Biogeogr 41 (4)629e643 httpdxdoiorg101111jbi12227

Royle JA Chandler RB Yackulic C Nichols JD 2012 Likelihood analysis ofspecies occurrence probability from presence-only data for modelling speciesdistributions Methods Ecol Evol 3 545e554 httpdxdoiorg101111j2041-210X201100182x

Sauer J Domisch S Nowak C Haase P 2011 Lowmountain ranges summit trapsfor montane freshwater species under climate change Biodivers Conserv 203133e3146 httpdxdoiorg101007s10531-011-0140-y

Songer M Delion M Biggs A Huang Q 2012 Modeling impacts of climatechange on Giant Panda Habitat Int J Ecol 2012 1e12 httpdxdoiorg1011552012108752

Soultan AE 2011 Antelope distribution in Egypt Modeling the Impact of ClimateChange on Antelopes Distribution in Egypt Lambert Academic Publishing

Standards IUCN Subcommittee Petitions 2010 Guidelines for Using the IUCNRed List Categories and Criteria Version 81

Stockwell DRB Peterson AT 2002 Effects of sample size on accuracy of speciesdistribution models Ecol Model 148 1e13 httpdxdoiorg101016S0304-3800(01)00388-X

Suarez AV Tsutsui ND 2004 The value of museum collections for research andsociety Bioscience 54 66 httpdxdoiorg1016410006-3568(2004)054[0066tvomcf]20co2

Taheri S 2010 Hyper Temporal NDVI Images for Modelling and Prediction theHabitat Distribution of Balkan Green Lizard (Lacerta Trilineata) e Case Study

Crete (Greece) University of Twente (ITC) the Netherlands p 54 MScthesis

Taubmann J Theissinger K Feldheim KA Laube I Graf W et al 2011Modelling range shifts and assessing genetic diversity distribution of themontane aquatic mayfly Ameletus inopinatus in Europe under climate changescenarios Conserv Genet 12 503e515 httpdxdoiorg101007s10592-010-0157-x

Teodoro AC Sillero N Alves S Duarte L 2013 Correlation between the habitatsproductivity and species richness (amphibians and reptiles) in Portugal throughremote sensed data In Proc SPIE 8887 Remote Sensing for Agriculture Eco-systems and Hydrology httpdxdoiorg101117122028502

Thomas CD Cameron A Green RE Bakkenes M Beaumont LJ et al 2004Extinction risk from climate change Nature 427 145e148 httpdxdoiorg101038nature02121

Thuiller W 2004 Patterns and uncertainties of species range shifts under climatechange Glob Change Biol 10 2020e2027 httpdxdoiorg101111j1365-2486200400859x

Thuiller W 2007 Biodiversity climate change and the ecologist Nature 448550e552 httpdxdoiorg101038448550a

Thuiller W Lavorel S Arauacutejo MB Sykes MT Prentice IC 2005 Climate changethreats to plant diversity in Europe Proc Natl Acad Sci U S A 1028245e8250 httpdxdoiorg101073pnas0409902102

Thuiller W Lafourcade B Engler R Arauacutejo MB 2009 BIOMOD e a platform forensemble forecasting of species distributions Ecography 32 (3) 369e373httpdxdoiorg101111j1600-0587200805742x

Tolba MK Saab NW 2009 Arab environment Climate Change - Impact ofClimate Change on Arab Countries Arab Forum for Environment andDvelopment

Trotta-Moreu N Lobo JM 2010 Deriving the species richness distribution ofGeotrupinae (Coleoptera Scarabaeoidea) in Mexico from the overlap of indi-vidual model predictions Environ Entomol 39 42e49 httpdxdoiorg101603EN08179

Warren DL Seifert SN 2011 Ecological niche modeling in Maxent the impor-tance of model complexity and the performance of model selection criteriaEcol Appl 21 (2) 335e342 httpdxdoiorg10189010-11711

Wilson KA Cabeza M Klein CJ 2009 Fundamental concepts of spatial con-servation prioritisation In Moilanen A Wilson KA Possingham HP (Eds)Spatial Conservation Prioritization Quantitative Methods and ComputationalTools Oxford University Press pp 16e27 ISBN 9780199547760

Page 3: Journal of Arid Environments - elgabbas.github.io · e Faculty of Sciences & Arts, Taibah University, Al-Ula Campus, Saudi Arabia f Department of Zoology, Suez Canal University, Ismailia,

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 213

represent current climate conditions and comprise 19 lsquobioclimaticrsquovariables derived from precipitation and temperature records(Table 1) Four (Bio-7 -14 -17 amp -18) were rejected a priori as notsufficiently variable across Egypt reflecting the countrys status asthe driest country in the world (FAO 2012) A resolution of 25 arc-minutes (~5 5 km) best matched the level of positional uncer-tainty that accompanies the records (most of which derive frommuseum records georeferenced by site name) and because theclimate data for Egypt were interpolated using relatively fewweather stations largely concentrated in the Nile Valley and Delta(see Fig 23 in Newbold 2009) This relatively coarse resolution isalso useful tominimise the effect of ignoring species interactions onmodelled distributions (Pearson and Dawson 2003)

Elevation data were obtained from the SRTM Digital ElevationDatabase v41 [available at httpwwwcgiar-csiorgdataelevation] The tiles that cover Egypt were downloaded at 90-mresolution and rescaled to 25 arc minutes We also used an Egyp-tian habitat map from the BioMAP project (A A Hassan unpub-lished data) In this map the habitat of Egypt is classified into 11classes (sea littoral coastal land cultivated land sand dune wadimetamorphic rock igneous rock gravels serir sand sheets sabkhasand sedimentary rocks e see Newbold et al 2009a) NormalizedDifference Vegetation Index (NDVI) data for seven years (from Jan2004 to Dec 2010) were downloaded from the SPOT Vegetationwebsite (see httpfreevgtvitobe) Two layers were derived fromthese maps and used as predictors in the models maximum NDVIvalue (indicating how much vegetation there is per pixel) and thedifference between maximum and minimum NDVI value (indi-cating variability in vegetation per pixel)

To reduce colinearity among predictors and over-fitting theVariance Inflation Factor (VIF) was calculated for the continuouspredictors using R v2150 (R Development Core Team 2012) andused to prune predictors before use (eg Bombi et al 2012) until allremaining variables yielded VIF values below 10 (Table 1) thesewere then used in the Maxent modelling

23 Climate models used

To estimate the potential impact of climate change currentdistribution models were projected into the future for three timeslices (2020s 2050s and 2080s each the median of a 30-yearwindow) Future climate data (IPCC 2007 downloaded fromhttpwwwccafs-climateorg) derived from four independentGlobal Circulation Models (GCMs) were used to obtain consensus

Table 1List of variables used to calculate VIF values rows shaded wiso used to run the models Four (Bio-7 -14 -17 amp -18) were

Altitude AltitudeNDVI_Max NDVI maximNDVI_Difference Absolute diffBio1 Annual meanBio2 Mean diurna

Bio3 IsothermalityBio4 TemperatureBio5 Maximum te

Bio6 Minimum teBio8 Mean tempeBio9 Mean tempe

Bio10 Mean tempeBio11 Mean tempeBio12 Annual preci

Bio13 PrecipitationBio15 Precipitation

Bio16 PrecipitationBio19 Precipitation

estimations about the fate of the herpetofauna These GCMs werethe Hadley Centre Coupled Model version 3 (HadCM3) the Cana-dian second-generation coupled global climate model (CGCM2-CCCma) the Australian model (CSIRO Mk2) and a Japanese model(NIES99) all frequently used in similar studies (eg Arauacutejo et al2006 Holt et al 2009) Our results assume that species try totrack their suitable habitat (dependent on dispersal ability) ratherthan accommodate or adapt to the new conditions

For each GCMmodel and time slice two emission scenarios (A2aand B2a) were used the most commonly used scenarios in climate-change assessments (Hannah 2011) They reflect two differentassumptions about the levels of CO2 emissions linked to assumeddemographic changes and socio-economic and technological de-velopments (Marini et al 2009) The A2a scenario assumes a lot ofclimate change caused by medium to high CO2 emission rates (theldquobusiness as usualrdquo scenario) while the B2a scenario assumesrelatively little change because of mitigation measures (theldquomoderaterdquo scenario) (Sauer et al 2011 Taubmann et al 2011)There are no data on how NDVI and habitat variables might changein Egypt in the future and so we assume they (and altitude) remainconstant Apart from the A2a scenario for 2080 nearly all the ex-pected climatic conditions are similar to current values (see Fig S4)

24 Species distribution modelling

We used Maxent v333k (Phillips et al 2006 e see httpwwwcsprincetonedu~schapiremaxent) to run the species dis-tribution models Maxent performs well in comparison withalternative algorithms and shows equal or higher predictive accu-racy on cross-validation (Elith et al 2006 Hernandez et al 2006)Crucial to our decision not to use ensemble modelling Maxent isvery robust to small numbers of training records outperformsother algorithms when using a relatively small number of occur-rence records (Franklin 2009) and is robust against moderatelocation errors (Graham et al 2008) We used lsquopermutationimportancersquo (see Phillips and ATampT Research 2011) to detect whichpredictor had the greatest influence on the model because it doesnot depend on the path used by the algorithm (Songer et al 2012)

Ten replicated runs with cross-validation and default settings(except iterations frac14 1000) were made for each species by itera-tively partitioning the records to use 90 for training and 10 fortesting (as recommended by Phillips and ATampT Research 2011)this gives a more stable model performance (avoiding overfitting)and minimises the effects of errors and biases The default logistic

th grey show variables with VIF values less than 10 andrejected a priori (see text)

um valueerence between the highest and lowest NDVI valuestemperature

l range (mean of monthly (max temp min temp))

(Bio2Bio7) (100)seasonality (standard deviation100)mperature of the warmest month

mperature of the coldest monthrature of the wettest quarterrature of the driest quarter

rature of the warmest quarterrature of the coldest quarterpitation

of the wettest monthseasonality (coefficient of variation)

of the wettest quarterof the coldest quarter

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221214

output format was chosen ie related to probability of suitableconditions ranging from 0 to 1 Logistic output has been criticizedrecently (eg Merow et al 2013) becausemany users assume that itrepresents the probability of occurrence to do that however thecalculation must assume that overall prevalence is fixed at 05 forall species which may not be accurate estimating true prevalenceis very difficult (Elith et al 2011) Both logistic and raw outputs aremonotonically related and rank locations identically (Elith et al2011) making rank-based measures such as AUC (for model eval-uation Elith et al 2011) and thresholding (using values frompredictions of the training presence locations eg 10th-percentilethreshold) identical whichever is used We used logistic output fortwomain reasons a) it standardises the expected habitat suitabilityvalues across species permitting the adding together of species tocreate the lsquoprobability richnessrsquo maps b) it allows Maxent auto-matically to convert the probabilities into expected presenceabsence (or more precisely suitablenon-suitable) habitat For thelatter we used the 10th-percentile training presence threshold(following Pearson et al 2007) For each species (and for each ofthe logistic and thresholded distributions) the ten replicate dis-tribution maps were converted to a single consensus map byMaxent for logistic values and manually for thresholded values(where each pixel was a consensus presence if it had presencevalues in more than half of the model runs ie gt5)

Distributions were then projected into three future time slices(2020 2050 and 2080) in each of four GCM models (HadCM3CCCma CSIRO and NIES99) and two emission scenarios (A2a andB2a) giving a total of 24 future projections Thresholded futuredistributions involved two extreme assumptions about dispersalability unlimited and no dispersal For each of the 24 projectionsthree consensus maps were produced one using the probability(logistic) output and two using the thresholded distributions(assuming unlimited and no-dispersal) The maps were then aver-aged across the different GCMs for thresholded distributions thisinvolved assigning a pixel as potentially suitable if it had a value ofpresence in more than two GCMs Thus there were 18 (three timeslices two scenarios three consensus) maps per species

25 Comparisons across species

Current and future maps of estimated species richness werecalculated in two different ways first by simply adding togetherthe average probability of suitable conditions for each pixel acrossthe distribution maps of all species (this assumes unlimiteddispersal eg Gilbert and Zalat 2008) and second by addingtogether the thresholded distribution maps (under unlimited- andno-dispersal assumptions eg Trotta-Moreu and Lobo 2010) Mapsof future species richness were obtained for each of the 24 futureprojections and then averaged across the four GCMs Future po-tential changes in species richness (but not composition) werecalculated by subtracting current species richness at each pixel

At the species level gains and losses in future distribution werecalculated using current and future consensus thresholded distri-butions Then the losses across all the 24 future potential distri-butions across all species were added together into a species lossmap showing which areas are expected to suffer most in losingspecies (for each dispersal assumption) The same was carried outfor gains in suitability to produce a species gain map showingwhich areas are expected to gain species (only for the unlimiteddispersal assumption) Species turnover an index of dissimilaritybetween current and future species composition (Thuiller 2004)was calculated for each future projection and dispersal assumption(following Thuiller et al 2005)

In order to assess future extinction risk and to determine whichspecies may require more protection in the future species range

changes were calculated as the percentage loss (or gain) in suitablehabitats carried out for each dispersal assumption Each specieswas classified into one of the following categories Extinct (ex-pected loss of the entire suitable habitat - 100) Critically Endan-gered (loss gt80) Endangered (loss 50e80) Vulnerable (loss30e50) Least Concern (loss lt30) gain 1 (gain lt30) gain 2(gain 30e50) gain 3 (gain 50e80) gain 4 (gain 80e100) andgain 5 (gain gt100) (modified from Thuiller et al 2005) (seeTables S2eS3)

26 Area prioritization for conservation

Several algorithms are now available to prioritize areas forconservation and conservation planning Using the SDM pre-dictions Zonation v31 (Moilanen et al 2012) created a nestedspatial conservation prioritization network to evaluate the effec-tiveness and performance of current Protected Area network inEgypt and identify new sites for conservation Zonation prioritizesthe conservation value of the landscape hierarchically based on theconservation value of sites (cells) (Moilanen et al 2012) by itera-tively removing the least valuable cells from the edges of thelandscape according to set rules The last to be removed are themost important areas (Moilanen et al 2012) We used two removalrules lsquobasic core-area functionrsquo to identify important (or poor)locations where a single or a few species have important occur-rences and lsquoadditive benefit functionrsquo to give more weight to lo-cations with high species richness (Moilanen et al 2012)Fragmentation of the chosen areas is minimized by an aggregationrule and we chose to use lsquodistribution smoothingrsquo for which anestimate of dispersal distance is required The dispersal ability ofalmost all terrestrial reptile species is very limited (Cadby et al2010 Edgar et al 2010) To be conservative in this study thedispersal distance of all reptiles was set to be equal to 1 km

Because some species are more important to conserve thanothers each species was given a weight Weights were developedfrom the global and national IUCN assessments the world distri-bution and distribution patterns within Egypt Following Gilbertand Zalat (2008) and Basuony et al (2010) we classified Egyptianreptiles according to their world distribution and their distributionpattern within Egypt We assessed their IUCN Red List status inEgypt according to IUCN guideline and categories (IUCN Standardsand Petitions Subcommittee 2010) although inevitably incom-plete We gave each element a score (modified from Leach et al2013) to indicate relative importance and then the sum of thescores gave the relative conservation weight of each species (seeTables S1 and S4) We used these values to weight the Zonationprioritization process

For current and future climates Zonation was run using themaps of habitat suitabilities for all species The number of cellsremoved at each iteration was set to 10 (frac14lsquowarp factorrsquo) Zonationproduces ASCII raster files with ranked pixel values scaled to befrom 0 to 1 values gt 07 were considered to be of high conservationimportance These important areas were then overlaid with thecurrent Protected Area network in Egypt to show if the network inEgypt is adequate to conserve the species in the face of climatechange and if there are areas outside the current Protected Areasboundaries that require special protection measures

3 Results

31 Model performance amp variable importance

In terms of mean AUC values for the 10 cross-validation runs theperformance of current-distribution models was good (seeTable S1) ranging from 078 to 099 with an overall mean of

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 215

093 plusmn 005 Only one species (Ptyodactylus siphonorhina) had amean AUC of less than 08 while in 36 species it was greater than095 As almost all species had a mean AUC gt08 all models wereaccepted and processed for further analyses More widespreadspecies tended to have lower AUC values (correlations [n frac14 75]with number of suitable pixels rs frac14 085 p lt 0005 extent ofoccurrence rs frac14 077 p lt 0005 number of pixels with recordsrs frac14 0433 p lt 0005)

Variables with highest mean permutation importance across allmodelled species were altitude (highest for 34 species) Bio4(temperature seasonality 10 species) and Bio13 (precipitation ofwettest month 9 species) Those with the lowest mean permuta-tion importance and thus not influencing the final models verymuch were the difference betweenmaximum andminimumNDVIBio15 (precipitation seasonality) habitat and Bio9 (mean temper-ature of driest quarter) the first and last of these were never themost influential variable for any species

32 Species richness

The patterns of current species richness (Fig 1) were similarusing either logistic or thresholded distributions both emphasizingthe lower Nile Valley the Suez Canal area and the coasts of theMediterranean and Suez Gulf (Fig S2 for locations map) The

Fig 1 Current estimated reptile species richness of Egypt using the sum of (a)Maxent logistic predictions and (b) thresholded predicted distributions

pattern is concordant with the species richness map in Baha El Din(2006) produced using records and qualitative criteria (seeFig S5a) Under climate change regardless of emission scenario ordispersal assumptions several areas are expected to have increasedspecies richness (Fig 2) with generally greater magnitudes underthe A2a scenario while others are expected to decline (eg the SuezCanal area) with greater declines under the B2a scenario by 2050and 2080

Areas with the highest expected loss of species include betweengreater Cairo Ismailia and Suez south of Suez on both sides of theSuez Gulf and Wadi El-Natrun much greater under the B2a sce-nario (Fig 3a) If species can disperse well then the highest gains inspecies are expected to include a large proportion of the northernhalf of the Western Desert again greater under the B2a scenario(Fig 3b) In terms of percentage change of the reptile assemblage ifdispersal is unlimited then the Western Desert is expected to un-dergo high turnover (Fig S6a) but this is because there are so fewspecies there currently Under no dispersal turnover is very limited(Fig S6b) again with highest values again in the Western desertbecause of the species poor assemblages there

33 Range and status changes

With unlimited dispersal no species is expected to becomeldquoextinctrdquo (100 loss of suitable habitat) under all or the average ofall GCMs There are a couple of species expected to become extinctby losing their entire area of suitable habitat in at least one of thefuture projections Tarentola mindiae and Hemidactylus robustusUsing the average gain or loss of suitable habitat across the fourdifferent GCMs eight species are expected to be classified as Crit-ically Endangered (loss gt 80) twenty as Endangered (loss50e80) and twenty-two as Vulnerable (loss 30e50) in at leastone mean future projection (details see Table S2)

When assuming no dispersal at all none of the species are ex-pected to lose all suitable habitats under all or the average of allGCMs but T mindiae and H robustus again experience completeloss of habitat under at least one of the projections Using theaverage loss of suitable habitat across the four different GCMs tenspecies are expected to be classified as Critically Endangeredtwenty-seven as Endangered and thirty-three as Vulnerable in atleast one mean future projection (details see Table S3)

34 Area prioritization for conservation

The areas with the highest current prioritization value for con-servation were similar for both cell-removal methods mainlyemphasizing the north coast Suez Canal area South Sinai theGebel Elba region Qattara Depression Wadi El-Natrun and aroundCairo amp Fayoum with the lsquoadditive benefitrsquo function putting greateremphasis on the Nile Delta and its fringing areas (Fig 4) Thisoverall pattern does not change much under future climate pro-jections though a few areas decline in priority and much greaterarea of the northern Western Desert increases These effects aremore pronounced with the lsquocore-area functionrsquo used as the cellremoval rule

The mean prioritization value in all models was higher in Pro-tected Areas than outside them (Fig 5) with both slightlyincreasing in future projections The difference in prioritizationvalue between inside and outside Protected Areas declined withtime especially by 2050 and 2080 under the B2a scenario

4 Discussion

The results of this study show for first time the potential impactsof climate change on the distribution of Egyptian reptiles Some

Fig 2 Estimated future changes in species richness as a result of anthropogenic climate change under two IPCC scenarios (A2a B2a) and at three times in the future (2020 20502080) Each map is the difference from the current species richness (shown in Fig 1a) and is averaged across four GCMmodels Maps made (a) using Maxent logistic output and (b)using thresholded distributions (see Methods)

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221216

reptiles are expected to lose much of their currently suitabilityareas (up to 80 in some cases) and some areas will lose or changea large proportion of their species Thus there will be a need forgreater conservation efforts in the future Some of these areas arealready inside Protected Areas and so hypothetically they are pro-tected while others are either just outside or very far from Pro-tected Areas Although overall Protected Areas contain better(higher priority) values compared to unprotected sites the currentProtected-Area network appears inadequate for future conserva-tion Our results together with others (eg Leach et al 2013) willhelp the Egyptian Environmental Affairs Agency and otherdecision-making parties in Egypt to direct conservation efforts andlimited conservation resources in the right direction

41 Model performance amp variable importance

The negative correlations found between AUC and the areaoccupied and with the extent of occurrence concur with findingsof many other studies (eg Brotons et al 2004 Elith et al 2006Hernandez et al 2006) Rare species are usually habitat special-ists with narrow environmental tolerances and are environmen-tally or geographically restricted compared to widespread species

(Elith et al 2006 Pandit et al 2009) Widespread species are likelyto be generalists occupying a wide range of habitats and climatesmaking it difficult to distinguish between suitable and unsuitablehabitats (Franklin et al 2009) However as pointed out by Elithet al (2006) this may be an artefact an inevitable result for spe-cies with small relative occurrence areas and perhaps a reason notto use AUC to compare model performances (Lobo et al 2008)Most (about 60) of the Egyptian herpetofauna are narrowlydistributed occupying less than 10 of Egypts area (Baha El Din2006) There is no evidence of a strong correlation betweenmodel accuracy and the number of records used (see Elith et al2006 Newbold et al 2009b) and the weak negative correlationfound here is unusual since most (eg Stockwell and Peterson2002 Hernandez et al 2006) find that models with larger sam-ple sizes have higher accuracy (but see de Pous et al 2011)

We found altitude to be the most effective predictor for manyspecies In contrast the NDVI predictors were not useful unlike inmany other cases (eg Egbert et al 2002 Anderson et al 2006Kgosiesele 2010 Taheri 2010) but perhaps typical for arid envi-ronments (eg Soultans (2011) study on Egyptian antelopes) or forherpetofauna (eg Teodoro et al 2013) Habitat categories likewisemade only a low contribution to the reptile models

Fig 3 Expected patterns of cumulative (a) losses and (b) gains of species as a result of climate change The losses of each species in each pixel (under both dispersal assumptions)and gains (under unlimited dispersal only) were averaged across GCM models accumulated across species and then plotted

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 217

42 Species richness

Species richness by itself is not adequate as an indicator of bioticchange since species composition can change independently Ourresults suggest that in the future some areas with currently highexpected species richness may lose many species or undergo largechanges in species composition Some of these areas are alreadyunder a certain degree of protection but even if they can bemaintained or enhanced (eg Siwa oasis Gebel Elba and the coastalareas over the Aqaba gulf Fig S2) current Protected Areas are notsufficient to prevent expected loss of species Other areas (eg WadiEl-Natrun the Suez Canal area Red Sea inland wadis the poorlyknown Qattara Depression and the areas around Gebel El-Gallalaand Gebel El-Hallal) are unprotected and measures are needed toconserve them There are similar situations elsewhere for examplehigh species loss and turnover in the reptiles of a priority conser-vation hotspot in Europe (the Iberian Peninsula) are expected (dePous 2011) with great variation in expected patterns of loss andchange with different projections and dispersal assumptions Otherstudies have also concluded that large changes are likely in chelo-nians worldwide (Ihlow et al 2012) Mediterranean mammals(Maiorano et al 2011) US trees and vertebrates (Currie 2001) andMesoAmerican trees (Golicher et al 2012)

43 Range and status changes

Two species are expected to lose their suitable habitats entirelyin at least one future projection 1) T mindiae is a near-endemicspecies recorded just from northwest Egypt and northern Cyrena-ica (eastern Libya) its distribution in Egypt is restricted to Siwaoasis the Qattara Depression and their periphery (Baha El Din2006) Future range changes suggest it will be Endangered (loss50e80) by 2020 and as Critically Endangered (loss gt80) by 2050and 2080 (under all emission scenarios and dispersal assumptions)2) H robustus is localized to the East African coast from Zanzibar tosouthern Egypt Arabia east to Pakistan its distribution in Egypt ison the Red Sea coast from El-Quseir southwards (Baha El Din2006) The A2a scenario suggests it will be Critically Endangeredby 2020 and 2050 while under the B2a scenario it will be Endan-gered by 2020 and Critically Endangered by 2050 and 2080 Morethan half of its current distribution in Egypt is located within theWadi El-Gemal and Gebel Elba Protected Areas so hypothetically itis protected However it is found mainly on buildings and henceassociated with man it may in fact be an invasive to Egypt

Another eight species are expected to be classified as CriticallyEndangered (loss gt80) in at least one averaged future projection(Acanthodactylus longipes Cerastes vipera Eryx jaculus Eumeces

Fig 4 Current conservation prioritizationl of areas according to the Zonation algo-rithm under two cell removal rules (a) core-area function (identifies areas where oneor more species have important occurrences) and (b) additive benefit function (whichgives greater emphasis to areas with high species richness) Darker colours representhigher conservation value (For interpretation of the references to colour in this figurelegend the reader is referred to the web version of this article)

Fig 5 Mean prioritization value for conservation (plusmn95 confidence limits) derivedfrom Zonation for Protected Areas (PAs) and outside Protected Areas (non-PAs) forcurrent and expected future conditions using the cell-removal rule of (a) core-areafunction and (b) additive benefit function

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221218

schneiderii Malpolon moilensis Ptyodactylus guttatusP siphonorhina and Trapelus mutabilis) According to the currentassessment twenty-seven species are classified as Endangered(loss 50e80) in at least one projection (mostly by 2080 in the A2scenario) two of these are classified assuming no dispersal asEndangered in all future projections (Uromastyx aegyptia and Pris-turus flavipunctatus for details see Tables S2eS3)

Reptiles are also likely to gain from global warming especially inmore northern areas of the globe if they can disperse and expandtheir distributions but under perhaps the more realistic assump-tion of no dispersal most species were expected to lose habitatsuitability by 2050 (Arauacutejo et al 2006) In Egypt areas of clearfuture expected gain (assuming unlimited dispersal) are the areabetween the Siwa oasis and Qattara Depression northwards almostto the northern coastal strip and some sparse locations in centralSinai There were greater gains under the B2a scenario in theeastern part of the Western Desert and some patchy areas aroundthe Nile (Fig 3b) This pattern is similar to that reported by Araujoet al (2006)

Some of the species identified here as being most at risk due tofuture climate change are in fact resilient species that are ecolog-ically very adaptable and are currently experiencing range expan-sion and population growth at least in Egypt (eg H robustus) This

disparity is probably because modelling is based solely on theprobable impacts of climate change on the spatial distribution ofsuitable habitats without taking evolution or ecological adapt-ability and resilience into account Species vary greatly in theirresponse to ecological change some adapting very well even tovery rapid anthropogenically induced changes (eg invasive spe-cies) while others are much less adaptable and tend to go extinctfirst

There are no published studies that discuss the effect of climatechange on Egyptian reptiles obviating any comparisons Informa-tion on the reptiles of the adjacent countries of Sudan and Libya arevery limited making it impossible to assess any movements thatmight compensate for Egyptian losses The patterns of expectedchanges in Leach et als (2013) similar study on mammals andbutterflies do not concur with ours Mammal species richness wasexpected to decline in the Mediterranean and Red Sea areas (by40e60) and increase elsewhere (by 80e100) while for butter-flies declines are expected over almost the whole of Egypt (by40e60) except the south which is expected to increase (by40e60) In the models of European reptiles (de Pous 2011)

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 219

between eight (unlimited dispersal) and 21 (no dispersal) specieswere expected to become extinct in at least one future projectionwith two more losing more than 80 of their currently suitablehabitats This is a much more damaging scenario than the expectedresponses of Egyptian reptiles in our study There are no othercomparable studies from North Africa or the Middle East

44 Area prioritization for conservation

In our study Protected Areas (Fig S3) were more valuable inconserving reptiles than areas outside them both currently and inthe future although the difference is expected to get smaller withtime (Fig 5) Although this is encouraging the Protected Areanetwork appears inadequate to conserve Egyptian reptiles NewProtected Areas that have already been proposed would go someway towards reducing this deficiency in coverage although otherswill be needed in the middle to north Sinai (especially aroundGebel El-Hallal) the Suez Canal area both sides of the Suez Gulf thenorthern Mediterranean coast coastal wadis of the Red Sea be-tween Hurghada and Mersa Alam and the Gebel El-Gallala areaMore effective protection is required in the established ProtectedAreas including those in South Sinai Gebel Elba and Siwa oasis dueto their relatively high expected future suitability loss The Pro-tected Areas of Siwa oasis and El-Gilf El-Kebir plus the areas ofWadiEl-Natrun the Gebel El-Hallal area and wadis of the Red Sea areexpected to have particularly high turnover in species composition(Fig S6) and hence probably need more conservation efforts Somecomparable information is available for Egypts butterflies andmammals (Gilbert and Zalat 2008 Basuony et al 2010 Leach et al2013) Egypts network of Protected Areas appears adequate toconserve Egyptian butterfly hotspots except for the Mediterraneancoast between Alexandria and Sallum (Gilbert and Zalat 2008) butinadequate to conserve important mammal areas with new Pro-tected Areas needed to be constructed in the lower Nile Valleyalong the north coast between Alexandria and Sallum top part ofthe Suez Gulf and perhaps the Qattara Depression (Basuony et al2010) Suggestions for conserving mammals are consistent to someextent with our results which also suggest new Protected Areasalong the Suez Gulf Mediterranean coast from Mersa Matruh toSallum and the Qattara Depression Leach et al (2013) concur withour results in suggesting priority areas along the western Medi-terranean coast around Greater Cairo central and north Sinai andthe Red Sea coast

45 Limitations

There are some limitations of the results shown in this study 1)Egyptian climate data in WorldClim are interpolated from ratherfew non-randomly placed weather stations in Egypt and adjacentcountries Although making the climate data less reliable thanelsewhere there are no alternative climate data available 2) Anensemble approach (eg using BIOMOD Thuiller et al 2009) isoften suggested for SDM modelling to avoid dependence on oneparticular algorithm This is not possible without a greater numberof records since other algorithms are not as flexible or robust asMaxent Despite all the caveats with using SDMs and criticisms ofMaxent in particular (Royle et al 2012 but see Hastie and Fithian2013 Elith et al 2006 Hernandez et al 2006) in practice coun-tries such as Egypt have little else to guide them in attempts atclimate change adaptation 3) There were signs of sampling bias inthe available records Although this violates the assumptions of allSDMs it is a problem common to nearly all datasets worldwideNew techniques may allow for the correction of sampling bias egusing target-group approach (Phillips et al 2009) although thedata requirements are currently impossible for Egypt and most

other developing countries) 4) Using Maxents default settings iscriticized by some authors they need to be modified dependent onthe data (Warren and Seifert 2011 Radosavljevic and Anderson2014 Merow et al 2013) but we think using the default settingsis enough in this study

5 Conclusion

Although Egyptian datasets exist knowledge of Egyptian spe-cies and conservation priorities are limited (with consequencesexplored in Leach et al 2013) nevertheless our results and those ofLeach et al (2013) encourage us to believe that useful insights canbe gained for conservation priorities by using the combination ofspecies distribution modelling plus spatial conservation prioriti-zation techniques (Hannah 2010 Klorvuttimontara et al 2011) Itis crucial for the Egyptian Environmental Affairs Agency to establishan accurate and reliable database of records assessed by experts tohelp improve the basic information for future conservation as-sessments and analyses

Although many SDM studies have been done on a variety oforganisms in recent decades aiming to solve conservation prob-lems evidence of their practical utility in real-world conservation issparse with very few studies clearly targeting conservation de-cisions (Guisan et al 2013) If implemented correctly SDMs canplay a critical role in supporting spatial conservation decision-making (Addison et al 2013) For reptiles there is a shortage ofstudies estimating their response to expected climate changes(Arauacutejo et al 2006) Although some suggest that reptiles (and to agreater extent amphibians) will be strongly affected by climatechange (eg Gibbons et al 2000 Arauacutejo et al 2006 Carvalho et al2010) very few of themwere concerned with arid environments orthe Middle East We hope that this and other similar studies thatestimate the fate of reptiles under climate change help decision-making authorities to take the measures necessary to conservethem

Acknowledgements

We thank the Italian Cooperation (Debt Swap) in Egypt forfunding the BioMAP project that allowed the collation of the da-tabases used in this work Dr Tim Newbold for allowing the use ofthe data from his field trips Dr Mustafa Fouda (Former Director ofthe Nature Conservation Sector EEAA) for facilities offered duringBioMAP project and all the BioMAP staff This work was supportedby the award of a Chevening Scholarship Masters grant to AEl-GWe are especially grateful to two anonymous referees whogreatly improved an earlier manuscript

Appendix A Supplementary data

Supplementary data related to this article can be found at httpdxdoiorg101016jjaridenv201512007

References

Addison PFE Rumpff L Bau SS Carey JM Chee YE Jarrad FC McBride MFBurgman MA 2013 Practical solutions for making models indispensable inconservation decision-making Divers Distrib 19 (5e6) 490e502 httpdxdoiorg101111ddi12054

Anderson RP Peterson AT Egbert SL 2006 Vegetation-index models predictareas vulnerable to purple loosestrife (Lythrum salicaria) invasion in KansasSouthwest Nat 51 471e480 httpdxdoiorg1018940038-4909(2006)51[471VMPAVT]20CO2

Arauacutejo MB 2009 Climate change and spatial conservation planning InMoilanen A Wilson KA Possingham HP (Eds) Spatial Conservation Prior-itization Quantitative Methods and Computational Tools Oxford UniversityPress pp 172e184 ISBN 9780199547760

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221220

Arauacutejo MB Rahbek C 2006 How does climate change affect biodiversity Sci-ence 313 1396e1397 httpdxdoiorg101126science1131758

Arauacutejo MB Thuiller W Pearson RG 2006 Climate warming and the decline ofamphibians and reptiles in Europe J Biogeogr 33 1712e1728 httpdxdoiorg101111j1365-2699200601482x

Baha El Din SM 2001 The Herpetofauna of Egypt Species Communities andAssemblages University of Nottingham p 435 Unpublished PhD thesis

Baha El Din SM 2006 A Guide to the Reptiles and Amphibians of Egypt TheAmerican University in Cairo Press ISBN 9789774249792

Baha El Din SM 2007 A new lizard of the Acanthodactylus scutellatus group(Squamata Lacertidae) from Egypt Zool Middle East 40 21e32 httpdxdoiorg10108009397140200710638200

Baldwin RA 2009 Use of maximum entropy modeling in wildlife research En-tropy 11 854e866 httpdxdoiorg103390e11040854

Ball IR Possingham HP Watts M 2009 Marxan and relatives software forspatial conservation prioritisation In Moilanen A Wilson KAPossingham HP (Eds) Spatial Conservation Prioritization QuantitativeMethods and Computational Tools Oxford University Press pp 185e195 ISBN9780199547760

Basuony MI Gilbert F Zalat S 2010 The Mammals of Egypt Atlas Red DataListing amp Conservation BioMAP amp CultNat EEAA amp Bibliotheca AlexandrinaCairo available at httpecologynottinghamacuk~plzfg

Bombi P Salvi D Bologna MA Pettorelli N 2012 Cross-scale predictions allowthe identification of local conservation priorities from atlas data Anim Conserv15 378e387 httpdxdoiorg101111j1469-1795201200526x

Brotons L Thuiller W Arauacutejo MB Hirzel AH 2004 Presence-absence versuspresence-only modelling methods for predicting bird habitat suitability Ecog-raphy 27 437e448 httpdxdoiorg101111j0906-7590200403764x

Cadby CD While GM Hobday AJ Uller T Wapstra E 2010 Multi-scaleapproach to understanding climate effects on offspring size at birth and date ofbirth in a reptile Integr Zool 5 164e175 httpdxdoiorg101111j1749-4877201000201x

Carvalho SB Brito JC Crespo EJ Possingham HP 2010 From climate changepredictions to actions e conserving vulnerable animal groups in hotspots at aregional scale Glob Change Biol 16 (12) 3257e3270 httpdxdoiorg101111j1365-2486201002212x

Currie DJ 2001 Projected effects of climate change on patterns of vertebrate andtree species richness in the conterminous United States Ecosystems 4216e225 httpdxdoiorg101007s10021-001-0005-4

de Pous P 2011 Climate Change and European Reptiles from Large-scale SpatialRange Shift Projections towards a More Realistic Assessment of Species Re-sponses University of East Anglia UK p 64 MSc thesis

de Pous P Beukema W Weterings M Duumlmmer I Geniez P 2011 Area priori-tization and performance evaluation of the conservation area network for theMoroccan herpetofauna a preliminary assessment Biodivers Conserv 2089e118 httpdxdoiorg101007s10531-010-9948-0

Edgar P Foster J Baker J 2010 Reptile Habitat Management HandbookAmphibian and Reptile Conservation Bournemouth

Egbert SL Martiacutenez-Meyer E Ortega-Huerta M Peterson AT 2002 Use ofdatasets derived from time-series AVHRR imagery as surrogates for land covermaps in predicting species distributions In Proceedings IEEE 2002 Interna-tional Geoscience and Remote Sensing Symposium (IGARSS) vol 4pp 2337e2339 httpdxdoiorg101109IGARSS20021026537

El Alqamy H Ismael A Abdelhameed A Nagy A Hamada A et al 2010 Pre-dicting the status and distribution of the Nubian Ibex (Capra nubiana) in thehigh-altitude mountains of south Sinai (Egypt) In Galemys Newsletter of theSpanish Society for the Preservation and Study of Mammals vol 22pp 517e530

Elith J Graham CH Anderson RP Dudiacutek M Ferrier S et al 2006 Novelmethods improve prediction of species distributions from occurrence dataEcography 29 129e151 httpdxdoiorg101111j20060906-759004596x

Elith J Phillips SJ Hastie T Dudiacutek M Chee YE Yates CJ 2011 A statisticalexplanation of MaxEnt for ecologists Divers Distrib 17 (1) 43e57 httpdxdoiorg101111j1472-4642201000725x

FAO 2012 Food and Agriculture Organisation World Development IndicatorsAverage Precipitation in Depth (mm per year) httpdataworldbankorgindicatorAGLNDPRCPMM

Franklin J 2009 Mapping Species Distributions Spatial Inference and PredictionCambridge University Press Cambridge UK ISBN 9780521700023

Franklin J Wejnert KE Hathaway SA Rochester CJ Fisher RN 2009 Effect ofspecies rarity on the accuracy of species distribution models for reptiles andamphibians in southern California Divers Distrib 15 167e177 httpdxdoiorg101111j1472-4642200800536x

Gibbons JW Scott DE Ryan TJ Buhlmann KA Tuberville TD Metts BSGreene JL Mills T et al 2000 The global decline of reptiles Deja Vu Am-phibians BioScience 50 (8) 653e666 httpdxdoiorg1016410006-3568(2000)050[0653TGDORD]20CO2

Gilbert F Zalat S 2008 The Butterflies of Egypt Atlas Red Data Listing amp Con-servation BioMAP EEAA Cairo available at httpecologynottinghamacuk~plzfg

Golicher DJ Cayuela L Newton AC 2012 Effects of climate change on the po-tential species richness of Mesoamerican forests Biotropica 44 284e293httpdxdoiorg101111j1744-7429201100815x

Graham CH Elith J Hijmans RJ Guisan A Peterson AT et al 2008 Theinfluence of spatial errors in species occurrence data used in distribution

models J Appl Ecol 45 239e247 httpdxdoiorg101111j1365-2664200701408x

Guisan A Tingley R Baumgartner JB Naujokaitis-Lewis I Sutcliffe PRTulloch AIT Regan TJ Brotons L et al 2013 Predicting species distributionsfor conservation decisions Ecol Lett 16 (12) 1424e1435 httpdxdoiorg101111ele12189

Hannah L 2010 A global conservation system for climate-change adaptationConserv Biol 24 70e77 httpdxdoiorg101111j1523-1739200901405x

Hannah L 2011 Climate Change Biology Elsevier Ltd ISBN 9780123741820Hannah L Lovejoy TE Schneider SH 2005 Biodiversity and climate change in

context In Lovejoy TE Hannah L (Eds) Climate Change and BiodiversityYale University Press New Haven CT USA

Hannah L Midgley G Andelman S Arauacutejo M Hughes G et al 2007 Protectedarea needs in a changing climate Front Ecol Environ 5 131e138 httpdxdoiorg1018901540-9295(2007)5[131PANIAC]20CO2

Hastie T Fithian W 2013 Inference from presence-only data the ongoing con-troversy Ecography 36 864e867 httpdxdoiorg101111j1600-0587201300321x

Hernandez PA Graham CH Master LL Albert DL 2006 The effect of samplesize and species characteristics on performance of different species distributionmodeling methods Ecography 29 773e785 httpdxdoiorg101111j0906-7590200604700x

Hijmans RJ Cameron SE Parra JL Jones PG Jarvis A 2005 Very high reso-lution interpolated climate surfaces for global land areas Int J Climatol 251965e1978 httpdxdoiorg101002Joc1276

Holt AC Salkeld DJ Fritz CL Tucker JR Gong P 2009 Spatial analysis ofplague in California niche modeling predictions of the current distribution andpotential response to climate change Int J Health Geogr 8 38 httpdxdoiorg1011861476-072X-8-38

Hoyle M James M 2005 Global warming human population pressure andviability of the worlds smallest butterfly Conserv Biol 19 1113e1124 httpdxdoiorg101111j1523-1739200500166x

Ihlow F Dambach J Engler JO Flecks M Hartmann T et al 2012 On the brinkof extinction How climate change may affect global chelonian species richnessand distribution Glob Change Biol 18 1520e1530 httpdxdoiorg101111j1365-2486201102623x

IPCC 2007 Climate Change 2007 Synthesis Report Contribution of WorkingGroups I II and III to the Fourth Assessment Report of the IntergovernmentalPanel on Climate Change IPCC Geneva Switzerland

Kgosiesele E 2010 Predictive Distribution Modelling of Timon Lepida in SpainUniversity of Twente (ITC) the Netherlands p 87 MSc thesis

Klorvuttimontara S McClean CJ Hill JK 2011 Evaluating the effectiveness ofprotected areas for conserving tropical forest butterflies of Thailand BiolConserv 144 2534e2540 httpdxdoiorg101016jbiocon201107012

Leach K Zalat S Gilbert F 2013 Egypts protected area network under futureclimate change Biol Conserv 159 490e500 httpdxdoiorg101016jbiocon201211025

Lobo JM Jimenez-Valverde A Real R 2008 AUC a misleading measure of theperformance of predictive distribution models Glob Ecol Biogeogr 17145e151 httpdxdoiorg101111j1466-8238200700358x

Maiorano L Falcucci A Zimmermann NE Psomas A Pottier J et al 2011 Thefuture of terrestrial mammals in the Mediterranean basin under climatechange Philosophical Trans R Soc Lond Ser B Biol Sci 366 2681e2692httpdxdoiorg101098rstb20110121

Marini MA Barbet-Massin M Lopes LE Jiguet F 2009 Predicted climate-driven bird distribution changes and forecasted conservation conflicts in aneotropical savanna Conserv Biol 23 1558e1567 httpdxdoiorg101111j1523-1739200901258x

Merow C Smith MJ Silander JA 2013 A practical guide to MaxEnt for modelingspecies distributions what it does and why inputs and settings matterEcography 36 (10) 1058e1069 httpdxdoiorg101111j1600-0587201307872x

Moilanen A Meller L Leppeuroanen J Pouzols FM Arponen A et al 2012 Zona-tion Spatial Conservation Planning Framework and Software v31 User Manualwwwhelsinkifibioscienceconsplan

Newbold T 2009 The Value of Species Distribution Models as a Tool for Conser-vation and Ecology in Egypt and Britain University of Nottingham UK p 312PhD thesis

Newbold T Gilbert F Zalat S El-Gabbas A Reader T 2009a Climate-basedmodels of spatial patterns of species richness in Egypts butterfly and mammalfauna J Biogeogr 36 2085e2095 httpdxdoiorg101111j1365-2699200902140x

Newbold T Reader T Zalat S El-Gabbas A Gilbert F 2009b Effect of charac-teristics of butterfly species on the accuracy of distribution models in an aridenvironment Biodivers Conserv 18 3629e3641 httpdxdoiorg101007s10531-009-9668-5

Pandit SN Kolasa J Cottenie K 2009 Contrasts between habitat generalists andspecialists an empirical extension to the basic metacommunity frameworkEcology 90 2253e2262 httpdxdoiorg10189008-08511

Pearson RG Dawson TP 2003 Predicting the impacts of climate change on thedistribution of species are bioclimate envelope models useful Glob EcolBiogeogr 12 361e371 httpdxdoiorg101046j1466-822X200300042x

Pearson RG Raxworthy CJ Nakamura M Peterson AT 2007 Predicting speciesdistributions from small numbers of occurrence records a test case using

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 221

cryptic geckos in Madagascar J Biogeogr 34 102e117 httpdxdoiorg101111j1365-2699200601594x

Phillips SJ ATampT Research 2011 A Brief Tutorial on Maxent Available at httpwwwcsprincetonedu~schapiremaxenttutorialtutorialdoc (Last accessed290512)

Phillips SJ Anderson RP Schapire RE 2006 Maximum entropy modeling ofspecies geographic distributions Ecol Model 190 231e259 httpdxdoiorg101016jecolmodel200503026

Phillips SJ Dudiacutek M Elith J Graham CH Lehmann A Leathwick J Ferrier S2009 Sample selection bias and presence-only distribution models implica-tions for background and pseudo-absence data Ecol Appl 19 181e197 httpdxdoiorg10189007-21531

R Development Core Team 2012 R a Language and Environment for StatisticalComputing R Foundation for Statistical Computing Vienna Austria ISBN 3-900051-07-0 httpwwwR-projectorg

Radosavljevic A Anderson RP 2014 Making better Maxent models of speciesdistributions complexity overfitting and evaluation J Biogeogr 41 (4)629e643 httpdxdoiorg101111jbi12227

Royle JA Chandler RB Yackulic C Nichols JD 2012 Likelihood analysis ofspecies occurrence probability from presence-only data for modelling speciesdistributions Methods Ecol Evol 3 545e554 httpdxdoiorg101111j2041-210X201100182x

Sauer J Domisch S Nowak C Haase P 2011 Lowmountain ranges summit trapsfor montane freshwater species under climate change Biodivers Conserv 203133e3146 httpdxdoiorg101007s10531-011-0140-y

Songer M Delion M Biggs A Huang Q 2012 Modeling impacts of climatechange on Giant Panda Habitat Int J Ecol 2012 1e12 httpdxdoiorg1011552012108752

Soultan AE 2011 Antelope distribution in Egypt Modeling the Impact of ClimateChange on Antelopes Distribution in Egypt Lambert Academic Publishing

Standards IUCN Subcommittee Petitions 2010 Guidelines for Using the IUCNRed List Categories and Criteria Version 81

Stockwell DRB Peterson AT 2002 Effects of sample size on accuracy of speciesdistribution models Ecol Model 148 1e13 httpdxdoiorg101016S0304-3800(01)00388-X

Suarez AV Tsutsui ND 2004 The value of museum collections for research andsociety Bioscience 54 66 httpdxdoiorg1016410006-3568(2004)054[0066tvomcf]20co2

Taheri S 2010 Hyper Temporal NDVI Images for Modelling and Prediction theHabitat Distribution of Balkan Green Lizard (Lacerta Trilineata) e Case Study

Crete (Greece) University of Twente (ITC) the Netherlands p 54 MScthesis

Taubmann J Theissinger K Feldheim KA Laube I Graf W et al 2011Modelling range shifts and assessing genetic diversity distribution of themontane aquatic mayfly Ameletus inopinatus in Europe under climate changescenarios Conserv Genet 12 503e515 httpdxdoiorg101007s10592-010-0157-x

Teodoro AC Sillero N Alves S Duarte L 2013 Correlation between the habitatsproductivity and species richness (amphibians and reptiles) in Portugal throughremote sensed data In Proc SPIE 8887 Remote Sensing for Agriculture Eco-systems and Hydrology httpdxdoiorg101117122028502

Thomas CD Cameron A Green RE Bakkenes M Beaumont LJ et al 2004Extinction risk from climate change Nature 427 145e148 httpdxdoiorg101038nature02121

Thuiller W 2004 Patterns and uncertainties of species range shifts under climatechange Glob Change Biol 10 2020e2027 httpdxdoiorg101111j1365-2486200400859x

Thuiller W 2007 Biodiversity climate change and the ecologist Nature 448550e552 httpdxdoiorg101038448550a

Thuiller W Lavorel S Arauacutejo MB Sykes MT Prentice IC 2005 Climate changethreats to plant diversity in Europe Proc Natl Acad Sci U S A 1028245e8250 httpdxdoiorg101073pnas0409902102

Thuiller W Lafourcade B Engler R Arauacutejo MB 2009 BIOMOD e a platform forensemble forecasting of species distributions Ecography 32 (3) 369e373httpdxdoiorg101111j1600-0587200805742x

Tolba MK Saab NW 2009 Arab environment Climate Change - Impact ofClimate Change on Arab Countries Arab Forum for Environment andDvelopment

Trotta-Moreu N Lobo JM 2010 Deriving the species richness distribution ofGeotrupinae (Coleoptera Scarabaeoidea) in Mexico from the overlap of indi-vidual model predictions Environ Entomol 39 42e49 httpdxdoiorg101603EN08179

Warren DL Seifert SN 2011 Ecological niche modeling in Maxent the impor-tance of model complexity and the performance of model selection criteriaEcol Appl 21 (2) 335e342 httpdxdoiorg10189010-11711

Wilson KA Cabeza M Klein CJ 2009 Fundamental concepts of spatial con-servation prioritisation In Moilanen A Wilson KA Possingham HP (Eds)Spatial Conservation Prioritization Quantitative Methods and ComputationalTools Oxford University Press pp 16e27 ISBN 9780199547760

Page 4: Journal of Arid Environments - elgabbas.github.io · e Faculty of Sciences & Arts, Taibah University, Al-Ula Campus, Saudi Arabia f Department of Zoology, Suez Canal University, Ismailia,

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221214

output format was chosen ie related to probability of suitableconditions ranging from 0 to 1 Logistic output has been criticizedrecently (eg Merow et al 2013) becausemany users assume that itrepresents the probability of occurrence to do that however thecalculation must assume that overall prevalence is fixed at 05 forall species which may not be accurate estimating true prevalenceis very difficult (Elith et al 2011) Both logistic and raw outputs aremonotonically related and rank locations identically (Elith et al2011) making rank-based measures such as AUC (for model eval-uation Elith et al 2011) and thresholding (using values frompredictions of the training presence locations eg 10th-percentilethreshold) identical whichever is used We used logistic output fortwomain reasons a) it standardises the expected habitat suitabilityvalues across species permitting the adding together of species tocreate the lsquoprobability richnessrsquo maps b) it allows Maxent auto-matically to convert the probabilities into expected presenceabsence (or more precisely suitablenon-suitable) habitat For thelatter we used the 10th-percentile training presence threshold(following Pearson et al 2007) For each species (and for each ofthe logistic and thresholded distributions) the ten replicate dis-tribution maps were converted to a single consensus map byMaxent for logistic values and manually for thresholded values(where each pixel was a consensus presence if it had presencevalues in more than half of the model runs ie gt5)

Distributions were then projected into three future time slices(2020 2050 and 2080) in each of four GCM models (HadCM3CCCma CSIRO and NIES99) and two emission scenarios (A2a andB2a) giving a total of 24 future projections Thresholded futuredistributions involved two extreme assumptions about dispersalability unlimited and no dispersal For each of the 24 projectionsthree consensus maps were produced one using the probability(logistic) output and two using the thresholded distributions(assuming unlimited and no-dispersal) The maps were then aver-aged across the different GCMs for thresholded distributions thisinvolved assigning a pixel as potentially suitable if it had a value ofpresence in more than two GCMs Thus there were 18 (three timeslices two scenarios three consensus) maps per species

25 Comparisons across species

Current and future maps of estimated species richness werecalculated in two different ways first by simply adding togetherthe average probability of suitable conditions for each pixel acrossthe distribution maps of all species (this assumes unlimiteddispersal eg Gilbert and Zalat 2008) and second by addingtogether the thresholded distribution maps (under unlimited- andno-dispersal assumptions eg Trotta-Moreu and Lobo 2010) Mapsof future species richness were obtained for each of the 24 futureprojections and then averaged across the four GCMs Future po-tential changes in species richness (but not composition) werecalculated by subtracting current species richness at each pixel

At the species level gains and losses in future distribution werecalculated using current and future consensus thresholded distri-butions Then the losses across all the 24 future potential distri-butions across all species were added together into a species lossmap showing which areas are expected to suffer most in losingspecies (for each dispersal assumption) The same was carried outfor gains in suitability to produce a species gain map showingwhich areas are expected to gain species (only for the unlimiteddispersal assumption) Species turnover an index of dissimilaritybetween current and future species composition (Thuiller 2004)was calculated for each future projection and dispersal assumption(following Thuiller et al 2005)

In order to assess future extinction risk and to determine whichspecies may require more protection in the future species range

changes were calculated as the percentage loss (or gain) in suitablehabitats carried out for each dispersal assumption Each specieswas classified into one of the following categories Extinct (ex-pected loss of the entire suitable habitat - 100) Critically Endan-gered (loss gt80) Endangered (loss 50e80) Vulnerable (loss30e50) Least Concern (loss lt30) gain 1 (gain lt30) gain 2(gain 30e50) gain 3 (gain 50e80) gain 4 (gain 80e100) andgain 5 (gain gt100) (modified from Thuiller et al 2005) (seeTables S2eS3)

26 Area prioritization for conservation

Several algorithms are now available to prioritize areas forconservation and conservation planning Using the SDM pre-dictions Zonation v31 (Moilanen et al 2012) created a nestedspatial conservation prioritization network to evaluate the effec-tiveness and performance of current Protected Area network inEgypt and identify new sites for conservation Zonation prioritizesthe conservation value of the landscape hierarchically based on theconservation value of sites (cells) (Moilanen et al 2012) by itera-tively removing the least valuable cells from the edges of thelandscape according to set rules The last to be removed are themost important areas (Moilanen et al 2012) We used two removalrules lsquobasic core-area functionrsquo to identify important (or poor)locations where a single or a few species have important occur-rences and lsquoadditive benefit functionrsquo to give more weight to lo-cations with high species richness (Moilanen et al 2012)Fragmentation of the chosen areas is minimized by an aggregationrule and we chose to use lsquodistribution smoothingrsquo for which anestimate of dispersal distance is required The dispersal ability ofalmost all terrestrial reptile species is very limited (Cadby et al2010 Edgar et al 2010) To be conservative in this study thedispersal distance of all reptiles was set to be equal to 1 km

Because some species are more important to conserve thanothers each species was given a weight Weights were developedfrom the global and national IUCN assessments the world distri-bution and distribution patterns within Egypt Following Gilbertand Zalat (2008) and Basuony et al (2010) we classified Egyptianreptiles according to their world distribution and their distributionpattern within Egypt We assessed their IUCN Red List status inEgypt according to IUCN guideline and categories (IUCN Standardsand Petitions Subcommittee 2010) although inevitably incom-plete We gave each element a score (modified from Leach et al2013) to indicate relative importance and then the sum of thescores gave the relative conservation weight of each species (seeTables S1 and S4) We used these values to weight the Zonationprioritization process

For current and future climates Zonation was run using themaps of habitat suitabilities for all species The number of cellsremoved at each iteration was set to 10 (frac14lsquowarp factorrsquo) Zonationproduces ASCII raster files with ranked pixel values scaled to befrom 0 to 1 values gt 07 were considered to be of high conservationimportance These important areas were then overlaid with thecurrent Protected Area network in Egypt to show if the network inEgypt is adequate to conserve the species in the face of climatechange and if there are areas outside the current Protected Areasboundaries that require special protection measures

3 Results

31 Model performance amp variable importance

In terms of mean AUC values for the 10 cross-validation runs theperformance of current-distribution models was good (seeTable S1) ranging from 078 to 099 with an overall mean of

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 215

093 plusmn 005 Only one species (Ptyodactylus siphonorhina) had amean AUC of less than 08 while in 36 species it was greater than095 As almost all species had a mean AUC gt08 all models wereaccepted and processed for further analyses More widespreadspecies tended to have lower AUC values (correlations [n frac14 75]with number of suitable pixels rs frac14 085 p lt 0005 extent ofoccurrence rs frac14 077 p lt 0005 number of pixels with recordsrs frac14 0433 p lt 0005)

Variables with highest mean permutation importance across allmodelled species were altitude (highest for 34 species) Bio4(temperature seasonality 10 species) and Bio13 (precipitation ofwettest month 9 species) Those with the lowest mean permuta-tion importance and thus not influencing the final models verymuch were the difference betweenmaximum andminimumNDVIBio15 (precipitation seasonality) habitat and Bio9 (mean temper-ature of driest quarter) the first and last of these were never themost influential variable for any species

32 Species richness

The patterns of current species richness (Fig 1) were similarusing either logistic or thresholded distributions both emphasizingthe lower Nile Valley the Suez Canal area and the coasts of theMediterranean and Suez Gulf (Fig S2 for locations map) The

Fig 1 Current estimated reptile species richness of Egypt using the sum of (a)Maxent logistic predictions and (b) thresholded predicted distributions

pattern is concordant with the species richness map in Baha El Din(2006) produced using records and qualitative criteria (seeFig S5a) Under climate change regardless of emission scenario ordispersal assumptions several areas are expected to have increasedspecies richness (Fig 2) with generally greater magnitudes underthe A2a scenario while others are expected to decline (eg the SuezCanal area) with greater declines under the B2a scenario by 2050and 2080

Areas with the highest expected loss of species include betweengreater Cairo Ismailia and Suez south of Suez on both sides of theSuez Gulf and Wadi El-Natrun much greater under the B2a sce-nario (Fig 3a) If species can disperse well then the highest gains inspecies are expected to include a large proportion of the northernhalf of the Western Desert again greater under the B2a scenario(Fig 3b) In terms of percentage change of the reptile assemblage ifdispersal is unlimited then the Western Desert is expected to un-dergo high turnover (Fig S6a) but this is because there are so fewspecies there currently Under no dispersal turnover is very limited(Fig S6b) again with highest values again in the Western desertbecause of the species poor assemblages there

33 Range and status changes

With unlimited dispersal no species is expected to becomeldquoextinctrdquo (100 loss of suitable habitat) under all or the average ofall GCMs There are a couple of species expected to become extinctby losing their entire area of suitable habitat in at least one of thefuture projections Tarentola mindiae and Hemidactylus robustusUsing the average gain or loss of suitable habitat across the fourdifferent GCMs eight species are expected to be classified as Crit-ically Endangered (loss gt 80) twenty as Endangered (loss50e80) and twenty-two as Vulnerable (loss 30e50) in at leastone mean future projection (details see Table S2)

When assuming no dispersal at all none of the species are ex-pected to lose all suitable habitats under all or the average of allGCMs but T mindiae and H robustus again experience completeloss of habitat under at least one of the projections Using theaverage loss of suitable habitat across the four different GCMs tenspecies are expected to be classified as Critically Endangeredtwenty-seven as Endangered and thirty-three as Vulnerable in atleast one mean future projection (details see Table S3)

34 Area prioritization for conservation

The areas with the highest current prioritization value for con-servation were similar for both cell-removal methods mainlyemphasizing the north coast Suez Canal area South Sinai theGebel Elba region Qattara Depression Wadi El-Natrun and aroundCairo amp Fayoum with the lsquoadditive benefitrsquo function putting greateremphasis on the Nile Delta and its fringing areas (Fig 4) Thisoverall pattern does not change much under future climate pro-jections though a few areas decline in priority and much greaterarea of the northern Western Desert increases These effects aremore pronounced with the lsquocore-area functionrsquo used as the cellremoval rule

The mean prioritization value in all models was higher in Pro-tected Areas than outside them (Fig 5) with both slightlyincreasing in future projections The difference in prioritizationvalue between inside and outside Protected Areas declined withtime especially by 2050 and 2080 under the B2a scenario

4 Discussion

The results of this study show for first time the potential impactsof climate change on the distribution of Egyptian reptiles Some

Fig 2 Estimated future changes in species richness as a result of anthropogenic climate change under two IPCC scenarios (A2a B2a) and at three times in the future (2020 20502080) Each map is the difference from the current species richness (shown in Fig 1a) and is averaged across four GCMmodels Maps made (a) using Maxent logistic output and (b)using thresholded distributions (see Methods)

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221216

reptiles are expected to lose much of their currently suitabilityareas (up to 80 in some cases) and some areas will lose or changea large proportion of their species Thus there will be a need forgreater conservation efforts in the future Some of these areas arealready inside Protected Areas and so hypothetically they are pro-tected while others are either just outside or very far from Pro-tected Areas Although overall Protected Areas contain better(higher priority) values compared to unprotected sites the currentProtected-Area network appears inadequate for future conserva-tion Our results together with others (eg Leach et al 2013) willhelp the Egyptian Environmental Affairs Agency and otherdecision-making parties in Egypt to direct conservation efforts andlimited conservation resources in the right direction

41 Model performance amp variable importance

The negative correlations found between AUC and the areaoccupied and with the extent of occurrence concur with findingsof many other studies (eg Brotons et al 2004 Elith et al 2006Hernandez et al 2006) Rare species are usually habitat special-ists with narrow environmental tolerances and are environmen-tally or geographically restricted compared to widespread species

(Elith et al 2006 Pandit et al 2009) Widespread species are likelyto be generalists occupying a wide range of habitats and climatesmaking it difficult to distinguish between suitable and unsuitablehabitats (Franklin et al 2009) However as pointed out by Elithet al (2006) this may be an artefact an inevitable result for spe-cies with small relative occurrence areas and perhaps a reason notto use AUC to compare model performances (Lobo et al 2008)Most (about 60) of the Egyptian herpetofauna are narrowlydistributed occupying less than 10 of Egypts area (Baha El Din2006) There is no evidence of a strong correlation betweenmodel accuracy and the number of records used (see Elith et al2006 Newbold et al 2009b) and the weak negative correlationfound here is unusual since most (eg Stockwell and Peterson2002 Hernandez et al 2006) find that models with larger sam-ple sizes have higher accuracy (but see de Pous et al 2011)

We found altitude to be the most effective predictor for manyspecies In contrast the NDVI predictors were not useful unlike inmany other cases (eg Egbert et al 2002 Anderson et al 2006Kgosiesele 2010 Taheri 2010) but perhaps typical for arid envi-ronments (eg Soultans (2011) study on Egyptian antelopes) or forherpetofauna (eg Teodoro et al 2013) Habitat categories likewisemade only a low contribution to the reptile models

Fig 3 Expected patterns of cumulative (a) losses and (b) gains of species as a result of climate change The losses of each species in each pixel (under both dispersal assumptions)and gains (under unlimited dispersal only) were averaged across GCM models accumulated across species and then plotted

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 217

42 Species richness

Species richness by itself is not adequate as an indicator of bioticchange since species composition can change independently Ourresults suggest that in the future some areas with currently highexpected species richness may lose many species or undergo largechanges in species composition Some of these areas are alreadyunder a certain degree of protection but even if they can bemaintained or enhanced (eg Siwa oasis Gebel Elba and the coastalareas over the Aqaba gulf Fig S2) current Protected Areas are notsufficient to prevent expected loss of species Other areas (eg WadiEl-Natrun the Suez Canal area Red Sea inland wadis the poorlyknown Qattara Depression and the areas around Gebel El-Gallalaand Gebel El-Hallal) are unprotected and measures are needed toconserve them There are similar situations elsewhere for examplehigh species loss and turnover in the reptiles of a priority conser-vation hotspot in Europe (the Iberian Peninsula) are expected (dePous 2011) with great variation in expected patterns of loss andchange with different projections and dispersal assumptions Otherstudies have also concluded that large changes are likely in chelo-nians worldwide (Ihlow et al 2012) Mediterranean mammals(Maiorano et al 2011) US trees and vertebrates (Currie 2001) andMesoAmerican trees (Golicher et al 2012)

43 Range and status changes

Two species are expected to lose their suitable habitats entirelyin at least one future projection 1) T mindiae is a near-endemicspecies recorded just from northwest Egypt and northern Cyrena-ica (eastern Libya) its distribution in Egypt is restricted to Siwaoasis the Qattara Depression and their periphery (Baha El Din2006) Future range changes suggest it will be Endangered (loss50e80) by 2020 and as Critically Endangered (loss gt80) by 2050and 2080 (under all emission scenarios and dispersal assumptions)2) H robustus is localized to the East African coast from Zanzibar tosouthern Egypt Arabia east to Pakistan its distribution in Egypt ison the Red Sea coast from El-Quseir southwards (Baha El Din2006) The A2a scenario suggests it will be Critically Endangeredby 2020 and 2050 while under the B2a scenario it will be Endan-gered by 2020 and Critically Endangered by 2050 and 2080 Morethan half of its current distribution in Egypt is located within theWadi El-Gemal and Gebel Elba Protected Areas so hypothetically itis protected However it is found mainly on buildings and henceassociated with man it may in fact be an invasive to Egypt

Another eight species are expected to be classified as CriticallyEndangered (loss gt80) in at least one averaged future projection(Acanthodactylus longipes Cerastes vipera Eryx jaculus Eumeces

Fig 4 Current conservation prioritizationl of areas according to the Zonation algo-rithm under two cell removal rules (a) core-area function (identifies areas where oneor more species have important occurrences) and (b) additive benefit function (whichgives greater emphasis to areas with high species richness) Darker colours representhigher conservation value (For interpretation of the references to colour in this figurelegend the reader is referred to the web version of this article)

Fig 5 Mean prioritization value for conservation (plusmn95 confidence limits) derivedfrom Zonation for Protected Areas (PAs) and outside Protected Areas (non-PAs) forcurrent and expected future conditions using the cell-removal rule of (a) core-areafunction and (b) additive benefit function

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221218

schneiderii Malpolon moilensis Ptyodactylus guttatusP siphonorhina and Trapelus mutabilis) According to the currentassessment twenty-seven species are classified as Endangered(loss 50e80) in at least one projection (mostly by 2080 in the A2scenario) two of these are classified assuming no dispersal asEndangered in all future projections (Uromastyx aegyptia and Pris-turus flavipunctatus for details see Tables S2eS3)

Reptiles are also likely to gain from global warming especially inmore northern areas of the globe if they can disperse and expandtheir distributions but under perhaps the more realistic assump-tion of no dispersal most species were expected to lose habitatsuitability by 2050 (Arauacutejo et al 2006) In Egypt areas of clearfuture expected gain (assuming unlimited dispersal) are the areabetween the Siwa oasis and Qattara Depression northwards almostto the northern coastal strip and some sparse locations in centralSinai There were greater gains under the B2a scenario in theeastern part of the Western Desert and some patchy areas aroundthe Nile (Fig 3b) This pattern is similar to that reported by Araujoet al (2006)

Some of the species identified here as being most at risk due tofuture climate change are in fact resilient species that are ecolog-ically very adaptable and are currently experiencing range expan-sion and population growth at least in Egypt (eg H robustus) This

disparity is probably because modelling is based solely on theprobable impacts of climate change on the spatial distribution ofsuitable habitats without taking evolution or ecological adapt-ability and resilience into account Species vary greatly in theirresponse to ecological change some adapting very well even tovery rapid anthropogenically induced changes (eg invasive spe-cies) while others are much less adaptable and tend to go extinctfirst

There are no published studies that discuss the effect of climatechange on Egyptian reptiles obviating any comparisons Informa-tion on the reptiles of the adjacent countries of Sudan and Libya arevery limited making it impossible to assess any movements thatmight compensate for Egyptian losses The patterns of expectedchanges in Leach et als (2013) similar study on mammals andbutterflies do not concur with ours Mammal species richness wasexpected to decline in the Mediterranean and Red Sea areas (by40e60) and increase elsewhere (by 80e100) while for butter-flies declines are expected over almost the whole of Egypt (by40e60) except the south which is expected to increase (by40e60) In the models of European reptiles (de Pous 2011)

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 219

between eight (unlimited dispersal) and 21 (no dispersal) specieswere expected to become extinct in at least one future projectionwith two more losing more than 80 of their currently suitablehabitats This is a much more damaging scenario than the expectedresponses of Egyptian reptiles in our study There are no othercomparable studies from North Africa or the Middle East

44 Area prioritization for conservation

In our study Protected Areas (Fig S3) were more valuable inconserving reptiles than areas outside them both currently and inthe future although the difference is expected to get smaller withtime (Fig 5) Although this is encouraging the Protected Areanetwork appears inadequate to conserve Egyptian reptiles NewProtected Areas that have already been proposed would go someway towards reducing this deficiency in coverage although otherswill be needed in the middle to north Sinai (especially aroundGebel El-Hallal) the Suez Canal area both sides of the Suez Gulf thenorthern Mediterranean coast coastal wadis of the Red Sea be-tween Hurghada and Mersa Alam and the Gebel El-Gallala areaMore effective protection is required in the established ProtectedAreas including those in South Sinai Gebel Elba and Siwa oasis dueto their relatively high expected future suitability loss The Pro-tected Areas of Siwa oasis and El-Gilf El-Kebir plus the areas ofWadiEl-Natrun the Gebel El-Hallal area and wadis of the Red Sea areexpected to have particularly high turnover in species composition(Fig S6) and hence probably need more conservation efforts Somecomparable information is available for Egypts butterflies andmammals (Gilbert and Zalat 2008 Basuony et al 2010 Leach et al2013) Egypts network of Protected Areas appears adequate toconserve Egyptian butterfly hotspots except for the Mediterraneancoast between Alexandria and Sallum (Gilbert and Zalat 2008) butinadequate to conserve important mammal areas with new Pro-tected Areas needed to be constructed in the lower Nile Valleyalong the north coast between Alexandria and Sallum top part ofthe Suez Gulf and perhaps the Qattara Depression (Basuony et al2010) Suggestions for conserving mammals are consistent to someextent with our results which also suggest new Protected Areasalong the Suez Gulf Mediterranean coast from Mersa Matruh toSallum and the Qattara Depression Leach et al (2013) concur withour results in suggesting priority areas along the western Medi-terranean coast around Greater Cairo central and north Sinai andthe Red Sea coast

45 Limitations

There are some limitations of the results shown in this study 1)Egyptian climate data in WorldClim are interpolated from ratherfew non-randomly placed weather stations in Egypt and adjacentcountries Although making the climate data less reliable thanelsewhere there are no alternative climate data available 2) Anensemble approach (eg using BIOMOD Thuiller et al 2009) isoften suggested for SDM modelling to avoid dependence on oneparticular algorithm This is not possible without a greater numberof records since other algorithms are not as flexible or robust asMaxent Despite all the caveats with using SDMs and criticisms ofMaxent in particular (Royle et al 2012 but see Hastie and Fithian2013 Elith et al 2006 Hernandez et al 2006) in practice coun-tries such as Egypt have little else to guide them in attempts atclimate change adaptation 3) There were signs of sampling bias inthe available records Although this violates the assumptions of allSDMs it is a problem common to nearly all datasets worldwideNew techniques may allow for the correction of sampling bias egusing target-group approach (Phillips et al 2009) although thedata requirements are currently impossible for Egypt and most

other developing countries) 4) Using Maxents default settings iscriticized by some authors they need to be modified dependent onthe data (Warren and Seifert 2011 Radosavljevic and Anderson2014 Merow et al 2013) but we think using the default settingsis enough in this study

5 Conclusion

Although Egyptian datasets exist knowledge of Egyptian spe-cies and conservation priorities are limited (with consequencesexplored in Leach et al 2013) nevertheless our results and those ofLeach et al (2013) encourage us to believe that useful insights canbe gained for conservation priorities by using the combination ofspecies distribution modelling plus spatial conservation prioriti-zation techniques (Hannah 2010 Klorvuttimontara et al 2011) Itis crucial for the Egyptian Environmental Affairs Agency to establishan accurate and reliable database of records assessed by experts tohelp improve the basic information for future conservation as-sessments and analyses

Although many SDM studies have been done on a variety oforganisms in recent decades aiming to solve conservation prob-lems evidence of their practical utility in real-world conservation issparse with very few studies clearly targeting conservation de-cisions (Guisan et al 2013) If implemented correctly SDMs canplay a critical role in supporting spatial conservation decision-making (Addison et al 2013) For reptiles there is a shortage ofstudies estimating their response to expected climate changes(Arauacutejo et al 2006) Although some suggest that reptiles (and to agreater extent amphibians) will be strongly affected by climatechange (eg Gibbons et al 2000 Arauacutejo et al 2006 Carvalho et al2010) very few of themwere concerned with arid environments orthe Middle East We hope that this and other similar studies thatestimate the fate of reptiles under climate change help decision-making authorities to take the measures necessary to conservethem

Acknowledgements

We thank the Italian Cooperation (Debt Swap) in Egypt forfunding the BioMAP project that allowed the collation of the da-tabases used in this work Dr Tim Newbold for allowing the use ofthe data from his field trips Dr Mustafa Fouda (Former Director ofthe Nature Conservation Sector EEAA) for facilities offered duringBioMAP project and all the BioMAP staff This work was supportedby the award of a Chevening Scholarship Masters grant to AEl-GWe are especially grateful to two anonymous referees whogreatly improved an earlier manuscript

Appendix A Supplementary data

Supplementary data related to this article can be found at httpdxdoiorg101016jjaridenv201512007

References

Addison PFE Rumpff L Bau SS Carey JM Chee YE Jarrad FC McBride MFBurgman MA 2013 Practical solutions for making models indispensable inconservation decision-making Divers Distrib 19 (5e6) 490e502 httpdxdoiorg101111ddi12054

Anderson RP Peterson AT Egbert SL 2006 Vegetation-index models predictareas vulnerable to purple loosestrife (Lythrum salicaria) invasion in KansasSouthwest Nat 51 471e480 httpdxdoiorg1018940038-4909(2006)51[471VMPAVT]20CO2

Arauacutejo MB 2009 Climate change and spatial conservation planning InMoilanen A Wilson KA Possingham HP (Eds) Spatial Conservation Prior-itization Quantitative Methods and Computational Tools Oxford UniversityPress pp 172e184 ISBN 9780199547760

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221220

Arauacutejo MB Rahbek C 2006 How does climate change affect biodiversity Sci-ence 313 1396e1397 httpdxdoiorg101126science1131758

Arauacutejo MB Thuiller W Pearson RG 2006 Climate warming and the decline ofamphibians and reptiles in Europe J Biogeogr 33 1712e1728 httpdxdoiorg101111j1365-2699200601482x

Baha El Din SM 2001 The Herpetofauna of Egypt Species Communities andAssemblages University of Nottingham p 435 Unpublished PhD thesis

Baha El Din SM 2006 A Guide to the Reptiles and Amphibians of Egypt TheAmerican University in Cairo Press ISBN 9789774249792

Baha El Din SM 2007 A new lizard of the Acanthodactylus scutellatus group(Squamata Lacertidae) from Egypt Zool Middle East 40 21e32 httpdxdoiorg10108009397140200710638200

Baldwin RA 2009 Use of maximum entropy modeling in wildlife research En-tropy 11 854e866 httpdxdoiorg103390e11040854

Ball IR Possingham HP Watts M 2009 Marxan and relatives software forspatial conservation prioritisation In Moilanen A Wilson KAPossingham HP (Eds) Spatial Conservation Prioritization QuantitativeMethods and Computational Tools Oxford University Press pp 185e195 ISBN9780199547760

Basuony MI Gilbert F Zalat S 2010 The Mammals of Egypt Atlas Red DataListing amp Conservation BioMAP amp CultNat EEAA amp Bibliotheca AlexandrinaCairo available at httpecologynottinghamacuk~plzfg

Bombi P Salvi D Bologna MA Pettorelli N 2012 Cross-scale predictions allowthe identification of local conservation priorities from atlas data Anim Conserv15 378e387 httpdxdoiorg101111j1469-1795201200526x

Brotons L Thuiller W Arauacutejo MB Hirzel AH 2004 Presence-absence versuspresence-only modelling methods for predicting bird habitat suitability Ecog-raphy 27 437e448 httpdxdoiorg101111j0906-7590200403764x

Cadby CD While GM Hobday AJ Uller T Wapstra E 2010 Multi-scaleapproach to understanding climate effects on offspring size at birth and date ofbirth in a reptile Integr Zool 5 164e175 httpdxdoiorg101111j1749-4877201000201x

Carvalho SB Brito JC Crespo EJ Possingham HP 2010 From climate changepredictions to actions e conserving vulnerable animal groups in hotspots at aregional scale Glob Change Biol 16 (12) 3257e3270 httpdxdoiorg101111j1365-2486201002212x

Currie DJ 2001 Projected effects of climate change on patterns of vertebrate andtree species richness in the conterminous United States Ecosystems 4216e225 httpdxdoiorg101007s10021-001-0005-4

de Pous P 2011 Climate Change and European Reptiles from Large-scale SpatialRange Shift Projections towards a More Realistic Assessment of Species Re-sponses University of East Anglia UK p 64 MSc thesis

de Pous P Beukema W Weterings M Duumlmmer I Geniez P 2011 Area priori-tization and performance evaluation of the conservation area network for theMoroccan herpetofauna a preliminary assessment Biodivers Conserv 2089e118 httpdxdoiorg101007s10531-010-9948-0

Edgar P Foster J Baker J 2010 Reptile Habitat Management HandbookAmphibian and Reptile Conservation Bournemouth

Egbert SL Martiacutenez-Meyer E Ortega-Huerta M Peterson AT 2002 Use ofdatasets derived from time-series AVHRR imagery as surrogates for land covermaps in predicting species distributions In Proceedings IEEE 2002 Interna-tional Geoscience and Remote Sensing Symposium (IGARSS) vol 4pp 2337e2339 httpdxdoiorg101109IGARSS20021026537

El Alqamy H Ismael A Abdelhameed A Nagy A Hamada A et al 2010 Pre-dicting the status and distribution of the Nubian Ibex (Capra nubiana) in thehigh-altitude mountains of south Sinai (Egypt) In Galemys Newsletter of theSpanish Society for the Preservation and Study of Mammals vol 22pp 517e530

Elith J Graham CH Anderson RP Dudiacutek M Ferrier S et al 2006 Novelmethods improve prediction of species distributions from occurrence dataEcography 29 129e151 httpdxdoiorg101111j20060906-759004596x

Elith J Phillips SJ Hastie T Dudiacutek M Chee YE Yates CJ 2011 A statisticalexplanation of MaxEnt for ecologists Divers Distrib 17 (1) 43e57 httpdxdoiorg101111j1472-4642201000725x

FAO 2012 Food and Agriculture Organisation World Development IndicatorsAverage Precipitation in Depth (mm per year) httpdataworldbankorgindicatorAGLNDPRCPMM

Franklin J 2009 Mapping Species Distributions Spatial Inference and PredictionCambridge University Press Cambridge UK ISBN 9780521700023

Franklin J Wejnert KE Hathaway SA Rochester CJ Fisher RN 2009 Effect ofspecies rarity on the accuracy of species distribution models for reptiles andamphibians in southern California Divers Distrib 15 167e177 httpdxdoiorg101111j1472-4642200800536x

Gibbons JW Scott DE Ryan TJ Buhlmann KA Tuberville TD Metts BSGreene JL Mills T et al 2000 The global decline of reptiles Deja Vu Am-phibians BioScience 50 (8) 653e666 httpdxdoiorg1016410006-3568(2000)050[0653TGDORD]20CO2

Gilbert F Zalat S 2008 The Butterflies of Egypt Atlas Red Data Listing amp Con-servation BioMAP EEAA Cairo available at httpecologynottinghamacuk~plzfg

Golicher DJ Cayuela L Newton AC 2012 Effects of climate change on the po-tential species richness of Mesoamerican forests Biotropica 44 284e293httpdxdoiorg101111j1744-7429201100815x

Graham CH Elith J Hijmans RJ Guisan A Peterson AT et al 2008 Theinfluence of spatial errors in species occurrence data used in distribution

models J Appl Ecol 45 239e247 httpdxdoiorg101111j1365-2664200701408x

Guisan A Tingley R Baumgartner JB Naujokaitis-Lewis I Sutcliffe PRTulloch AIT Regan TJ Brotons L et al 2013 Predicting species distributionsfor conservation decisions Ecol Lett 16 (12) 1424e1435 httpdxdoiorg101111ele12189

Hannah L 2010 A global conservation system for climate-change adaptationConserv Biol 24 70e77 httpdxdoiorg101111j1523-1739200901405x

Hannah L 2011 Climate Change Biology Elsevier Ltd ISBN 9780123741820Hannah L Lovejoy TE Schneider SH 2005 Biodiversity and climate change in

context In Lovejoy TE Hannah L (Eds) Climate Change and BiodiversityYale University Press New Haven CT USA

Hannah L Midgley G Andelman S Arauacutejo M Hughes G et al 2007 Protectedarea needs in a changing climate Front Ecol Environ 5 131e138 httpdxdoiorg1018901540-9295(2007)5[131PANIAC]20CO2

Hastie T Fithian W 2013 Inference from presence-only data the ongoing con-troversy Ecography 36 864e867 httpdxdoiorg101111j1600-0587201300321x

Hernandez PA Graham CH Master LL Albert DL 2006 The effect of samplesize and species characteristics on performance of different species distributionmodeling methods Ecography 29 773e785 httpdxdoiorg101111j0906-7590200604700x

Hijmans RJ Cameron SE Parra JL Jones PG Jarvis A 2005 Very high reso-lution interpolated climate surfaces for global land areas Int J Climatol 251965e1978 httpdxdoiorg101002Joc1276

Holt AC Salkeld DJ Fritz CL Tucker JR Gong P 2009 Spatial analysis ofplague in California niche modeling predictions of the current distribution andpotential response to climate change Int J Health Geogr 8 38 httpdxdoiorg1011861476-072X-8-38

Hoyle M James M 2005 Global warming human population pressure andviability of the worlds smallest butterfly Conserv Biol 19 1113e1124 httpdxdoiorg101111j1523-1739200500166x

Ihlow F Dambach J Engler JO Flecks M Hartmann T et al 2012 On the brinkof extinction How climate change may affect global chelonian species richnessand distribution Glob Change Biol 18 1520e1530 httpdxdoiorg101111j1365-2486201102623x

IPCC 2007 Climate Change 2007 Synthesis Report Contribution of WorkingGroups I II and III to the Fourth Assessment Report of the IntergovernmentalPanel on Climate Change IPCC Geneva Switzerland

Kgosiesele E 2010 Predictive Distribution Modelling of Timon Lepida in SpainUniversity of Twente (ITC) the Netherlands p 87 MSc thesis

Klorvuttimontara S McClean CJ Hill JK 2011 Evaluating the effectiveness ofprotected areas for conserving tropical forest butterflies of Thailand BiolConserv 144 2534e2540 httpdxdoiorg101016jbiocon201107012

Leach K Zalat S Gilbert F 2013 Egypts protected area network under futureclimate change Biol Conserv 159 490e500 httpdxdoiorg101016jbiocon201211025

Lobo JM Jimenez-Valverde A Real R 2008 AUC a misleading measure of theperformance of predictive distribution models Glob Ecol Biogeogr 17145e151 httpdxdoiorg101111j1466-8238200700358x

Maiorano L Falcucci A Zimmermann NE Psomas A Pottier J et al 2011 Thefuture of terrestrial mammals in the Mediterranean basin under climatechange Philosophical Trans R Soc Lond Ser B Biol Sci 366 2681e2692httpdxdoiorg101098rstb20110121

Marini MA Barbet-Massin M Lopes LE Jiguet F 2009 Predicted climate-driven bird distribution changes and forecasted conservation conflicts in aneotropical savanna Conserv Biol 23 1558e1567 httpdxdoiorg101111j1523-1739200901258x

Merow C Smith MJ Silander JA 2013 A practical guide to MaxEnt for modelingspecies distributions what it does and why inputs and settings matterEcography 36 (10) 1058e1069 httpdxdoiorg101111j1600-0587201307872x

Moilanen A Meller L Leppeuroanen J Pouzols FM Arponen A et al 2012 Zona-tion Spatial Conservation Planning Framework and Software v31 User Manualwwwhelsinkifibioscienceconsplan

Newbold T 2009 The Value of Species Distribution Models as a Tool for Conser-vation and Ecology in Egypt and Britain University of Nottingham UK p 312PhD thesis

Newbold T Gilbert F Zalat S El-Gabbas A Reader T 2009a Climate-basedmodels of spatial patterns of species richness in Egypts butterfly and mammalfauna J Biogeogr 36 2085e2095 httpdxdoiorg101111j1365-2699200902140x

Newbold T Reader T Zalat S El-Gabbas A Gilbert F 2009b Effect of charac-teristics of butterfly species on the accuracy of distribution models in an aridenvironment Biodivers Conserv 18 3629e3641 httpdxdoiorg101007s10531-009-9668-5

Pandit SN Kolasa J Cottenie K 2009 Contrasts between habitat generalists andspecialists an empirical extension to the basic metacommunity frameworkEcology 90 2253e2262 httpdxdoiorg10189008-08511

Pearson RG Dawson TP 2003 Predicting the impacts of climate change on thedistribution of species are bioclimate envelope models useful Glob EcolBiogeogr 12 361e371 httpdxdoiorg101046j1466-822X200300042x

Pearson RG Raxworthy CJ Nakamura M Peterson AT 2007 Predicting speciesdistributions from small numbers of occurrence records a test case using

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 221

cryptic geckos in Madagascar J Biogeogr 34 102e117 httpdxdoiorg101111j1365-2699200601594x

Phillips SJ ATampT Research 2011 A Brief Tutorial on Maxent Available at httpwwwcsprincetonedu~schapiremaxenttutorialtutorialdoc (Last accessed290512)

Phillips SJ Anderson RP Schapire RE 2006 Maximum entropy modeling ofspecies geographic distributions Ecol Model 190 231e259 httpdxdoiorg101016jecolmodel200503026

Phillips SJ Dudiacutek M Elith J Graham CH Lehmann A Leathwick J Ferrier S2009 Sample selection bias and presence-only distribution models implica-tions for background and pseudo-absence data Ecol Appl 19 181e197 httpdxdoiorg10189007-21531

R Development Core Team 2012 R a Language and Environment for StatisticalComputing R Foundation for Statistical Computing Vienna Austria ISBN 3-900051-07-0 httpwwwR-projectorg

Radosavljevic A Anderson RP 2014 Making better Maxent models of speciesdistributions complexity overfitting and evaluation J Biogeogr 41 (4)629e643 httpdxdoiorg101111jbi12227

Royle JA Chandler RB Yackulic C Nichols JD 2012 Likelihood analysis ofspecies occurrence probability from presence-only data for modelling speciesdistributions Methods Ecol Evol 3 545e554 httpdxdoiorg101111j2041-210X201100182x

Sauer J Domisch S Nowak C Haase P 2011 Lowmountain ranges summit trapsfor montane freshwater species under climate change Biodivers Conserv 203133e3146 httpdxdoiorg101007s10531-011-0140-y

Songer M Delion M Biggs A Huang Q 2012 Modeling impacts of climatechange on Giant Panda Habitat Int J Ecol 2012 1e12 httpdxdoiorg1011552012108752

Soultan AE 2011 Antelope distribution in Egypt Modeling the Impact of ClimateChange on Antelopes Distribution in Egypt Lambert Academic Publishing

Standards IUCN Subcommittee Petitions 2010 Guidelines for Using the IUCNRed List Categories and Criteria Version 81

Stockwell DRB Peterson AT 2002 Effects of sample size on accuracy of speciesdistribution models Ecol Model 148 1e13 httpdxdoiorg101016S0304-3800(01)00388-X

Suarez AV Tsutsui ND 2004 The value of museum collections for research andsociety Bioscience 54 66 httpdxdoiorg1016410006-3568(2004)054[0066tvomcf]20co2

Taheri S 2010 Hyper Temporal NDVI Images for Modelling and Prediction theHabitat Distribution of Balkan Green Lizard (Lacerta Trilineata) e Case Study

Crete (Greece) University of Twente (ITC) the Netherlands p 54 MScthesis

Taubmann J Theissinger K Feldheim KA Laube I Graf W et al 2011Modelling range shifts and assessing genetic diversity distribution of themontane aquatic mayfly Ameletus inopinatus in Europe under climate changescenarios Conserv Genet 12 503e515 httpdxdoiorg101007s10592-010-0157-x

Teodoro AC Sillero N Alves S Duarte L 2013 Correlation between the habitatsproductivity and species richness (amphibians and reptiles) in Portugal throughremote sensed data In Proc SPIE 8887 Remote Sensing for Agriculture Eco-systems and Hydrology httpdxdoiorg101117122028502

Thomas CD Cameron A Green RE Bakkenes M Beaumont LJ et al 2004Extinction risk from climate change Nature 427 145e148 httpdxdoiorg101038nature02121

Thuiller W 2004 Patterns and uncertainties of species range shifts under climatechange Glob Change Biol 10 2020e2027 httpdxdoiorg101111j1365-2486200400859x

Thuiller W 2007 Biodiversity climate change and the ecologist Nature 448550e552 httpdxdoiorg101038448550a

Thuiller W Lavorel S Arauacutejo MB Sykes MT Prentice IC 2005 Climate changethreats to plant diversity in Europe Proc Natl Acad Sci U S A 1028245e8250 httpdxdoiorg101073pnas0409902102

Thuiller W Lafourcade B Engler R Arauacutejo MB 2009 BIOMOD e a platform forensemble forecasting of species distributions Ecography 32 (3) 369e373httpdxdoiorg101111j1600-0587200805742x

Tolba MK Saab NW 2009 Arab environment Climate Change - Impact ofClimate Change on Arab Countries Arab Forum for Environment andDvelopment

Trotta-Moreu N Lobo JM 2010 Deriving the species richness distribution ofGeotrupinae (Coleoptera Scarabaeoidea) in Mexico from the overlap of indi-vidual model predictions Environ Entomol 39 42e49 httpdxdoiorg101603EN08179

Warren DL Seifert SN 2011 Ecological niche modeling in Maxent the impor-tance of model complexity and the performance of model selection criteriaEcol Appl 21 (2) 335e342 httpdxdoiorg10189010-11711

Wilson KA Cabeza M Klein CJ 2009 Fundamental concepts of spatial con-servation prioritisation In Moilanen A Wilson KA Possingham HP (Eds)Spatial Conservation Prioritization Quantitative Methods and ComputationalTools Oxford University Press pp 16e27 ISBN 9780199547760

Page 5: Journal of Arid Environments - elgabbas.github.io · e Faculty of Sciences & Arts, Taibah University, Al-Ula Campus, Saudi Arabia f Department of Zoology, Suez Canal University, Ismailia,

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 215

093 plusmn 005 Only one species (Ptyodactylus siphonorhina) had amean AUC of less than 08 while in 36 species it was greater than095 As almost all species had a mean AUC gt08 all models wereaccepted and processed for further analyses More widespreadspecies tended to have lower AUC values (correlations [n frac14 75]with number of suitable pixels rs frac14 085 p lt 0005 extent ofoccurrence rs frac14 077 p lt 0005 number of pixels with recordsrs frac14 0433 p lt 0005)

Variables with highest mean permutation importance across allmodelled species were altitude (highest for 34 species) Bio4(temperature seasonality 10 species) and Bio13 (precipitation ofwettest month 9 species) Those with the lowest mean permuta-tion importance and thus not influencing the final models verymuch were the difference betweenmaximum andminimumNDVIBio15 (precipitation seasonality) habitat and Bio9 (mean temper-ature of driest quarter) the first and last of these were never themost influential variable for any species

32 Species richness

The patterns of current species richness (Fig 1) were similarusing either logistic or thresholded distributions both emphasizingthe lower Nile Valley the Suez Canal area and the coasts of theMediterranean and Suez Gulf (Fig S2 for locations map) The

Fig 1 Current estimated reptile species richness of Egypt using the sum of (a)Maxent logistic predictions and (b) thresholded predicted distributions

pattern is concordant with the species richness map in Baha El Din(2006) produced using records and qualitative criteria (seeFig S5a) Under climate change regardless of emission scenario ordispersal assumptions several areas are expected to have increasedspecies richness (Fig 2) with generally greater magnitudes underthe A2a scenario while others are expected to decline (eg the SuezCanal area) with greater declines under the B2a scenario by 2050and 2080

Areas with the highest expected loss of species include betweengreater Cairo Ismailia and Suez south of Suez on both sides of theSuez Gulf and Wadi El-Natrun much greater under the B2a sce-nario (Fig 3a) If species can disperse well then the highest gains inspecies are expected to include a large proportion of the northernhalf of the Western Desert again greater under the B2a scenario(Fig 3b) In terms of percentage change of the reptile assemblage ifdispersal is unlimited then the Western Desert is expected to un-dergo high turnover (Fig S6a) but this is because there are so fewspecies there currently Under no dispersal turnover is very limited(Fig S6b) again with highest values again in the Western desertbecause of the species poor assemblages there

33 Range and status changes

With unlimited dispersal no species is expected to becomeldquoextinctrdquo (100 loss of suitable habitat) under all or the average ofall GCMs There are a couple of species expected to become extinctby losing their entire area of suitable habitat in at least one of thefuture projections Tarentola mindiae and Hemidactylus robustusUsing the average gain or loss of suitable habitat across the fourdifferent GCMs eight species are expected to be classified as Crit-ically Endangered (loss gt 80) twenty as Endangered (loss50e80) and twenty-two as Vulnerable (loss 30e50) in at leastone mean future projection (details see Table S2)

When assuming no dispersal at all none of the species are ex-pected to lose all suitable habitats under all or the average of allGCMs but T mindiae and H robustus again experience completeloss of habitat under at least one of the projections Using theaverage loss of suitable habitat across the four different GCMs tenspecies are expected to be classified as Critically Endangeredtwenty-seven as Endangered and thirty-three as Vulnerable in atleast one mean future projection (details see Table S3)

34 Area prioritization for conservation

The areas with the highest current prioritization value for con-servation were similar for both cell-removal methods mainlyemphasizing the north coast Suez Canal area South Sinai theGebel Elba region Qattara Depression Wadi El-Natrun and aroundCairo amp Fayoum with the lsquoadditive benefitrsquo function putting greateremphasis on the Nile Delta and its fringing areas (Fig 4) Thisoverall pattern does not change much under future climate pro-jections though a few areas decline in priority and much greaterarea of the northern Western Desert increases These effects aremore pronounced with the lsquocore-area functionrsquo used as the cellremoval rule

The mean prioritization value in all models was higher in Pro-tected Areas than outside them (Fig 5) with both slightlyincreasing in future projections The difference in prioritizationvalue between inside and outside Protected Areas declined withtime especially by 2050 and 2080 under the B2a scenario

4 Discussion

The results of this study show for first time the potential impactsof climate change on the distribution of Egyptian reptiles Some

Fig 2 Estimated future changes in species richness as a result of anthropogenic climate change under two IPCC scenarios (A2a B2a) and at three times in the future (2020 20502080) Each map is the difference from the current species richness (shown in Fig 1a) and is averaged across four GCMmodels Maps made (a) using Maxent logistic output and (b)using thresholded distributions (see Methods)

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221216

reptiles are expected to lose much of their currently suitabilityareas (up to 80 in some cases) and some areas will lose or changea large proportion of their species Thus there will be a need forgreater conservation efforts in the future Some of these areas arealready inside Protected Areas and so hypothetically they are pro-tected while others are either just outside or very far from Pro-tected Areas Although overall Protected Areas contain better(higher priority) values compared to unprotected sites the currentProtected-Area network appears inadequate for future conserva-tion Our results together with others (eg Leach et al 2013) willhelp the Egyptian Environmental Affairs Agency and otherdecision-making parties in Egypt to direct conservation efforts andlimited conservation resources in the right direction

41 Model performance amp variable importance

The negative correlations found between AUC and the areaoccupied and with the extent of occurrence concur with findingsof many other studies (eg Brotons et al 2004 Elith et al 2006Hernandez et al 2006) Rare species are usually habitat special-ists with narrow environmental tolerances and are environmen-tally or geographically restricted compared to widespread species

(Elith et al 2006 Pandit et al 2009) Widespread species are likelyto be generalists occupying a wide range of habitats and climatesmaking it difficult to distinguish between suitable and unsuitablehabitats (Franklin et al 2009) However as pointed out by Elithet al (2006) this may be an artefact an inevitable result for spe-cies with small relative occurrence areas and perhaps a reason notto use AUC to compare model performances (Lobo et al 2008)Most (about 60) of the Egyptian herpetofauna are narrowlydistributed occupying less than 10 of Egypts area (Baha El Din2006) There is no evidence of a strong correlation betweenmodel accuracy and the number of records used (see Elith et al2006 Newbold et al 2009b) and the weak negative correlationfound here is unusual since most (eg Stockwell and Peterson2002 Hernandez et al 2006) find that models with larger sam-ple sizes have higher accuracy (but see de Pous et al 2011)

We found altitude to be the most effective predictor for manyspecies In contrast the NDVI predictors were not useful unlike inmany other cases (eg Egbert et al 2002 Anderson et al 2006Kgosiesele 2010 Taheri 2010) but perhaps typical for arid envi-ronments (eg Soultans (2011) study on Egyptian antelopes) or forherpetofauna (eg Teodoro et al 2013) Habitat categories likewisemade only a low contribution to the reptile models

Fig 3 Expected patterns of cumulative (a) losses and (b) gains of species as a result of climate change The losses of each species in each pixel (under both dispersal assumptions)and gains (under unlimited dispersal only) were averaged across GCM models accumulated across species and then plotted

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 217

42 Species richness

Species richness by itself is not adequate as an indicator of bioticchange since species composition can change independently Ourresults suggest that in the future some areas with currently highexpected species richness may lose many species or undergo largechanges in species composition Some of these areas are alreadyunder a certain degree of protection but even if they can bemaintained or enhanced (eg Siwa oasis Gebel Elba and the coastalareas over the Aqaba gulf Fig S2) current Protected Areas are notsufficient to prevent expected loss of species Other areas (eg WadiEl-Natrun the Suez Canal area Red Sea inland wadis the poorlyknown Qattara Depression and the areas around Gebel El-Gallalaand Gebel El-Hallal) are unprotected and measures are needed toconserve them There are similar situations elsewhere for examplehigh species loss and turnover in the reptiles of a priority conser-vation hotspot in Europe (the Iberian Peninsula) are expected (dePous 2011) with great variation in expected patterns of loss andchange with different projections and dispersal assumptions Otherstudies have also concluded that large changes are likely in chelo-nians worldwide (Ihlow et al 2012) Mediterranean mammals(Maiorano et al 2011) US trees and vertebrates (Currie 2001) andMesoAmerican trees (Golicher et al 2012)

43 Range and status changes

Two species are expected to lose their suitable habitats entirelyin at least one future projection 1) T mindiae is a near-endemicspecies recorded just from northwest Egypt and northern Cyrena-ica (eastern Libya) its distribution in Egypt is restricted to Siwaoasis the Qattara Depression and their periphery (Baha El Din2006) Future range changes suggest it will be Endangered (loss50e80) by 2020 and as Critically Endangered (loss gt80) by 2050and 2080 (under all emission scenarios and dispersal assumptions)2) H robustus is localized to the East African coast from Zanzibar tosouthern Egypt Arabia east to Pakistan its distribution in Egypt ison the Red Sea coast from El-Quseir southwards (Baha El Din2006) The A2a scenario suggests it will be Critically Endangeredby 2020 and 2050 while under the B2a scenario it will be Endan-gered by 2020 and Critically Endangered by 2050 and 2080 Morethan half of its current distribution in Egypt is located within theWadi El-Gemal and Gebel Elba Protected Areas so hypothetically itis protected However it is found mainly on buildings and henceassociated with man it may in fact be an invasive to Egypt

Another eight species are expected to be classified as CriticallyEndangered (loss gt80) in at least one averaged future projection(Acanthodactylus longipes Cerastes vipera Eryx jaculus Eumeces

Fig 4 Current conservation prioritizationl of areas according to the Zonation algo-rithm under two cell removal rules (a) core-area function (identifies areas where oneor more species have important occurrences) and (b) additive benefit function (whichgives greater emphasis to areas with high species richness) Darker colours representhigher conservation value (For interpretation of the references to colour in this figurelegend the reader is referred to the web version of this article)

Fig 5 Mean prioritization value for conservation (plusmn95 confidence limits) derivedfrom Zonation for Protected Areas (PAs) and outside Protected Areas (non-PAs) forcurrent and expected future conditions using the cell-removal rule of (a) core-areafunction and (b) additive benefit function

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221218

schneiderii Malpolon moilensis Ptyodactylus guttatusP siphonorhina and Trapelus mutabilis) According to the currentassessment twenty-seven species are classified as Endangered(loss 50e80) in at least one projection (mostly by 2080 in the A2scenario) two of these are classified assuming no dispersal asEndangered in all future projections (Uromastyx aegyptia and Pris-turus flavipunctatus for details see Tables S2eS3)

Reptiles are also likely to gain from global warming especially inmore northern areas of the globe if they can disperse and expandtheir distributions but under perhaps the more realistic assump-tion of no dispersal most species were expected to lose habitatsuitability by 2050 (Arauacutejo et al 2006) In Egypt areas of clearfuture expected gain (assuming unlimited dispersal) are the areabetween the Siwa oasis and Qattara Depression northwards almostto the northern coastal strip and some sparse locations in centralSinai There were greater gains under the B2a scenario in theeastern part of the Western Desert and some patchy areas aroundthe Nile (Fig 3b) This pattern is similar to that reported by Araujoet al (2006)

Some of the species identified here as being most at risk due tofuture climate change are in fact resilient species that are ecolog-ically very adaptable and are currently experiencing range expan-sion and population growth at least in Egypt (eg H robustus) This

disparity is probably because modelling is based solely on theprobable impacts of climate change on the spatial distribution ofsuitable habitats without taking evolution or ecological adapt-ability and resilience into account Species vary greatly in theirresponse to ecological change some adapting very well even tovery rapid anthropogenically induced changes (eg invasive spe-cies) while others are much less adaptable and tend to go extinctfirst

There are no published studies that discuss the effect of climatechange on Egyptian reptiles obviating any comparisons Informa-tion on the reptiles of the adjacent countries of Sudan and Libya arevery limited making it impossible to assess any movements thatmight compensate for Egyptian losses The patterns of expectedchanges in Leach et als (2013) similar study on mammals andbutterflies do not concur with ours Mammal species richness wasexpected to decline in the Mediterranean and Red Sea areas (by40e60) and increase elsewhere (by 80e100) while for butter-flies declines are expected over almost the whole of Egypt (by40e60) except the south which is expected to increase (by40e60) In the models of European reptiles (de Pous 2011)

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 219

between eight (unlimited dispersal) and 21 (no dispersal) specieswere expected to become extinct in at least one future projectionwith two more losing more than 80 of their currently suitablehabitats This is a much more damaging scenario than the expectedresponses of Egyptian reptiles in our study There are no othercomparable studies from North Africa or the Middle East

44 Area prioritization for conservation

In our study Protected Areas (Fig S3) were more valuable inconserving reptiles than areas outside them both currently and inthe future although the difference is expected to get smaller withtime (Fig 5) Although this is encouraging the Protected Areanetwork appears inadequate to conserve Egyptian reptiles NewProtected Areas that have already been proposed would go someway towards reducing this deficiency in coverage although otherswill be needed in the middle to north Sinai (especially aroundGebel El-Hallal) the Suez Canal area both sides of the Suez Gulf thenorthern Mediterranean coast coastal wadis of the Red Sea be-tween Hurghada and Mersa Alam and the Gebel El-Gallala areaMore effective protection is required in the established ProtectedAreas including those in South Sinai Gebel Elba and Siwa oasis dueto their relatively high expected future suitability loss The Pro-tected Areas of Siwa oasis and El-Gilf El-Kebir plus the areas ofWadiEl-Natrun the Gebel El-Hallal area and wadis of the Red Sea areexpected to have particularly high turnover in species composition(Fig S6) and hence probably need more conservation efforts Somecomparable information is available for Egypts butterflies andmammals (Gilbert and Zalat 2008 Basuony et al 2010 Leach et al2013) Egypts network of Protected Areas appears adequate toconserve Egyptian butterfly hotspots except for the Mediterraneancoast between Alexandria and Sallum (Gilbert and Zalat 2008) butinadequate to conserve important mammal areas with new Pro-tected Areas needed to be constructed in the lower Nile Valleyalong the north coast between Alexandria and Sallum top part ofthe Suez Gulf and perhaps the Qattara Depression (Basuony et al2010) Suggestions for conserving mammals are consistent to someextent with our results which also suggest new Protected Areasalong the Suez Gulf Mediterranean coast from Mersa Matruh toSallum and the Qattara Depression Leach et al (2013) concur withour results in suggesting priority areas along the western Medi-terranean coast around Greater Cairo central and north Sinai andthe Red Sea coast

45 Limitations

There are some limitations of the results shown in this study 1)Egyptian climate data in WorldClim are interpolated from ratherfew non-randomly placed weather stations in Egypt and adjacentcountries Although making the climate data less reliable thanelsewhere there are no alternative climate data available 2) Anensemble approach (eg using BIOMOD Thuiller et al 2009) isoften suggested for SDM modelling to avoid dependence on oneparticular algorithm This is not possible without a greater numberof records since other algorithms are not as flexible or robust asMaxent Despite all the caveats with using SDMs and criticisms ofMaxent in particular (Royle et al 2012 but see Hastie and Fithian2013 Elith et al 2006 Hernandez et al 2006) in practice coun-tries such as Egypt have little else to guide them in attempts atclimate change adaptation 3) There were signs of sampling bias inthe available records Although this violates the assumptions of allSDMs it is a problem common to nearly all datasets worldwideNew techniques may allow for the correction of sampling bias egusing target-group approach (Phillips et al 2009) although thedata requirements are currently impossible for Egypt and most

other developing countries) 4) Using Maxents default settings iscriticized by some authors they need to be modified dependent onthe data (Warren and Seifert 2011 Radosavljevic and Anderson2014 Merow et al 2013) but we think using the default settingsis enough in this study

5 Conclusion

Although Egyptian datasets exist knowledge of Egyptian spe-cies and conservation priorities are limited (with consequencesexplored in Leach et al 2013) nevertheless our results and those ofLeach et al (2013) encourage us to believe that useful insights canbe gained for conservation priorities by using the combination ofspecies distribution modelling plus spatial conservation prioriti-zation techniques (Hannah 2010 Klorvuttimontara et al 2011) Itis crucial for the Egyptian Environmental Affairs Agency to establishan accurate and reliable database of records assessed by experts tohelp improve the basic information for future conservation as-sessments and analyses

Although many SDM studies have been done on a variety oforganisms in recent decades aiming to solve conservation prob-lems evidence of their practical utility in real-world conservation issparse with very few studies clearly targeting conservation de-cisions (Guisan et al 2013) If implemented correctly SDMs canplay a critical role in supporting spatial conservation decision-making (Addison et al 2013) For reptiles there is a shortage ofstudies estimating their response to expected climate changes(Arauacutejo et al 2006) Although some suggest that reptiles (and to agreater extent amphibians) will be strongly affected by climatechange (eg Gibbons et al 2000 Arauacutejo et al 2006 Carvalho et al2010) very few of themwere concerned with arid environments orthe Middle East We hope that this and other similar studies thatestimate the fate of reptiles under climate change help decision-making authorities to take the measures necessary to conservethem

Acknowledgements

We thank the Italian Cooperation (Debt Swap) in Egypt forfunding the BioMAP project that allowed the collation of the da-tabases used in this work Dr Tim Newbold for allowing the use ofthe data from his field trips Dr Mustafa Fouda (Former Director ofthe Nature Conservation Sector EEAA) for facilities offered duringBioMAP project and all the BioMAP staff This work was supportedby the award of a Chevening Scholarship Masters grant to AEl-GWe are especially grateful to two anonymous referees whogreatly improved an earlier manuscript

Appendix A Supplementary data

Supplementary data related to this article can be found at httpdxdoiorg101016jjaridenv201512007

References

Addison PFE Rumpff L Bau SS Carey JM Chee YE Jarrad FC McBride MFBurgman MA 2013 Practical solutions for making models indispensable inconservation decision-making Divers Distrib 19 (5e6) 490e502 httpdxdoiorg101111ddi12054

Anderson RP Peterson AT Egbert SL 2006 Vegetation-index models predictareas vulnerable to purple loosestrife (Lythrum salicaria) invasion in KansasSouthwest Nat 51 471e480 httpdxdoiorg1018940038-4909(2006)51[471VMPAVT]20CO2

Arauacutejo MB 2009 Climate change and spatial conservation planning InMoilanen A Wilson KA Possingham HP (Eds) Spatial Conservation Prior-itization Quantitative Methods and Computational Tools Oxford UniversityPress pp 172e184 ISBN 9780199547760

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221220

Arauacutejo MB Rahbek C 2006 How does climate change affect biodiversity Sci-ence 313 1396e1397 httpdxdoiorg101126science1131758

Arauacutejo MB Thuiller W Pearson RG 2006 Climate warming and the decline ofamphibians and reptiles in Europe J Biogeogr 33 1712e1728 httpdxdoiorg101111j1365-2699200601482x

Baha El Din SM 2001 The Herpetofauna of Egypt Species Communities andAssemblages University of Nottingham p 435 Unpublished PhD thesis

Baha El Din SM 2006 A Guide to the Reptiles and Amphibians of Egypt TheAmerican University in Cairo Press ISBN 9789774249792

Baha El Din SM 2007 A new lizard of the Acanthodactylus scutellatus group(Squamata Lacertidae) from Egypt Zool Middle East 40 21e32 httpdxdoiorg10108009397140200710638200

Baldwin RA 2009 Use of maximum entropy modeling in wildlife research En-tropy 11 854e866 httpdxdoiorg103390e11040854

Ball IR Possingham HP Watts M 2009 Marxan and relatives software forspatial conservation prioritisation In Moilanen A Wilson KAPossingham HP (Eds) Spatial Conservation Prioritization QuantitativeMethods and Computational Tools Oxford University Press pp 185e195 ISBN9780199547760

Basuony MI Gilbert F Zalat S 2010 The Mammals of Egypt Atlas Red DataListing amp Conservation BioMAP amp CultNat EEAA amp Bibliotheca AlexandrinaCairo available at httpecologynottinghamacuk~plzfg

Bombi P Salvi D Bologna MA Pettorelli N 2012 Cross-scale predictions allowthe identification of local conservation priorities from atlas data Anim Conserv15 378e387 httpdxdoiorg101111j1469-1795201200526x

Brotons L Thuiller W Arauacutejo MB Hirzel AH 2004 Presence-absence versuspresence-only modelling methods for predicting bird habitat suitability Ecog-raphy 27 437e448 httpdxdoiorg101111j0906-7590200403764x

Cadby CD While GM Hobday AJ Uller T Wapstra E 2010 Multi-scaleapproach to understanding climate effects on offspring size at birth and date ofbirth in a reptile Integr Zool 5 164e175 httpdxdoiorg101111j1749-4877201000201x

Carvalho SB Brito JC Crespo EJ Possingham HP 2010 From climate changepredictions to actions e conserving vulnerable animal groups in hotspots at aregional scale Glob Change Biol 16 (12) 3257e3270 httpdxdoiorg101111j1365-2486201002212x

Currie DJ 2001 Projected effects of climate change on patterns of vertebrate andtree species richness in the conterminous United States Ecosystems 4216e225 httpdxdoiorg101007s10021-001-0005-4

de Pous P 2011 Climate Change and European Reptiles from Large-scale SpatialRange Shift Projections towards a More Realistic Assessment of Species Re-sponses University of East Anglia UK p 64 MSc thesis

de Pous P Beukema W Weterings M Duumlmmer I Geniez P 2011 Area priori-tization and performance evaluation of the conservation area network for theMoroccan herpetofauna a preliminary assessment Biodivers Conserv 2089e118 httpdxdoiorg101007s10531-010-9948-0

Edgar P Foster J Baker J 2010 Reptile Habitat Management HandbookAmphibian and Reptile Conservation Bournemouth

Egbert SL Martiacutenez-Meyer E Ortega-Huerta M Peterson AT 2002 Use ofdatasets derived from time-series AVHRR imagery as surrogates for land covermaps in predicting species distributions In Proceedings IEEE 2002 Interna-tional Geoscience and Remote Sensing Symposium (IGARSS) vol 4pp 2337e2339 httpdxdoiorg101109IGARSS20021026537

El Alqamy H Ismael A Abdelhameed A Nagy A Hamada A et al 2010 Pre-dicting the status and distribution of the Nubian Ibex (Capra nubiana) in thehigh-altitude mountains of south Sinai (Egypt) In Galemys Newsletter of theSpanish Society for the Preservation and Study of Mammals vol 22pp 517e530

Elith J Graham CH Anderson RP Dudiacutek M Ferrier S et al 2006 Novelmethods improve prediction of species distributions from occurrence dataEcography 29 129e151 httpdxdoiorg101111j20060906-759004596x

Elith J Phillips SJ Hastie T Dudiacutek M Chee YE Yates CJ 2011 A statisticalexplanation of MaxEnt for ecologists Divers Distrib 17 (1) 43e57 httpdxdoiorg101111j1472-4642201000725x

FAO 2012 Food and Agriculture Organisation World Development IndicatorsAverage Precipitation in Depth (mm per year) httpdataworldbankorgindicatorAGLNDPRCPMM

Franklin J 2009 Mapping Species Distributions Spatial Inference and PredictionCambridge University Press Cambridge UK ISBN 9780521700023

Franklin J Wejnert KE Hathaway SA Rochester CJ Fisher RN 2009 Effect ofspecies rarity on the accuracy of species distribution models for reptiles andamphibians in southern California Divers Distrib 15 167e177 httpdxdoiorg101111j1472-4642200800536x

Gibbons JW Scott DE Ryan TJ Buhlmann KA Tuberville TD Metts BSGreene JL Mills T et al 2000 The global decline of reptiles Deja Vu Am-phibians BioScience 50 (8) 653e666 httpdxdoiorg1016410006-3568(2000)050[0653TGDORD]20CO2

Gilbert F Zalat S 2008 The Butterflies of Egypt Atlas Red Data Listing amp Con-servation BioMAP EEAA Cairo available at httpecologynottinghamacuk~plzfg

Golicher DJ Cayuela L Newton AC 2012 Effects of climate change on the po-tential species richness of Mesoamerican forests Biotropica 44 284e293httpdxdoiorg101111j1744-7429201100815x

Graham CH Elith J Hijmans RJ Guisan A Peterson AT et al 2008 Theinfluence of spatial errors in species occurrence data used in distribution

models J Appl Ecol 45 239e247 httpdxdoiorg101111j1365-2664200701408x

Guisan A Tingley R Baumgartner JB Naujokaitis-Lewis I Sutcliffe PRTulloch AIT Regan TJ Brotons L et al 2013 Predicting species distributionsfor conservation decisions Ecol Lett 16 (12) 1424e1435 httpdxdoiorg101111ele12189

Hannah L 2010 A global conservation system for climate-change adaptationConserv Biol 24 70e77 httpdxdoiorg101111j1523-1739200901405x

Hannah L 2011 Climate Change Biology Elsevier Ltd ISBN 9780123741820Hannah L Lovejoy TE Schneider SH 2005 Biodiversity and climate change in

context In Lovejoy TE Hannah L (Eds) Climate Change and BiodiversityYale University Press New Haven CT USA

Hannah L Midgley G Andelman S Arauacutejo M Hughes G et al 2007 Protectedarea needs in a changing climate Front Ecol Environ 5 131e138 httpdxdoiorg1018901540-9295(2007)5[131PANIAC]20CO2

Hastie T Fithian W 2013 Inference from presence-only data the ongoing con-troversy Ecography 36 864e867 httpdxdoiorg101111j1600-0587201300321x

Hernandez PA Graham CH Master LL Albert DL 2006 The effect of samplesize and species characteristics on performance of different species distributionmodeling methods Ecography 29 773e785 httpdxdoiorg101111j0906-7590200604700x

Hijmans RJ Cameron SE Parra JL Jones PG Jarvis A 2005 Very high reso-lution interpolated climate surfaces for global land areas Int J Climatol 251965e1978 httpdxdoiorg101002Joc1276

Holt AC Salkeld DJ Fritz CL Tucker JR Gong P 2009 Spatial analysis ofplague in California niche modeling predictions of the current distribution andpotential response to climate change Int J Health Geogr 8 38 httpdxdoiorg1011861476-072X-8-38

Hoyle M James M 2005 Global warming human population pressure andviability of the worlds smallest butterfly Conserv Biol 19 1113e1124 httpdxdoiorg101111j1523-1739200500166x

Ihlow F Dambach J Engler JO Flecks M Hartmann T et al 2012 On the brinkof extinction How climate change may affect global chelonian species richnessand distribution Glob Change Biol 18 1520e1530 httpdxdoiorg101111j1365-2486201102623x

IPCC 2007 Climate Change 2007 Synthesis Report Contribution of WorkingGroups I II and III to the Fourth Assessment Report of the IntergovernmentalPanel on Climate Change IPCC Geneva Switzerland

Kgosiesele E 2010 Predictive Distribution Modelling of Timon Lepida in SpainUniversity of Twente (ITC) the Netherlands p 87 MSc thesis

Klorvuttimontara S McClean CJ Hill JK 2011 Evaluating the effectiveness ofprotected areas for conserving tropical forest butterflies of Thailand BiolConserv 144 2534e2540 httpdxdoiorg101016jbiocon201107012

Leach K Zalat S Gilbert F 2013 Egypts protected area network under futureclimate change Biol Conserv 159 490e500 httpdxdoiorg101016jbiocon201211025

Lobo JM Jimenez-Valverde A Real R 2008 AUC a misleading measure of theperformance of predictive distribution models Glob Ecol Biogeogr 17145e151 httpdxdoiorg101111j1466-8238200700358x

Maiorano L Falcucci A Zimmermann NE Psomas A Pottier J et al 2011 Thefuture of terrestrial mammals in the Mediterranean basin under climatechange Philosophical Trans R Soc Lond Ser B Biol Sci 366 2681e2692httpdxdoiorg101098rstb20110121

Marini MA Barbet-Massin M Lopes LE Jiguet F 2009 Predicted climate-driven bird distribution changes and forecasted conservation conflicts in aneotropical savanna Conserv Biol 23 1558e1567 httpdxdoiorg101111j1523-1739200901258x

Merow C Smith MJ Silander JA 2013 A practical guide to MaxEnt for modelingspecies distributions what it does and why inputs and settings matterEcography 36 (10) 1058e1069 httpdxdoiorg101111j1600-0587201307872x

Moilanen A Meller L Leppeuroanen J Pouzols FM Arponen A et al 2012 Zona-tion Spatial Conservation Planning Framework and Software v31 User Manualwwwhelsinkifibioscienceconsplan

Newbold T 2009 The Value of Species Distribution Models as a Tool for Conser-vation and Ecology in Egypt and Britain University of Nottingham UK p 312PhD thesis

Newbold T Gilbert F Zalat S El-Gabbas A Reader T 2009a Climate-basedmodels of spatial patterns of species richness in Egypts butterfly and mammalfauna J Biogeogr 36 2085e2095 httpdxdoiorg101111j1365-2699200902140x

Newbold T Reader T Zalat S El-Gabbas A Gilbert F 2009b Effect of charac-teristics of butterfly species on the accuracy of distribution models in an aridenvironment Biodivers Conserv 18 3629e3641 httpdxdoiorg101007s10531-009-9668-5

Pandit SN Kolasa J Cottenie K 2009 Contrasts between habitat generalists andspecialists an empirical extension to the basic metacommunity frameworkEcology 90 2253e2262 httpdxdoiorg10189008-08511

Pearson RG Dawson TP 2003 Predicting the impacts of climate change on thedistribution of species are bioclimate envelope models useful Glob EcolBiogeogr 12 361e371 httpdxdoiorg101046j1466-822X200300042x

Pearson RG Raxworthy CJ Nakamura M Peterson AT 2007 Predicting speciesdistributions from small numbers of occurrence records a test case using

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 221

cryptic geckos in Madagascar J Biogeogr 34 102e117 httpdxdoiorg101111j1365-2699200601594x

Phillips SJ ATampT Research 2011 A Brief Tutorial on Maxent Available at httpwwwcsprincetonedu~schapiremaxenttutorialtutorialdoc (Last accessed290512)

Phillips SJ Anderson RP Schapire RE 2006 Maximum entropy modeling ofspecies geographic distributions Ecol Model 190 231e259 httpdxdoiorg101016jecolmodel200503026

Phillips SJ Dudiacutek M Elith J Graham CH Lehmann A Leathwick J Ferrier S2009 Sample selection bias and presence-only distribution models implica-tions for background and pseudo-absence data Ecol Appl 19 181e197 httpdxdoiorg10189007-21531

R Development Core Team 2012 R a Language and Environment for StatisticalComputing R Foundation for Statistical Computing Vienna Austria ISBN 3-900051-07-0 httpwwwR-projectorg

Radosavljevic A Anderson RP 2014 Making better Maxent models of speciesdistributions complexity overfitting and evaluation J Biogeogr 41 (4)629e643 httpdxdoiorg101111jbi12227

Royle JA Chandler RB Yackulic C Nichols JD 2012 Likelihood analysis ofspecies occurrence probability from presence-only data for modelling speciesdistributions Methods Ecol Evol 3 545e554 httpdxdoiorg101111j2041-210X201100182x

Sauer J Domisch S Nowak C Haase P 2011 Lowmountain ranges summit trapsfor montane freshwater species under climate change Biodivers Conserv 203133e3146 httpdxdoiorg101007s10531-011-0140-y

Songer M Delion M Biggs A Huang Q 2012 Modeling impacts of climatechange on Giant Panda Habitat Int J Ecol 2012 1e12 httpdxdoiorg1011552012108752

Soultan AE 2011 Antelope distribution in Egypt Modeling the Impact of ClimateChange on Antelopes Distribution in Egypt Lambert Academic Publishing

Standards IUCN Subcommittee Petitions 2010 Guidelines for Using the IUCNRed List Categories and Criteria Version 81

Stockwell DRB Peterson AT 2002 Effects of sample size on accuracy of speciesdistribution models Ecol Model 148 1e13 httpdxdoiorg101016S0304-3800(01)00388-X

Suarez AV Tsutsui ND 2004 The value of museum collections for research andsociety Bioscience 54 66 httpdxdoiorg1016410006-3568(2004)054[0066tvomcf]20co2

Taheri S 2010 Hyper Temporal NDVI Images for Modelling and Prediction theHabitat Distribution of Balkan Green Lizard (Lacerta Trilineata) e Case Study

Crete (Greece) University of Twente (ITC) the Netherlands p 54 MScthesis

Taubmann J Theissinger K Feldheim KA Laube I Graf W et al 2011Modelling range shifts and assessing genetic diversity distribution of themontane aquatic mayfly Ameletus inopinatus in Europe under climate changescenarios Conserv Genet 12 503e515 httpdxdoiorg101007s10592-010-0157-x

Teodoro AC Sillero N Alves S Duarte L 2013 Correlation between the habitatsproductivity and species richness (amphibians and reptiles) in Portugal throughremote sensed data In Proc SPIE 8887 Remote Sensing for Agriculture Eco-systems and Hydrology httpdxdoiorg101117122028502

Thomas CD Cameron A Green RE Bakkenes M Beaumont LJ et al 2004Extinction risk from climate change Nature 427 145e148 httpdxdoiorg101038nature02121

Thuiller W 2004 Patterns and uncertainties of species range shifts under climatechange Glob Change Biol 10 2020e2027 httpdxdoiorg101111j1365-2486200400859x

Thuiller W 2007 Biodiversity climate change and the ecologist Nature 448550e552 httpdxdoiorg101038448550a

Thuiller W Lavorel S Arauacutejo MB Sykes MT Prentice IC 2005 Climate changethreats to plant diversity in Europe Proc Natl Acad Sci U S A 1028245e8250 httpdxdoiorg101073pnas0409902102

Thuiller W Lafourcade B Engler R Arauacutejo MB 2009 BIOMOD e a platform forensemble forecasting of species distributions Ecography 32 (3) 369e373httpdxdoiorg101111j1600-0587200805742x

Tolba MK Saab NW 2009 Arab environment Climate Change - Impact ofClimate Change on Arab Countries Arab Forum for Environment andDvelopment

Trotta-Moreu N Lobo JM 2010 Deriving the species richness distribution ofGeotrupinae (Coleoptera Scarabaeoidea) in Mexico from the overlap of indi-vidual model predictions Environ Entomol 39 42e49 httpdxdoiorg101603EN08179

Warren DL Seifert SN 2011 Ecological niche modeling in Maxent the impor-tance of model complexity and the performance of model selection criteriaEcol Appl 21 (2) 335e342 httpdxdoiorg10189010-11711

Wilson KA Cabeza M Klein CJ 2009 Fundamental concepts of spatial con-servation prioritisation In Moilanen A Wilson KA Possingham HP (Eds)Spatial Conservation Prioritization Quantitative Methods and ComputationalTools Oxford University Press pp 16e27 ISBN 9780199547760

Page 6: Journal of Arid Environments - elgabbas.github.io · e Faculty of Sciences & Arts, Taibah University, Al-Ula Campus, Saudi Arabia f Department of Zoology, Suez Canal University, Ismailia,

Fig 2 Estimated future changes in species richness as a result of anthropogenic climate change under two IPCC scenarios (A2a B2a) and at three times in the future (2020 20502080) Each map is the difference from the current species richness (shown in Fig 1a) and is averaged across four GCMmodels Maps made (a) using Maxent logistic output and (b)using thresholded distributions (see Methods)

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221216

reptiles are expected to lose much of their currently suitabilityareas (up to 80 in some cases) and some areas will lose or changea large proportion of their species Thus there will be a need forgreater conservation efforts in the future Some of these areas arealready inside Protected Areas and so hypothetically they are pro-tected while others are either just outside or very far from Pro-tected Areas Although overall Protected Areas contain better(higher priority) values compared to unprotected sites the currentProtected-Area network appears inadequate for future conserva-tion Our results together with others (eg Leach et al 2013) willhelp the Egyptian Environmental Affairs Agency and otherdecision-making parties in Egypt to direct conservation efforts andlimited conservation resources in the right direction

41 Model performance amp variable importance

The negative correlations found between AUC and the areaoccupied and with the extent of occurrence concur with findingsof many other studies (eg Brotons et al 2004 Elith et al 2006Hernandez et al 2006) Rare species are usually habitat special-ists with narrow environmental tolerances and are environmen-tally or geographically restricted compared to widespread species

(Elith et al 2006 Pandit et al 2009) Widespread species are likelyto be generalists occupying a wide range of habitats and climatesmaking it difficult to distinguish between suitable and unsuitablehabitats (Franklin et al 2009) However as pointed out by Elithet al (2006) this may be an artefact an inevitable result for spe-cies with small relative occurrence areas and perhaps a reason notto use AUC to compare model performances (Lobo et al 2008)Most (about 60) of the Egyptian herpetofauna are narrowlydistributed occupying less than 10 of Egypts area (Baha El Din2006) There is no evidence of a strong correlation betweenmodel accuracy and the number of records used (see Elith et al2006 Newbold et al 2009b) and the weak negative correlationfound here is unusual since most (eg Stockwell and Peterson2002 Hernandez et al 2006) find that models with larger sam-ple sizes have higher accuracy (but see de Pous et al 2011)

We found altitude to be the most effective predictor for manyspecies In contrast the NDVI predictors were not useful unlike inmany other cases (eg Egbert et al 2002 Anderson et al 2006Kgosiesele 2010 Taheri 2010) but perhaps typical for arid envi-ronments (eg Soultans (2011) study on Egyptian antelopes) or forherpetofauna (eg Teodoro et al 2013) Habitat categories likewisemade only a low contribution to the reptile models

Fig 3 Expected patterns of cumulative (a) losses and (b) gains of species as a result of climate change The losses of each species in each pixel (under both dispersal assumptions)and gains (under unlimited dispersal only) were averaged across GCM models accumulated across species and then plotted

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 217

42 Species richness

Species richness by itself is not adequate as an indicator of bioticchange since species composition can change independently Ourresults suggest that in the future some areas with currently highexpected species richness may lose many species or undergo largechanges in species composition Some of these areas are alreadyunder a certain degree of protection but even if they can bemaintained or enhanced (eg Siwa oasis Gebel Elba and the coastalareas over the Aqaba gulf Fig S2) current Protected Areas are notsufficient to prevent expected loss of species Other areas (eg WadiEl-Natrun the Suez Canal area Red Sea inland wadis the poorlyknown Qattara Depression and the areas around Gebel El-Gallalaand Gebel El-Hallal) are unprotected and measures are needed toconserve them There are similar situations elsewhere for examplehigh species loss and turnover in the reptiles of a priority conser-vation hotspot in Europe (the Iberian Peninsula) are expected (dePous 2011) with great variation in expected patterns of loss andchange with different projections and dispersal assumptions Otherstudies have also concluded that large changes are likely in chelo-nians worldwide (Ihlow et al 2012) Mediterranean mammals(Maiorano et al 2011) US trees and vertebrates (Currie 2001) andMesoAmerican trees (Golicher et al 2012)

43 Range and status changes

Two species are expected to lose their suitable habitats entirelyin at least one future projection 1) T mindiae is a near-endemicspecies recorded just from northwest Egypt and northern Cyrena-ica (eastern Libya) its distribution in Egypt is restricted to Siwaoasis the Qattara Depression and their periphery (Baha El Din2006) Future range changes suggest it will be Endangered (loss50e80) by 2020 and as Critically Endangered (loss gt80) by 2050and 2080 (under all emission scenarios and dispersal assumptions)2) H robustus is localized to the East African coast from Zanzibar tosouthern Egypt Arabia east to Pakistan its distribution in Egypt ison the Red Sea coast from El-Quseir southwards (Baha El Din2006) The A2a scenario suggests it will be Critically Endangeredby 2020 and 2050 while under the B2a scenario it will be Endan-gered by 2020 and Critically Endangered by 2050 and 2080 Morethan half of its current distribution in Egypt is located within theWadi El-Gemal and Gebel Elba Protected Areas so hypothetically itis protected However it is found mainly on buildings and henceassociated with man it may in fact be an invasive to Egypt

Another eight species are expected to be classified as CriticallyEndangered (loss gt80) in at least one averaged future projection(Acanthodactylus longipes Cerastes vipera Eryx jaculus Eumeces

Fig 4 Current conservation prioritizationl of areas according to the Zonation algo-rithm under two cell removal rules (a) core-area function (identifies areas where oneor more species have important occurrences) and (b) additive benefit function (whichgives greater emphasis to areas with high species richness) Darker colours representhigher conservation value (For interpretation of the references to colour in this figurelegend the reader is referred to the web version of this article)

Fig 5 Mean prioritization value for conservation (plusmn95 confidence limits) derivedfrom Zonation for Protected Areas (PAs) and outside Protected Areas (non-PAs) forcurrent and expected future conditions using the cell-removal rule of (a) core-areafunction and (b) additive benefit function

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221218

schneiderii Malpolon moilensis Ptyodactylus guttatusP siphonorhina and Trapelus mutabilis) According to the currentassessment twenty-seven species are classified as Endangered(loss 50e80) in at least one projection (mostly by 2080 in the A2scenario) two of these are classified assuming no dispersal asEndangered in all future projections (Uromastyx aegyptia and Pris-turus flavipunctatus for details see Tables S2eS3)

Reptiles are also likely to gain from global warming especially inmore northern areas of the globe if they can disperse and expandtheir distributions but under perhaps the more realistic assump-tion of no dispersal most species were expected to lose habitatsuitability by 2050 (Arauacutejo et al 2006) In Egypt areas of clearfuture expected gain (assuming unlimited dispersal) are the areabetween the Siwa oasis and Qattara Depression northwards almostto the northern coastal strip and some sparse locations in centralSinai There were greater gains under the B2a scenario in theeastern part of the Western Desert and some patchy areas aroundthe Nile (Fig 3b) This pattern is similar to that reported by Araujoet al (2006)

Some of the species identified here as being most at risk due tofuture climate change are in fact resilient species that are ecolog-ically very adaptable and are currently experiencing range expan-sion and population growth at least in Egypt (eg H robustus) This

disparity is probably because modelling is based solely on theprobable impacts of climate change on the spatial distribution ofsuitable habitats without taking evolution or ecological adapt-ability and resilience into account Species vary greatly in theirresponse to ecological change some adapting very well even tovery rapid anthropogenically induced changes (eg invasive spe-cies) while others are much less adaptable and tend to go extinctfirst

There are no published studies that discuss the effect of climatechange on Egyptian reptiles obviating any comparisons Informa-tion on the reptiles of the adjacent countries of Sudan and Libya arevery limited making it impossible to assess any movements thatmight compensate for Egyptian losses The patterns of expectedchanges in Leach et als (2013) similar study on mammals andbutterflies do not concur with ours Mammal species richness wasexpected to decline in the Mediterranean and Red Sea areas (by40e60) and increase elsewhere (by 80e100) while for butter-flies declines are expected over almost the whole of Egypt (by40e60) except the south which is expected to increase (by40e60) In the models of European reptiles (de Pous 2011)

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 219

between eight (unlimited dispersal) and 21 (no dispersal) specieswere expected to become extinct in at least one future projectionwith two more losing more than 80 of their currently suitablehabitats This is a much more damaging scenario than the expectedresponses of Egyptian reptiles in our study There are no othercomparable studies from North Africa or the Middle East

44 Area prioritization for conservation

In our study Protected Areas (Fig S3) were more valuable inconserving reptiles than areas outside them both currently and inthe future although the difference is expected to get smaller withtime (Fig 5) Although this is encouraging the Protected Areanetwork appears inadequate to conserve Egyptian reptiles NewProtected Areas that have already been proposed would go someway towards reducing this deficiency in coverage although otherswill be needed in the middle to north Sinai (especially aroundGebel El-Hallal) the Suez Canal area both sides of the Suez Gulf thenorthern Mediterranean coast coastal wadis of the Red Sea be-tween Hurghada and Mersa Alam and the Gebel El-Gallala areaMore effective protection is required in the established ProtectedAreas including those in South Sinai Gebel Elba and Siwa oasis dueto their relatively high expected future suitability loss The Pro-tected Areas of Siwa oasis and El-Gilf El-Kebir plus the areas ofWadiEl-Natrun the Gebel El-Hallal area and wadis of the Red Sea areexpected to have particularly high turnover in species composition(Fig S6) and hence probably need more conservation efforts Somecomparable information is available for Egypts butterflies andmammals (Gilbert and Zalat 2008 Basuony et al 2010 Leach et al2013) Egypts network of Protected Areas appears adequate toconserve Egyptian butterfly hotspots except for the Mediterraneancoast between Alexandria and Sallum (Gilbert and Zalat 2008) butinadequate to conserve important mammal areas with new Pro-tected Areas needed to be constructed in the lower Nile Valleyalong the north coast between Alexandria and Sallum top part ofthe Suez Gulf and perhaps the Qattara Depression (Basuony et al2010) Suggestions for conserving mammals are consistent to someextent with our results which also suggest new Protected Areasalong the Suez Gulf Mediterranean coast from Mersa Matruh toSallum and the Qattara Depression Leach et al (2013) concur withour results in suggesting priority areas along the western Medi-terranean coast around Greater Cairo central and north Sinai andthe Red Sea coast

45 Limitations

There are some limitations of the results shown in this study 1)Egyptian climate data in WorldClim are interpolated from ratherfew non-randomly placed weather stations in Egypt and adjacentcountries Although making the climate data less reliable thanelsewhere there are no alternative climate data available 2) Anensemble approach (eg using BIOMOD Thuiller et al 2009) isoften suggested for SDM modelling to avoid dependence on oneparticular algorithm This is not possible without a greater numberof records since other algorithms are not as flexible or robust asMaxent Despite all the caveats with using SDMs and criticisms ofMaxent in particular (Royle et al 2012 but see Hastie and Fithian2013 Elith et al 2006 Hernandez et al 2006) in practice coun-tries such as Egypt have little else to guide them in attempts atclimate change adaptation 3) There were signs of sampling bias inthe available records Although this violates the assumptions of allSDMs it is a problem common to nearly all datasets worldwideNew techniques may allow for the correction of sampling bias egusing target-group approach (Phillips et al 2009) although thedata requirements are currently impossible for Egypt and most

other developing countries) 4) Using Maxents default settings iscriticized by some authors they need to be modified dependent onthe data (Warren and Seifert 2011 Radosavljevic and Anderson2014 Merow et al 2013) but we think using the default settingsis enough in this study

5 Conclusion

Although Egyptian datasets exist knowledge of Egyptian spe-cies and conservation priorities are limited (with consequencesexplored in Leach et al 2013) nevertheless our results and those ofLeach et al (2013) encourage us to believe that useful insights canbe gained for conservation priorities by using the combination ofspecies distribution modelling plus spatial conservation prioriti-zation techniques (Hannah 2010 Klorvuttimontara et al 2011) Itis crucial for the Egyptian Environmental Affairs Agency to establishan accurate and reliable database of records assessed by experts tohelp improve the basic information for future conservation as-sessments and analyses

Although many SDM studies have been done on a variety oforganisms in recent decades aiming to solve conservation prob-lems evidence of their practical utility in real-world conservation issparse with very few studies clearly targeting conservation de-cisions (Guisan et al 2013) If implemented correctly SDMs canplay a critical role in supporting spatial conservation decision-making (Addison et al 2013) For reptiles there is a shortage ofstudies estimating their response to expected climate changes(Arauacutejo et al 2006) Although some suggest that reptiles (and to agreater extent amphibians) will be strongly affected by climatechange (eg Gibbons et al 2000 Arauacutejo et al 2006 Carvalho et al2010) very few of themwere concerned with arid environments orthe Middle East We hope that this and other similar studies thatestimate the fate of reptiles under climate change help decision-making authorities to take the measures necessary to conservethem

Acknowledgements

We thank the Italian Cooperation (Debt Swap) in Egypt forfunding the BioMAP project that allowed the collation of the da-tabases used in this work Dr Tim Newbold for allowing the use ofthe data from his field trips Dr Mustafa Fouda (Former Director ofthe Nature Conservation Sector EEAA) for facilities offered duringBioMAP project and all the BioMAP staff This work was supportedby the award of a Chevening Scholarship Masters grant to AEl-GWe are especially grateful to two anonymous referees whogreatly improved an earlier manuscript

Appendix A Supplementary data

Supplementary data related to this article can be found at httpdxdoiorg101016jjaridenv201512007

References

Addison PFE Rumpff L Bau SS Carey JM Chee YE Jarrad FC McBride MFBurgman MA 2013 Practical solutions for making models indispensable inconservation decision-making Divers Distrib 19 (5e6) 490e502 httpdxdoiorg101111ddi12054

Anderson RP Peterson AT Egbert SL 2006 Vegetation-index models predictareas vulnerable to purple loosestrife (Lythrum salicaria) invasion in KansasSouthwest Nat 51 471e480 httpdxdoiorg1018940038-4909(2006)51[471VMPAVT]20CO2

Arauacutejo MB 2009 Climate change and spatial conservation planning InMoilanen A Wilson KA Possingham HP (Eds) Spatial Conservation Prior-itization Quantitative Methods and Computational Tools Oxford UniversityPress pp 172e184 ISBN 9780199547760

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221220

Arauacutejo MB Rahbek C 2006 How does climate change affect biodiversity Sci-ence 313 1396e1397 httpdxdoiorg101126science1131758

Arauacutejo MB Thuiller W Pearson RG 2006 Climate warming and the decline ofamphibians and reptiles in Europe J Biogeogr 33 1712e1728 httpdxdoiorg101111j1365-2699200601482x

Baha El Din SM 2001 The Herpetofauna of Egypt Species Communities andAssemblages University of Nottingham p 435 Unpublished PhD thesis

Baha El Din SM 2006 A Guide to the Reptiles and Amphibians of Egypt TheAmerican University in Cairo Press ISBN 9789774249792

Baha El Din SM 2007 A new lizard of the Acanthodactylus scutellatus group(Squamata Lacertidae) from Egypt Zool Middle East 40 21e32 httpdxdoiorg10108009397140200710638200

Baldwin RA 2009 Use of maximum entropy modeling in wildlife research En-tropy 11 854e866 httpdxdoiorg103390e11040854

Ball IR Possingham HP Watts M 2009 Marxan and relatives software forspatial conservation prioritisation In Moilanen A Wilson KAPossingham HP (Eds) Spatial Conservation Prioritization QuantitativeMethods and Computational Tools Oxford University Press pp 185e195 ISBN9780199547760

Basuony MI Gilbert F Zalat S 2010 The Mammals of Egypt Atlas Red DataListing amp Conservation BioMAP amp CultNat EEAA amp Bibliotheca AlexandrinaCairo available at httpecologynottinghamacuk~plzfg

Bombi P Salvi D Bologna MA Pettorelli N 2012 Cross-scale predictions allowthe identification of local conservation priorities from atlas data Anim Conserv15 378e387 httpdxdoiorg101111j1469-1795201200526x

Brotons L Thuiller W Arauacutejo MB Hirzel AH 2004 Presence-absence versuspresence-only modelling methods for predicting bird habitat suitability Ecog-raphy 27 437e448 httpdxdoiorg101111j0906-7590200403764x

Cadby CD While GM Hobday AJ Uller T Wapstra E 2010 Multi-scaleapproach to understanding climate effects on offspring size at birth and date ofbirth in a reptile Integr Zool 5 164e175 httpdxdoiorg101111j1749-4877201000201x

Carvalho SB Brito JC Crespo EJ Possingham HP 2010 From climate changepredictions to actions e conserving vulnerable animal groups in hotspots at aregional scale Glob Change Biol 16 (12) 3257e3270 httpdxdoiorg101111j1365-2486201002212x

Currie DJ 2001 Projected effects of climate change on patterns of vertebrate andtree species richness in the conterminous United States Ecosystems 4216e225 httpdxdoiorg101007s10021-001-0005-4

de Pous P 2011 Climate Change and European Reptiles from Large-scale SpatialRange Shift Projections towards a More Realistic Assessment of Species Re-sponses University of East Anglia UK p 64 MSc thesis

de Pous P Beukema W Weterings M Duumlmmer I Geniez P 2011 Area priori-tization and performance evaluation of the conservation area network for theMoroccan herpetofauna a preliminary assessment Biodivers Conserv 2089e118 httpdxdoiorg101007s10531-010-9948-0

Edgar P Foster J Baker J 2010 Reptile Habitat Management HandbookAmphibian and Reptile Conservation Bournemouth

Egbert SL Martiacutenez-Meyer E Ortega-Huerta M Peterson AT 2002 Use ofdatasets derived from time-series AVHRR imagery as surrogates for land covermaps in predicting species distributions In Proceedings IEEE 2002 Interna-tional Geoscience and Remote Sensing Symposium (IGARSS) vol 4pp 2337e2339 httpdxdoiorg101109IGARSS20021026537

El Alqamy H Ismael A Abdelhameed A Nagy A Hamada A et al 2010 Pre-dicting the status and distribution of the Nubian Ibex (Capra nubiana) in thehigh-altitude mountains of south Sinai (Egypt) In Galemys Newsletter of theSpanish Society for the Preservation and Study of Mammals vol 22pp 517e530

Elith J Graham CH Anderson RP Dudiacutek M Ferrier S et al 2006 Novelmethods improve prediction of species distributions from occurrence dataEcography 29 129e151 httpdxdoiorg101111j20060906-759004596x

Elith J Phillips SJ Hastie T Dudiacutek M Chee YE Yates CJ 2011 A statisticalexplanation of MaxEnt for ecologists Divers Distrib 17 (1) 43e57 httpdxdoiorg101111j1472-4642201000725x

FAO 2012 Food and Agriculture Organisation World Development IndicatorsAverage Precipitation in Depth (mm per year) httpdataworldbankorgindicatorAGLNDPRCPMM

Franklin J 2009 Mapping Species Distributions Spatial Inference and PredictionCambridge University Press Cambridge UK ISBN 9780521700023

Franklin J Wejnert KE Hathaway SA Rochester CJ Fisher RN 2009 Effect ofspecies rarity on the accuracy of species distribution models for reptiles andamphibians in southern California Divers Distrib 15 167e177 httpdxdoiorg101111j1472-4642200800536x

Gibbons JW Scott DE Ryan TJ Buhlmann KA Tuberville TD Metts BSGreene JL Mills T et al 2000 The global decline of reptiles Deja Vu Am-phibians BioScience 50 (8) 653e666 httpdxdoiorg1016410006-3568(2000)050[0653TGDORD]20CO2

Gilbert F Zalat S 2008 The Butterflies of Egypt Atlas Red Data Listing amp Con-servation BioMAP EEAA Cairo available at httpecologynottinghamacuk~plzfg

Golicher DJ Cayuela L Newton AC 2012 Effects of climate change on the po-tential species richness of Mesoamerican forests Biotropica 44 284e293httpdxdoiorg101111j1744-7429201100815x

Graham CH Elith J Hijmans RJ Guisan A Peterson AT et al 2008 Theinfluence of spatial errors in species occurrence data used in distribution

models J Appl Ecol 45 239e247 httpdxdoiorg101111j1365-2664200701408x

Guisan A Tingley R Baumgartner JB Naujokaitis-Lewis I Sutcliffe PRTulloch AIT Regan TJ Brotons L et al 2013 Predicting species distributionsfor conservation decisions Ecol Lett 16 (12) 1424e1435 httpdxdoiorg101111ele12189

Hannah L 2010 A global conservation system for climate-change adaptationConserv Biol 24 70e77 httpdxdoiorg101111j1523-1739200901405x

Hannah L 2011 Climate Change Biology Elsevier Ltd ISBN 9780123741820Hannah L Lovejoy TE Schneider SH 2005 Biodiversity and climate change in

context In Lovejoy TE Hannah L (Eds) Climate Change and BiodiversityYale University Press New Haven CT USA

Hannah L Midgley G Andelman S Arauacutejo M Hughes G et al 2007 Protectedarea needs in a changing climate Front Ecol Environ 5 131e138 httpdxdoiorg1018901540-9295(2007)5[131PANIAC]20CO2

Hastie T Fithian W 2013 Inference from presence-only data the ongoing con-troversy Ecography 36 864e867 httpdxdoiorg101111j1600-0587201300321x

Hernandez PA Graham CH Master LL Albert DL 2006 The effect of samplesize and species characteristics on performance of different species distributionmodeling methods Ecography 29 773e785 httpdxdoiorg101111j0906-7590200604700x

Hijmans RJ Cameron SE Parra JL Jones PG Jarvis A 2005 Very high reso-lution interpolated climate surfaces for global land areas Int J Climatol 251965e1978 httpdxdoiorg101002Joc1276

Holt AC Salkeld DJ Fritz CL Tucker JR Gong P 2009 Spatial analysis ofplague in California niche modeling predictions of the current distribution andpotential response to climate change Int J Health Geogr 8 38 httpdxdoiorg1011861476-072X-8-38

Hoyle M James M 2005 Global warming human population pressure andviability of the worlds smallest butterfly Conserv Biol 19 1113e1124 httpdxdoiorg101111j1523-1739200500166x

Ihlow F Dambach J Engler JO Flecks M Hartmann T et al 2012 On the brinkof extinction How climate change may affect global chelonian species richnessand distribution Glob Change Biol 18 1520e1530 httpdxdoiorg101111j1365-2486201102623x

IPCC 2007 Climate Change 2007 Synthesis Report Contribution of WorkingGroups I II and III to the Fourth Assessment Report of the IntergovernmentalPanel on Climate Change IPCC Geneva Switzerland

Kgosiesele E 2010 Predictive Distribution Modelling of Timon Lepida in SpainUniversity of Twente (ITC) the Netherlands p 87 MSc thesis

Klorvuttimontara S McClean CJ Hill JK 2011 Evaluating the effectiveness ofprotected areas for conserving tropical forest butterflies of Thailand BiolConserv 144 2534e2540 httpdxdoiorg101016jbiocon201107012

Leach K Zalat S Gilbert F 2013 Egypts protected area network under futureclimate change Biol Conserv 159 490e500 httpdxdoiorg101016jbiocon201211025

Lobo JM Jimenez-Valverde A Real R 2008 AUC a misleading measure of theperformance of predictive distribution models Glob Ecol Biogeogr 17145e151 httpdxdoiorg101111j1466-8238200700358x

Maiorano L Falcucci A Zimmermann NE Psomas A Pottier J et al 2011 Thefuture of terrestrial mammals in the Mediterranean basin under climatechange Philosophical Trans R Soc Lond Ser B Biol Sci 366 2681e2692httpdxdoiorg101098rstb20110121

Marini MA Barbet-Massin M Lopes LE Jiguet F 2009 Predicted climate-driven bird distribution changes and forecasted conservation conflicts in aneotropical savanna Conserv Biol 23 1558e1567 httpdxdoiorg101111j1523-1739200901258x

Merow C Smith MJ Silander JA 2013 A practical guide to MaxEnt for modelingspecies distributions what it does and why inputs and settings matterEcography 36 (10) 1058e1069 httpdxdoiorg101111j1600-0587201307872x

Moilanen A Meller L Leppeuroanen J Pouzols FM Arponen A et al 2012 Zona-tion Spatial Conservation Planning Framework and Software v31 User Manualwwwhelsinkifibioscienceconsplan

Newbold T 2009 The Value of Species Distribution Models as a Tool for Conser-vation and Ecology in Egypt and Britain University of Nottingham UK p 312PhD thesis

Newbold T Gilbert F Zalat S El-Gabbas A Reader T 2009a Climate-basedmodels of spatial patterns of species richness in Egypts butterfly and mammalfauna J Biogeogr 36 2085e2095 httpdxdoiorg101111j1365-2699200902140x

Newbold T Reader T Zalat S El-Gabbas A Gilbert F 2009b Effect of charac-teristics of butterfly species on the accuracy of distribution models in an aridenvironment Biodivers Conserv 18 3629e3641 httpdxdoiorg101007s10531-009-9668-5

Pandit SN Kolasa J Cottenie K 2009 Contrasts between habitat generalists andspecialists an empirical extension to the basic metacommunity frameworkEcology 90 2253e2262 httpdxdoiorg10189008-08511

Pearson RG Dawson TP 2003 Predicting the impacts of climate change on thedistribution of species are bioclimate envelope models useful Glob EcolBiogeogr 12 361e371 httpdxdoiorg101046j1466-822X200300042x

Pearson RG Raxworthy CJ Nakamura M Peterson AT 2007 Predicting speciesdistributions from small numbers of occurrence records a test case using

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 221

cryptic geckos in Madagascar J Biogeogr 34 102e117 httpdxdoiorg101111j1365-2699200601594x

Phillips SJ ATampT Research 2011 A Brief Tutorial on Maxent Available at httpwwwcsprincetonedu~schapiremaxenttutorialtutorialdoc (Last accessed290512)

Phillips SJ Anderson RP Schapire RE 2006 Maximum entropy modeling ofspecies geographic distributions Ecol Model 190 231e259 httpdxdoiorg101016jecolmodel200503026

Phillips SJ Dudiacutek M Elith J Graham CH Lehmann A Leathwick J Ferrier S2009 Sample selection bias and presence-only distribution models implica-tions for background and pseudo-absence data Ecol Appl 19 181e197 httpdxdoiorg10189007-21531

R Development Core Team 2012 R a Language and Environment for StatisticalComputing R Foundation for Statistical Computing Vienna Austria ISBN 3-900051-07-0 httpwwwR-projectorg

Radosavljevic A Anderson RP 2014 Making better Maxent models of speciesdistributions complexity overfitting and evaluation J Biogeogr 41 (4)629e643 httpdxdoiorg101111jbi12227

Royle JA Chandler RB Yackulic C Nichols JD 2012 Likelihood analysis ofspecies occurrence probability from presence-only data for modelling speciesdistributions Methods Ecol Evol 3 545e554 httpdxdoiorg101111j2041-210X201100182x

Sauer J Domisch S Nowak C Haase P 2011 Lowmountain ranges summit trapsfor montane freshwater species under climate change Biodivers Conserv 203133e3146 httpdxdoiorg101007s10531-011-0140-y

Songer M Delion M Biggs A Huang Q 2012 Modeling impacts of climatechange on Giant Panda Habitat Int J Ecol 2012 1e12 httpdxdoiorg1011552012108752

Soultan AE 2011 Antelope distribution in Egypt Modeling the Impact of ClimateChange on Antelopes Distribution in Egypt Lambert Academic Publishing

Standards IUCN Subcommittee Petitions 2010 Guidelines for Using the IUCNRed List Categories and Criteria Version 81

Stockwell DRB Peterson AT 2002 Effects of sample size on accuracy of speciesdistribution models Ecol Model 148 1e13 httpdxdoiorg101016S0304-3800(01)00388-X

Suarez AV Tsutsui ND 2004 The value of museum collections for research andsociety Bioscience 54 66 httpdxdoiorg1016410006-3568(2004)054[0066tvomcf]20co2

Taheri S 2010 Hyper Temporal NDVI Images for Modelling and Prediction theHabitat Distribution of Balkan Green Lizard (Lacerta Trilineata) e Case Study

Crete (Greece) University of Twente (ITC) the Netherlands p 54 MScthesis

Taubmann J Theissinger K Feldheim KA Laube I Graf W et al 2011Modelling range shifts and assessing genetic diversity distribution of themontane aquatic mayfly Ameletus inopinatus in Europe under climate changescenarios Conserv Genet 12 503e515 httpdxdoiorg101007s10592-010-0157-x

Teodoro AC Sillero N Alves S Duarte L 2013 Correlation between the habitatsproductivity and species richness (amphibians and reptiles) in Portugal throughremote sensed data In Proc SPIE 8887 Remote Sensing for Agriculture Eco-systems and Hydrology httpdxdoiorg101117122028502

Thomas CD Cameron A Green RE Bakkenes M Beaumont LJ et al 2004Extinction risk from climate change Nature 427 145e148 httpdxdoiorg101038nature02121

Thuiller W 2004 Patterns and uncertainties of species range shifts under climatechange Glob Change Biol 10 2020e2027 httpdxdoiorg101111j1365-2486200400859x

Thuiller W 2007 Biodiversity climate change and the ecologist Nature 448550e552 httpdxdoiorg101038448550a

Thuiller W Lavorel S Arauacutejo MB Sykes MT Prentice IC 2005 Climate changethreats to plant diversity in Europe Proc Natl Acad Sci U S A 1028245e8250 httpdxdoiorg101073pnas0409902102

Thuiller W Lafourcade B Engler R Arauacutejo MB 2009 BIOMOD e a platform forensemble forecasting of species distributions Ecography 32 (3) 369e373httpdxdoiorg101111j1600-0587200805742x

Tolba MK Saab NW 2009 Arab environment Climate Change - Impact ofClimate Change on Arab Countries Arab Forum for Environment andDvelopment

Trotta-Moreu N Lobo JM 2010 Deriving the species richness distribution ofGeotrupinae (Coleoptera Scarabaeoidea) in Mexico from the overlap of indi-vidual model predictions Environ Entomol 39 42e49 httpdxdoiorg101603EN08179

Warren DL Seifert SN 2011 Ecological niche modeling in Maxent the impor-tance of model complexity and the performance of model selection criteriaEcol Appl 21 (2) 335e342 httpdxdoiorg10189010-11711

Wilson KA Cabeza M Klein CJ 2009 Fundamental concepts of spatial con-servation prioritisation In Moilanen A Wilson KA Possingham HP (Eds)Spatial Conservation Prioritization Quantitative Methods and ComputationalTools Oxford University Press pp 16e27 ISBN 9780199547760

Page 7: Journal of Arid Environments - elgabbas.github.io · e Faculty of Sciences & Arts, Taibah University, Al-Ula Campus, Saudi Arabia f Department of Zoology, Suez Canal University, Ismailia,

Fig 3 Expected patterns of cumulative (a) losses and (b) gains of species as a result of climate change The losses of each species in each pixel (under both dispersal assumptions)and gains (under unlimited dispersal only) were averaged across GCM models accumulated across species and then plotted

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 217

42 Species richness

Species richness by itself is not adequate as an indicator of bioticchange since species composition can change independently Ourresults suggest that in the future some areas with currently highexpected species richness may lose many species or undergo largechanges in species composition Some of these areas are alreadyunder a certain degree of protection but even if they can bemaintained or enhanced (eg Siwa oasis Gebel Elba and the coastalareas over the Aqaba gulf Fig S2) current Protected Areas are notsufficient to prevent expected loss of species Other areas (eg WadiEl-Natrun the Suez Canal area Red Sea inland wadis the poorlyknown Qattara Depression and the areas around Gebel El-Gallalaand Gebel El-Hallal) are unprotected and measures are needed toconserve them There are similar situations elsewhere for examplehigh species loss and turnover in the reptiles of a priority conser-vation hotspot in Europe (the Iberian Peninsula) are expected (dePous 2011) with great variation in expected patterns of loss andchange with different projections and dispersal assumptions Otherstudies have also concluded that large changes are likely in chelo-nians worldwide (Ihlow et al 2012) Mediterranean mammals(Maiorano et al 2011) US trees and vertebrates (Currie 2001) andMesoAmerican trees (Golicher et al 2012)

43 Range and status changes

Two species are expected to lose their suitable habitats entirelyin at least one future projection 1) T mindiae is a near-endemicspecies recorded just from northwest Egypt and northern Cyrena-ica (eastern Libya) its distribution in Egypt is restricted to Siwaoasis the Qattara Depression and their periphery (Baha El Din2006) Future range changes suggest it will be Endangered (loss50e80) by 2020 and as Critically Endangered (loss gt80) by 2050and 2080 (under all emission scenarios and dispersal assumptions)2) H robustus is localized to the East African coast from Zanzibar tosouthern Egypt Arabia east to Pakistan its distribution in Egypt ison the Red Sea coast from El-Quseir southwards (Baha El Din2006) The A2a scenario suggests it will be Critically Endangeredby 2020 and 2050 while under the B2a scenario it will be Endan-gered by 2020 and Critically Endangered by 2050 and 2080 Morethan half of its current distribution in Egypt is located within theWadi El-Gemal and Gebel Elba Protected Areas so hypothetically itis protected However it is found mainly on buildings and henceassociated with man it may in fact be an invasive to Egypt

Another eight species are expected to be classified as CriticallyEndangered (loss gt80) in at least one averaged future projection(Acanthodactylus longipes Cerastes vipera Eryx jaculus Eumeces

Fig 4 Current conservation prioritizationl of areas according to the Zonation algo-rithm under two cell removal rules (a) core-area function (identifies areas where oneor more species have important occurrences) and (b) additive benefit function (whichgives greater emphasis to areas with high species richness) Darker colours representhigher conservation value (For interpretation of the references to colour in this figurelegend the reader is referred to the web version of this article)

Fig 5 Mean prioritization value for conservation (plusmn95 confidence limits) derivedfrom Zonation for Protected Areas (PAs) and outside Protected Areas (non-PAs) forcurrent and expected future conditions using the cell-removal rule of (a) core-areafunction and (b) additive benefit function

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221218

schneiderii Malpolon moilensis Ptyodactylus guttatusP siphonorhina and Trapelus mutabilis) According to the currentassessment twenty-seven species are classified as Endangered(loss 50e80) in at least one projection (mostly by 2080 in the A2scenario) two of these are classified assuming no dispersal asEndangered in all future projections (Uromastyx aegyptia and Pris-turus flavipunctatus for details see Tables S2eS3)

Reptiles are also likely to gain from global warming especially inmore northern areas of the globe if they can disperse and expandtheir distributions but under perhaps the more realistic assump-tion of no dispersal most species were expected to lose habitatsuitability by 2050 (Arauacutejo et al 2006) In Egypt areas of clearfuture expected gain (assuming unlimited dispersal) are the areabetween the Siwa oasis and Qattara Depression northwards almostto the northern coastal strip and some sparse locations in centralSinai There were greater gains under the B2a scenario in theeastern part of the Western Desert and some patchy areas aroundthe Nile (Fig 3b) This pattern is similar to that reported by Araujoet al (2006)

Some of the species identified here as being most at risk due tofuture climate change are in fact resilient species that are ecolog-ically very adaptable and are currently experiencing range expan-sion and population growth at least in Egypt (eg H robustus) This

disparity is probably because modelling is based solely on theprobable impacts of climate change on the spatial distribution ofsuitable habitats without taking evolution or ecological adapt-ability and resilience into account Species vary greatly in theirresponse to ecological change some adapting very well even tovery rapid anthropogenically induced changes (eg invasive spe-cies) while others are much less adaptable and tend to go extinctfirst

There are no published studies that discuss the effect of climatechange on Egyptian reptiles obviating any comparisons Informa-tion on the reptiles of the adjacent countries of Sudan and Libya arevery limited making it impossible to assess any movements thatmight compensate for Egyptian losses The patterns of expectedchanges in Leach et als (2013) similar study on mammals andbutterflies do not concur with ours Mammal species richness wasexpected to decline in the Mediterranean and Red Sea areas (by40e60) and increase elsewhere (by 80e100) while for butter-flies declines are expected over almost the whole of Egypt (by40e60) except the south which is expected to increase (by40e60) In the models of European reptiles (de Pous 2011)

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 219

between eight (unlimited dispersal) and 21 (no dispersal) specieswere expected to become extinct in at least one future projectionwith two more losing more than 80 of their currently suitablehabitats This is a much more damaging scenario than the expectedresponses of Egyptian reptiles in our study There are no othercomparable studies from North Africa or the Middle East

44 Area prioritization for conservation

In our study Protected Areas (Fig S3) were more valuable inconserving reptiles than areas outside them both currently and inthe future although the difference is expected to get smaller withtime (Fig 5) Although this is encouraging the Protected Areanetwork appears inadequate to conserve Egyptian reptiles NewProtected Areas that have already been proposed would go someway towards reducing this deficiency in coverage although otherswill be needed in the middle to north Sinai (especially aroundGebel El-Hallal) the Suez Canal area both sides of the Suez Gulf thenorthern Mediterranean coast coastal wadis of the Red Sea be-tween Hurghada and Mersa Alam and the Gebel El-Gallala areaMore effective protection is required in the established ProtectedAreas including those in South Sinai Gebel Elba and Siwa oasis dueto their relatively high expected future suitability loss The Pro-tected Areas of Siwa oasis and El-Gilf El-Kebir plus the areas ofWadiEl-Natrun the Gebel El-Hallal area and wadis of the Red Sea areexpected to have particularly high turnover in species composition(Fig S6) and hence probably need more conservation efforts Somecomparable information is available for Egypts butterflies andmammals (Gilbert and Zalat 2008 Basuony et al 2010 Leach et al2013) Egypts network of Protected Areas appears adequate toconserve Egyptian butterfly hotspots except for the Mediterraneancoast between Alexandria and Sallum (Gilbert and Zalat 2008) butinadequate to conserve important mammal areas with new Pro-tected Areas needed to be constructed in the lower Nile Valleyalong the north coast between Alexandria and Sallum top part ofthe Suez Gulf and perhaps the Qattara Depression (Basuony et al2010) Suggestions for conserving mammals are consistent to someextent with our results which also suggest new Protected Areasalong the Suez Gulf Mediterranean coast from Mersa Matruh toSallum and the Qattara Depression Leach et al (2013) concur withour results in suggesting priority areas along the western Medi-terranean coast around Greater Cairo central and north Sinai andthe Red Sea coast

45 Limitations

There are some limitations of the results shown in this study 1)Egyptian climate data in WorldClim are interpolated from ratherfew non-randomly placed weather stations in Egypt and adjacentcountries Although making the climate data less reliable thanelsewhere there are no alternative climate data available 2) Anensemble approach (eg using BIOMOD Thuiller et al 2009) isoften suggested for SDM modelling to avoid dependence on oneparticular algorithm This is not possible without a greater numberof records since other algorithms are not as flexible or robust asMaxent Despite all the caveats with using SDMs and criticisms ofMaxent in particular (Royle et al 2012 but see Hastie and Fithian2013 Elith et al 2006 Hernandez et al 2006) in practice coun-tries such as Egypt have little else to guide them in attempts atclimate change adaptation 3) There were signs of sampling bias inthe available records Although this violates the assumptions of allSDMs it is a problem common to nearly all datasets worldwideNew techniques may allow for the correction of sampling bias egusing target-group approach (Phillips et al 2009) although thedata requirements are currently impossible for Egypt and most

other developing countries) 4) Using Maxents default settings iscriticized by some authors they need to be modified dependent onthe data (Warren and Seifert 2011 Radosavljevic and Anderson2014 Merow et al 2013) but we think using the default settingsis enough in this study

5 Conclusion

Although Egyptian datasets exist knowledge of Egyptian spe-cies and conservation priorities are limited (with consequencesexplored in Leach et al 2013) nevertheless our results and those ofLeach et al (2013) encourage us to believe that useful insights canbe gained for conservation priorities by using the combination ofspecies distribution modelling plus spatial conservation prioriti-zation techniques (Hannah 2010 Klorvuttimontara et al 2011) Itis crucial for the Egyptian Environmental Affairs Agency to establishan accurate and reliable database of records assessed by experts tohelp improve the basic information for future conservation as-sessments and analyses

Although many SDM studies have been done on a variety oforganisms in recent decades aiming to solve conservation prob-lems evidence of their practical utility in real-world conservation issparse with very few studies clearly targeting conservation de-cisions (Guisan et al 2013) If implemented correctly SDMs canplay a critical role in supporting spatial conservation decision-making (Addison et al 2013) For reptiles there is a shortage ofstudies estimating their response to expected climate changes(Arauacutejo et al 2006) Although some suggest that reptiles (and to agreater extent amphibians) will be strongly affected by climatechange (eg Gibbons et al 2000 Arauacutejo et al 2006 Carvalho et al2010) very few of themwere concerned with arid environments orthe Middle East We hope that this and other similar studies thatestimate the fate of reptiles under climate change help decision-making authorities to take the measures necessary to conservethem

Acknowledgements

We thank the Italian Cooperation (Debt Swap) in Egypt forfunding the BioMAP project that allowed the collation of the da-tabases used in this work Dr Tim Newbold for allowing the use ofthe data from his field trips Dr Mustafa Fouda (Former Director ofthe Nature Conservation Sector EEAA) for facilities offered duringBioMAP project and all the BioMAP staff This work was supportedby the award of a Chevening Scholarship Masters grant to AEl-GWe are especially grateful to two anonymous referees whogreatly improved an earlier manuscript

Appendix A Supplementary data

Supplementary data related to this article can be found at httpdxdoiorg101016jjaridenv201512007

References

Addison PFE Rumpff L Bau SS Carey JM Chee YE Jarrad FC McBride MFBurgman MA 2013 Practical solutions for making models indispensable inconservation decision-making Divers Distrib 19 (5e6) 490e502 httpdxdoiorg101111ddi12054

Anderson RP Peterson AT Egbert SL 2006 Vegetation-index models predictareas vulnerable to purple loosestrife (Lythrum salicaria) invasion in KansasSouthwest Nat 51 471e480 httpdxdoiorg1018940038-4909(2006)51[471VMPAVT]20CO2

Arauacutejo MB 2009 Climate change and spatial conservation planning InMoilanen A Wilson KA Possingham HP (Eds) Spatial Conservation Prior-itization Quantitative Methods and Computational Tools Oxford UniversityPress pp 172e184 ISBN 9780199547760

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221220

Arauacutejo MB Rahbek C 2006 How does climate change affect biodiversity Sci-ence 313 1396e1397 httpdxdoiorg101126science1131758

Arauacutejo MB Thuiller W Pearson RG 2006 Climate warming and the decline ofamphibians and reptiles in Europe J Biogeogr 33 1712e1728 httpdxdoiorg101111j1365-2699200601482x

Baha El Din SM 2001 The Herpetofauna of Egypt Species Communities andAssemblages University of Nottingham p 435 Unpublished PhD thesis

Baha El Din SM 2006 A Guide to the Reptiles and Amphibians of Egypt TheAmerican University in Cairo Press ISBN 9789774249792

Baha El Din SM 2007 A new lizard of the Acanthodactylus scutellatus group(Squamata Lacertidae) from Egypt Zool Middle East 40 21e32 httpdxdoiorg10108009397140200710638200

Baldwin RA 2009 Use of maximum entropy modeling in wildlife research En-tropy 11 854e866 httpdxdoiorg103390e11040854

Ball IR Possingham HP Watts M 2009 Marxan and relatives software forspatial conservation prioritisation In Moilanen A Wilson KAPossingham HP (Eds) Spatial Conservation Prioritization QuantitativeMethods and Computational Tools Oxford University Press pp 185e195 ISBN9780199547760

Basuony MI Gilbert F Zalat S 2010 The Mammals of Egypt Atlas Red DataListing amp Conservation BioMAP amp CultNat EEAA amp Bibliotheca AlexandrinaCairo available at httpecologynottinghamacuk~plzfg

Bombi P Salvi D Bologna MA Pettorelli N 2012 Cross-scale predictions allowthe identification of local conservation priorities from atlas data Anim Conserv15 378e387 httpdxdoiorg101111j1469-1795201200526x

Brotons L Thuiller W Arauacutejo MB Hirzel AH 2004 Presence-absence versuspresence-only modelling methods for predicting bird habitat suitability Ecog-raphy 27 437e448 httpdxdoiorg101111j0906-7590200403764x

Cadby CD While GM Hobday AJ Uller T Wapstra E 2010 Multi-scaleapproach to understanding climate effects on offspring size at birth and date ofbirth in a reptile Integr Zool 5 164e175 httpdxdoiorg101111j1749-4877201000201x

Carvalho SB Brito JC Crespo EJ Possingham HP 2010 From climate changepredictions to actions e conserving vulnerable animal groups in hotspots at aregional scale Glob Change Biol 16 (12) 3257e3270 httpdxdoiorg101111j1365-2486201002212x

Currie DJ 2001 Projected effects of climate change on patterns of vertebrate andtree species richness in the conterminous United States Ecosystems 4216e225 httpdxdoiorg101007s10021-001-0005-4

de Pous P 2011 Climate Change and European Reptiles from Large-scale SpatialRange Shift Projections towards a More Realistic Assessment of Species Re-sponses University of East Anglia UK p 64 MSc thesis

de Pous P Beukema W Weterings M Duumlmmer I Geniez P 2011 Area priori-tization and performance evaluation of the conservation area network for theMoroccan herpetofauna a preliminary assessment Biodivers Conserv 2089e118 httpdxdoiorg101007s10531-010-9948-0

Edgar P Foster J Baker J 2010 Reptile Habitat Management HandbookAmphibian and Reptile Conservation Bournemouth

Egbert SL Martiacutenez-Meyer E Ortega-Huerta M Peterson AT 2002 Use ofdatasets derived from time-series AVHRR imagery as surrogates for land covermaps in predicting species distributions In Proceedings IEEE 2002 Interna-tional Geoscience and Remote Sensing Symposium (IGARSS) vol 4pp 2337e2339 httpdxdoiorg101109IGARSS20021026537

El Alqamy H Ismael A Abdelhameed A Nagy A Hamada A et al 2010 Pre-dicting the status and distribution of the Nubian Ibex (Capra nubiana) in thehigh-altitude mountains of south Sinai (Egypt) In Galemys Newsletter of theSpanish Society for the Preservation and Study of Mammals vol 22pp 517e530

Elith J Graham CH Anderson RP Dudiacutek M Ferrier S et al 2006 Novelmethods improve prediction of species distributions from occurrence dataEcography 29 129e151 httpdxdoiorg101111j20060906-759004596x

Elith J Phillips SJ Hastie T Dudiacutek M Chee YE Yates CJ 2011 A statisticalexplanation of MaxEnt for ecologists Divers Distrib 17 (1) 43e57 httpdxdoiorg101111j1472-4642201000725x

FAO 2012 Food and Agriculture Organisation World Development IndicatorsAverage Precipitation in Depth (mm per year) httpdataworldbankorgindicatorAGLNDPRCPMM

Franklin J 2009 Mapping Species Distributions Spatial Inference and PredictionCambridge University Press Cambridge UK ISBN 9780521700023

Franklin J Wejnert KE Hathaway SA Rochester CJ Fisher RN 2009 Effect ofspecies rarity on the accuracy of species distribution models for reptiles andamphibians in southern California Divers Distrib 15 167e177 httpdxdoiorg101111j1472-4642200800536x

Gibbons JW Scott DE Ryan TJ Buhlmann KA Tuberville TD Metts BSGreene JL Mills T et al 2000 The global decline of reptiles Deja Vu Am-phibians BioScience 50 (8) 653e666 httpdxdoiorg1016410006-3568(2000)050[0653TGDORD]20CO2

Gilbert F Zalat S 2008 The Butterflies of Egypt Atlas Red Data Listing amp Con-servation BioMAP EEAA Cairo available at httpecologynottinghamacuk~plzfg

Golicher DJ Cayuela L Newton AC 2012 Effects of climate change on the po-tential species richness of Mesoamerican forests Biotropica 44 284e293httpdxdoiorg101111j1744-7429201100815x

Graham CH Elith J Hijmans RJ Guisan A Peterson AT et al 2008 Theinfluence of spatial errors in species occurrence data used in distribution

models J Appl Ecol 45 239e247 httpdxdoiorg101111j1365-2664200701408x

Guisan A Tingley R Baumgartner JB Naujokaitis-Lewis I Sutcliffe PRTulloch AIT Regan TJ Brotons L et al 2013 Predicting species distributionsfor conservation decisions Ecol Lett 16 (12) 1424e1435 httpdxdoiorg101111ele12189

Hannah L 2010 A global conservation system for climate-change adaptationConserv Biol 24 70e77 httpdxdoiorg101111j1523-1739200901405x

Hannah L 2011 Climate Change Biology Elsevier Ltd ISBN 9780123741820Hannah L Lovejoy TE Schneider SH 2005 Biodiversity and climate change in

context In Lovejoy TE Hannah L (Eds) Climate Change and BiodiversityYale University Press New Haven CT USA

Hannah L Midgley G Andelman S Arauacutejo M Hughes G et al 2007 Protectedarea needs in a changing climate Front Ecol Environ 5 131e138 httpdxdoiorg1018901540-9295(2007)5[131PANIAC]20CO2

Hastie T Fithian W 2013 Inference from presence-only data the ongoing con-troversy Ecography 36 864e867 httpdxdoiorg101111j1600-0587201300321x

Hernandez PA Graham CH Master LL Albert DL 2006 The effect of samplesize and species characteristics on performance of different species distributionmodeling methods Ecography 29 773e785 httpdxdoiorg101111j0906-7590200604700x

Hijmans RJ Cameron SE Parra JL Jones PG Jarvis A 2005 Very high reso-lution interpolated climate surfaces for global land areas Int J Climatol 251965e1978 httpdxdoiorg101002Joc1276

Holt AC Salkeld DJ Fritz CL Tucker JR Gong P 2009 Spatial analysis ofplague in California niche modeling predictions of the current distribution andpotential response to climate change Int J Health Geogr 8 38 httpdxdoiorg1011861476-072X-8-38

Hoyle M James M 2005 Global warming human population pressure andviability of the worlds smallest butterfly Conserv Biol 19 1113e1124 httpdxdoiorg101111j1523-1739200500166x

Ihlow F Dambach J Engler JO Flecks M Hartmann T et al 2012 On the brinkof extinction How climate change may affect global chelonian species richnessand distribution Glob Change Biol 18 1520e1530 httpdxdoiorg101111j1365-2486201102623x

IPCC 2007 Climate Change 2007 Synthesis Report Contribution of WorkingGroups I II and III to the Fourth Assessment Report of the IntergovernmentalPanel on Climate Change IPCC Geneva Switzerland

Kgosiesele E 2010 Predictive Distribution Modelling of Timon Lepida in SpainUniversity of Twente (ITC) the Netherlands p 87 MSc thesis

Klorvuttimontara S McClean CJ Hill JK 2011 Evaluating the effectiveness ofprotected areas for conserving tropical forest butterflies of Thailand BiolConserv 144 2534e2540 httpdxdoiorg101016jbiocon201107012

Leach K Zalat S Gilbert F 2013 Egypts protected area network under futureclimate change Biol Conserv 159 490e500 httpdxdoiorg101016jbiocon201211025

Lobo JM Jimenez-Valverde A Real R 2008 AUC a misleading measure of theperformance of predictive distribution models Glob Ecol Biogeogr 17145e151 httpdxdoiorg101111j1466-8238200700358x

Maiorano L Falcucci A Zimmermann NE Psomas A Pottier J et al 2011 Thefuture of terrestrial mammals in the Mediterranean basin under climatechange Philosophical Trans R Soc Lond Ser B Biol Sci 366 2681e2692httpdxdoiorg101098rstb20110121

Marini MA Barbet-Massin M Lopes LE Jiguet F 2009 Predicted climate-driven bird distribution changes and forecasted conservation conflicts in aneotropical savanna Conserv Biol 23 1558e1567 httpdxdoiorg101111j1523-1739200901258x

Merow C Smith MJ Silander JA 2013 A practical guide to MaxEnt for modelingspecies distributions what it does and why inputs and settings matterEcography 36 (10) 1058e1069 httpdxdoiorg101111j1600-0587201307872x

Moilanen A Meller L Leppeuroanen J Pouzols FM Arponen A et al 2012 Zona-tion Spatial Conservation Planning Framework and Software v31 User Manualwwwhelsinkifibioscienceconsplan

Newbold T 2009 The Value of Species Distribution Models as a Tool for Conser-vation and Ecology in Egypt and Britain University of Nottingham UK p 312PhD thesis

Newbold T Gilbert F Zalat S El-Gabbas A Reader T 2009a Climate-basedmodels of spatial patterns of species richness in Egypts butterfly and mammalfauna J Biogeogr 36 2085e2095 httpdxdoiorg101111j1365-2699200902140x

Newbold T Reader T Zalat S El-Gabbas A Gilbert F 2009b Effect of charac-teristics of butterfly species on the accuracy of distribution models in an aridenvironment Biodivers Conserv 18 3629e3641 httpdxdoiorg101007s10531-009-9668-5

Pandit SN Kolasa J Cottenie K 2009 Contrasts between habitat generalists andspecialists an empirical extension to the basic metacommunity frameworkEcology 90 2253e2262 httpdxdoiorg10189008-08511

Pearson RG Dawson TP 2003 Predicting the impacts of climate change on thedistribution of species are bioclimate envelope models useful Glob EcolBiogeogr 12 361e371 httpdxdoiorg101046j1466-822X200300042x

Pearson RG Raxworthy CJ Nakamura M Peterson AT 2007 Predicting speciesdistributions from small numbers of occurrence records a test case using

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 221

cryptic geckos in Madagascar J Biogeogr 34 102e117 httpdxdoiorg101111j1365-2699200601594x

Phillips SJ ATampT Research 2011 A Brief Tutorial on Maxent Available at httpwwwcsprincetonedu~schapiremaxenttutorialtutorialdoc (Last accessed290512)

Phillips SJ Anderson RP Schapire RE 2006 Maximum entropy modeling ofspecies geographic distributions Ecol Model 190 231e259 httpdxdoiorg101016jecolmodel200503026

Phillips SJ Dudiacutek M Elith J Graham CH Lehmann A Leathwick J Ferrier S2009 Sample selection bias and presence-only distribution models implica-tions for background and pseudo-absence data Ecol Appl 19 181e197 httpdxdoiorg10189007-21531

R Development Core Team 2012 R a Language and Environment for StatisticalComputing R Foundation for Statistical Computing Vienna Austria ISBN 3-900051-07-0 httpwwwR-projectorg

Radosavljevic A Anderson RP 2014 Making better Maxent models of speciesdistributions complexity overfitting and evaluation J Biogeogr 41 (4)629e643 httpdxdoiorg101111jbi12227

Royle JA Chandler RB Yackulic C Nichols JD 2012 Likelihood analysis ofspecies occurrence probability from presence-only data for modelling speciesdistributions Methods Ecol Evol 3 545e554 httpdxdoiorg101111j2041-210X201100182x

Sauer J Domisch S Nowak C Haase P 2011 Lowmountain ranges summit trapsfor montane freshwater species under climate change Biodivers Conserv 203133e3146 httpdxdoiorg101007s10531-011-0140-y

Songer M Delion M Biggs A Huang Q 2012 Modeling impacts of climatechange on Giant Panda Habitat Int J Ecol 2012 1e12 httpdxdoiorg1011552012108752

Soultan AE 2011 Antelope distribution in Egypt Modeling the Impact of ClimateChange on Antelopes Distribution in Egypt Lambert Academic Publishing

Standards IUCN Subcommittee Petitions 2010 Guidelines for Using the IUCNRed List Categories and Criteria Version 81

Stockwell DRB Peterson AT 2002 Effects of sample size on accuracy of speciesdistribution models Ecol Model 148 1e13 httpdxdoiorg101016S0304-3800(01)00388-X

Suarez AV Tsutsui ND 2004 The value of museum collections for research andsociety Bioscience 54 66 httpdxdoiorg1016410006-3568(2004)054[0066tvomcf]20co2

Taheri S 2010 Hyper Temporal NDVI Images for Modelling and Prediction theHabitat Distribution of Balkan Green Lizard (Lacerta Trilineata) e Case Study

Crete (Greece) University of Twente (ITC) the Netherlands p 54 MScthesis

Taubmann J Theissinger K Feldheim KA Laube I Graf W et al 2011Modelling range shifts and assessing genetic diversity distribution of themontane aquatic mayfly Ameletus inopinatus in Europe under climate changescenarios Conserv Genet 12 503e515 httpdxdoiorg101007s10592-010-0157-x

Teodoro AC Sillero N Alves S Duarte L 2013 Correlation between the habitatsproductivity and species richness (amphibians and reptiles) in Portugal throughremote sensed data In Proc SPIE 8887 Remote Sensing for Agriculture Eco-systems and Hydrology httpdxdoiorg101117122028502

Thomas CD Cameron A Green RE Bakkenes M Beaumont LJ et al 2004Extinction risk from climate change Nature 427 145e148 httpdxdoiorg101038nature02121

Thuiller W 2004 Patterns and uncertainties of species range shifts under climatechange Glob Change Biol 10 2020e2027 httpdxdoiorg101111j1365-2486200400859x

Thuiller W 2007 Biodiversity climate change and the ecologist Nature 448550e552 httpdxdoiorg101038448550a

Thuiller W Lavorel S Arauacutejo MB Sykes MT Prentice IC 2005 Climate changethreats to plant diversity in Europe Proc Natl Acad Sci U S A 1028245e8250 httpdxdoiorg101073pnas0409902102

Thuiller W Lafourcade B Engler R Arauacutejo MB 2009 BIOMOD e a platform forensemble forecasting of species distributions Ecography 32 (3) 369e373httpdxdoiorg101111j1600-0587200805742x

Tolba MK Saab NW 2009 Arab environment Climate Change - Impact ofClimate Change on Arab Countries Arab Forum for Environment andDvelopment

Trotta-Moreu N Lobo JM 2010 Deriving the species richness distribution ofGeotrupinae (Coleoptera Scarabaeoidea) in Mexico from the overlap of indi-vidual model predictions Environ Entomol 39 42e49 httpdxdoiorg101603EN08179

Warren DL Seifert SN 2011 Ecological niche modeling in Maxent the impor-tance of model complexity and the performance of model selection criteriaEcol Appl 21 (2) 335e342 httpdxdoiorg10189010-11711

Wilson KA Cabeza M Klein CJ 2009 Fundamental concepts of spatial con-servation prioritisation In Moilanen A Wilson KA Possingham HP (Eds)Spatial Conservation Prioritization Quantitative Methods and ComputationalTools Oxford University Press pp 16e27 ISBN 9780199547760

Page 8: Journal of Arid Environments - elgabbas.github.io · e Faculty of Sciences & Arts, Taibah University, Al-Ula Campus, Saudi Arabia f Department of Zoology, Suez Canal University, Ismailia,

Fig 4 Current conservation prioritizationl of areas according to the Zonation algo-rithm under two cell removal rules (a) core-area function (identifies areas where oneor more species have important occurrences) and (b) additive benefit function (whichgives greater emphasis to areas with high species richness) Darker colours representhigher conservation value (For interpretation of the references to colour in this figurelegend the reader is referred to the web version of this article)

Fig 5 Mean prioritization value for conservation (plusmn95 confidence limits) derivedfrom Zonation for Protected Areas (PAs) and outside Protected Areas (non-PAs) forcurrent and expected future conditions using the cell-removal rule of (a) core-areafunction and (b) additive benefit function

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221218

schneiderii Malpolon moilensis Ptyodactylus guttatusP siphonorhina and Trapelus mutabilis) According to the currentassessment twenty-seven species are classified as Endangered(loss 50e80) in at least one projection (mostly by 2080 in the A2scenario) two of these are classified assuming no dispersal asEndangered in all future projections (Uromastyx aegyptia and Pris-turus flavipunctatus for details see Tables S2eS3)

Reptiles are also likely to gain from global warming especially inmore northern areas of the globe if they can disperse and expandtheir distributions but under perhaps the more realistic assump-tion of no dispersal most species were expected to lose habitatsuitability by 2050 (Arauacutejo et al 2006) In Egypt areas of clearfuture expected gain (assuming unlimited dispersal) are the areabetween the Siwa oasis and Qattara Depression northwards almostto the northern coastal strip and some sparse locations in centralSinai There were greater gains under the B2a scenario in theeastern part of the Western Desert and some patchy areas aroundthe Nile (Fig 3b) This pattern is similar to that reported by Araujoet al (2006)

Some of the species identified here as being most at risk due tofuture climate change are in fact resilient species that are ecolog-ically very adaptable and are currently experiencing range expan-sion and population growth at least in Egypt (eg H robustus) This

disparity is probably because modelling is based solely on theprobable impacts of climate change on the spatial distribution ofsuitable habitats without taking evolution or ecological adapt-ability and resilience into account Species vary greatly in theirresponse to ecological change some adapting very well even tovery rapid anthropogenically induced changes (eg invasive spe-cies) while others are much less adaptable and tend to go extinctfirst

There are no published studies that discuss the effect of climatechange on Egyptian reptiles obviating any comparisons Informa-tion on the reptiles of the adjacent countries of Sudan and Libya arevery limited making it impossible to assess any movements thatmight compensate for Egyptian losses The patterns of expectedchanges in Leach et als (2013) similar study on mammals andbutterflies do not concur with ours Mammal species richness wasexpected to decline in the Mediterranean and Red Sea areas (by40e60) and increase elsewhere (by 80e100) while for butter-flies declines are expected over almost the whole of Egypt (by40e60) except the south which is expected to increase (by40e60) In the models of European reptiles (de Pous 2011)

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 219

between eight (unlimited dispersal) and 21 (no dispersal) specieswere expected to become extinct in at least one future projectionwith two more losing more than 80 of their currently suitablehabitats This is a much more damaging scenario than the expectedresponses of Egyptian reptiles in our study There are no othercomparable studies from North Africa or the Middle East

44 Area prioritization for conservation

In our study Protected Areas (Fig S3) were more valuable inconserving reptiles than areas outside them both currently and inthe future although the difference is expected to get smaller withtime (Fig 5) Although this is encouraging the Protected Areanetwork appears inadequate to conserve Egyptian reptiles NewProtected Areas that have already been proposed would go someway towards reducing this deficiency in coverage although otherswill be needed in the middle to north Sinai (especially aroundGebel El-Hallal) the Suez Canal area both sides of the Suez Gulf thenorthern Mediterranean coast coastal wadis of the Red Sea be-tween Hurghada and Mersa Alam and the Gebel El-Gallala areaMore effective protection is required in the established ProtectedAreas including those in South Sinai Gebel Elba and Siwa oasis dueto their relatively high expected future suitability loss The Pro-tected Areas of Siwa oasis and El-Gilf El-Kebir plus the areas ofWadiEl-Natrun the Gebel El-Hallal area and wadis of the Red Sea areexpected to have particularly high turnover in species composition(Fig S6) and hence probably need more conservation efforts Somecomparable information is available for Egypts butterflies andmammals (Gilbert and Zalat 2008 Basuony et al 2010 Leach et al2013) Egypts network of Protected Areas appears adequate toconserve Egyptian butterfly hotspots except for the Mediterraneancoast between Alexandria and Sallum (Gilbert and Zalat 2008) butinadequate to conserve important mammal areas with new Pro-tected Areas needed to be constructed in the lower Nile Valleyalong the north coast between Alexandria and Sallum top part ofthe Suez Gulf and perhaps the Qattara Depression (Basuony et al2010) Suggestions for conserving mammals are consistent to someextent with our results which also suggest new Protected Areasalong the Suez Gulf Mediterranean coast from Mersa Matruh toSallum and the Qattara Depression Leach et al (2013) concur withour results in suggesting priority areas along the western Medi-terranean coast around Greater Cairo central and north Sinai andthe Red Sea coast

45 Limitations

There are some limitations of the results shown in this study 1)Egyptian climate data in WorldClim are interpolated from ratherfew non-randomly placed weather stations in Egypt and adjacentcountries Although making the climate data less reliable thanelsewhere there are no alternative climate data available 2) Anensemble approach (eg using BIOMOD Thuiller et al 2009) isoften suggested for SDM modelling to avoid dependence on oneparticular algorithm This is not possible without a greater numberof records since other algorithms are not as flexible or robust asMaxent Despite all the caveats with using SDMs and criticisms ofMaxent in particular (Royle et al 2012 but see Hastie and Fithian2013 Elith et al 2006 Hernandez et al 2006) in practice coun-tries such as Egypt have little else to guide them in attempts atclimate change adaptation 3) There were signs of sampling bias inthe available records Although this violates the assumptions of allSDMs it is a problem common to nearly all datasets worldwideNew techniques may allow for the correction of sampling bias egusing target-group approach (Phillips et al 2009) although thedata requirements are currently impossible for Egypt and most

other developing countries) 4) Using Maxents default settings iscriticized by some authors they need to be modified dependent onthe data (Warren and Seifert 2011 Radosavljevic and Anderson2014 Merow et al 2013) but we think using the default settingsis enough in this study

5 Conclusion

Although Egyptian datasets exist knowledge of Egyptian spe-cies and conservation priorities are limited (with consequencesexplored in Leach et al 2013) nevertheless our results and those ofLeach et al (2013) encourage us to believe that useful insights canbe gained for conservation priorities by using the combination ofspecies distribution modelling plus spatial conservation prioriti-zation techniques (Hannah 2010 Klorvuttimontara et al 2011) Itis crucial for the Egyptian Environmental Affairs Agency to establishan accurate and reliable database of records assessed by experts tohelp improve the basic information for future conservation as-sessments and analyses

Although many SDM studies have been done on a variety oforganisms in recent decades aiming to solve conservation prob-lems evidence of their practical utility in real-world conservation issparse with very few studies clearly targeting conservation de-cisions (Guisan et al 2013) If implemented correctly SDMs canplay a critical role in supporting spatial conservation decision-making (Addison et al 2013) For reptiles there is a shortage ofstudies estimating their response to expected climate changes(Arauacutejo et al 2006) Although some suggest that reptiles (and to agreater extent amphibians) will be strongly affected by climatechange (eg Gibbons et al 2000 Arauacutejo et al 2006 Carvalho et al2010) very few of themwere concerned with arid environments orthe Middle East We hope that this and other similar studies thatestimate the fate of reptiles under climate change help decision-making authorities to take the measures necessary to conservethem

Acknowledgements

We thank the Italian Cooperation (Debt Swap) in Egypt forfunding the BioMAP project that allowed the collation of the da-tabases used in this work Dr Tim Newbold for allowing the use ofthe data from his field trips Dr Mustafa Fouda (Former Director ofthe Nature Conservation Sector EEAA) for facilities offered duringBioMAP project and all the BioMAP staff This work was supportedby the award of a Chevening Scholarship Masters grant to AEl-GWe are especially grateful to two anonymous referees whogreatly improved an earlier manuscript

Appendix A Supplementary data

Supplementary data related to this article can be found at httpdxdoiorg101016jjaridenv201512007

References

Addison PFE Rumpff L Bau SS Carey JM Chee YE Jarrad FC McBride MFBurgman MA 2013 Practical solutions for making models indispensable inconservation decision-making Divers Distrib 19 (5e6) 490e502 httpdxdoiorg101111ddi12054

Anderson RP Peterson AT Egbert SL 2006 Vegetation-index models predictareas vulnerable to purple loosestrife (Lythrum salicaria) invasion in KansasSouthwest Nat 51 471e480 httpdxdoiorg1018940038-4909(2006)51[471VMPAVT]20CO2

Arauacutejo MB 2009 Climate change and spatial conservation planning InMoilanen A Wilson KA Possingham HP (Eds) Spatial Conservation Prior-itization Quantitative Methods and Computational Tools Oxford UniversityPress pp 172e184 ISBN 9780199547760

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221220

Arauacutejo MB Rahbek C 2006 How does climate change affect biodiversity Sci-ence 313 1396e1397 httpdxdoiorg101126science1131758

Arauacutejo MB Thuiller W Pearson RG 2006 Climate warming and the decline ofamphibians and reptiles in Europe J Biogeogr 33 1712e1728 httpdxdoiorg101111j1365-2699200601482x

Baha El Din SM 2001 The Herpetofauna of Egypt Species Communities andAssemblages University of Nottingham p 435 Unpublished PhD thesis

Baha El Din SM 2006 A Guide to the Reptiles and Amphibians of Egypt TheAmerican University in Cairo Press ISBN 9789774249792

Baha El Din SM 2007 A new lizard of the Acanthodactylus scutellatus group(Squamata Lacertidae) from Egypt Zool Middle East 40 21e32 httpdxdoiorg10108009397140200710638200

Baldwin RA 2009 Use of maximum entropy modeling in wildlife research En-tropy 11 854e866 httpdxdoiorg103390e11040854

Ball IR Possingham HP Watts M 2009 Marxan and relatives software forspatial conservation prioritisation In Moilanen A Wilson KAPossingham HP (Eds) Spatial Conservation Prioritization QuantitativeMethods and Computational Tools Oxford University Press pp 185e195 ISBN9780199547760

Basuony MI Gilbert F Zalat S 2010 The Mammals of Egypt Atlas Red DataListing amp Conservation BioMAP amp CultNat EEAA amp Bibliotheca AlexandrinaCairo available at httpecologynottinghamacuk~plzfg

Bombi P Salvi D Bologna MA Pettorelli N 2012 Cross-scale predictions allowthe identification of local conservation priorities from atlas data Anim Conserv15 378e387 httpdxdoiorg101111j1469-1795201200526x

Brotons L Thuiller W Arauacutejo MB Hirzel AH 2004 Presence-absence versuspresence-only modelling methods for predicting bird habitat suitability Ecog-raphy 27 437e448 httpdxdoiorg101111j0906-7590200403764x

Cadby CD While GM Hobday AJ Uller T Wapstra E 2010 Multi-scaleapproach to understanding climate effects on offspring size at birth and date ofbirth in a reptile Integr Zool 5 164e175 httpdxdoiorg101111j1749-4877201000201x

Carvalho SB Brito JC Crespo EJ Possingham HP 2010 From climate changepredictions to actions e conserving vulnerable animal groups in hotspots at aregional scale Glob Change Biol 16 (12) 3257e3270 httpdxdoiorg101111j1365-2486201002212x

Currie DJ 2001 Projected effects of climate change on patterns of vertebrate andtree species richness in the conterminous United States Ecosystems 4216e225 httpdxdoiorg101007s10021-001-0005-4

de Pous P 2011 Climate Change and European Reptiles from Large-scale SpatialRange Shift Projections towards a More Realistic Assessment of Species Re-sponses University of East Anglia UK p 64 MSc thesis

de Pous P Beukema W Weterings M Duumlmmer I Geniez P 2011 Area priori-tization and performance evaluation of the conservation area network for theMoroccan herpetofauna a preliminary assessment Biodivers Conserv 2089e118 httpdxdoiorg101007s10531-010-9948-0

Edgar P Foster J Baker J 2010 Reptile Habitat Management HandbookAmphibian and Reptile Conservation Bournemouth

Egbert SL Martiacutenez-Meyer E Ortega-Huerta M Peterson AT 2002 Use ofdatasets derived from time-series AVHRR imagery as surrogates for land covermaps in predicting species distributions In Proceedings IEEE 2002 Interna-tional Geoscience and Remote Sensing Symposium (IGARSS) vol 4pp 2337e2339 httpdxdoiorg101109IGARSS20021026537

El Alqamy H Ismael A Abdelhameed A Nagy A Hamada A et al 2010 Pre-dicting the status and distribution of the Nubian Ibex (Capra nubiana) in thehigh-altitude mountains of south Sinai (Egypt) In Galemys Newsletter of theSpanish Society for the Preservation and Study of Mammals vol 22pp 517e530

Elith J Graham CH Anderson RP Dudiacutek M Ferrier S et al 2006 Novelmethods improve prediction of species distributions from occurrence dataEcography 29 129e151 httpdxdoiorg101111j20060906-759004596x

Elith J Phillips SJ Hastie T Dudiacutek M Chee YE Yates CJ 2011 A statisticalexplanation of MaxEnt for ecologists Divers Distrib 17 (1) 43e57 httpdxdoiorg101111j1472-4642201000725x

FAO 2012 Food and Agriculture Organisation World Development IndicatorsAverage Precipitation in Depth (mm per year) httpdataworldbankorgindicatorAGLNDPRCPMM

Franklin J 2009 Mapping Species Distributions Spatial Inference and PredictionCambridge University Press Cambridge UK ISBN 9780521700023

Franklin J Wejnert KE Hathaway SA Rochester CJ Fisher RN 2009 Effect ofspecies rarity on the accuracy of species distribution models for reptiles andamphibians in southern California Divers Distrib 15 167e177 httpdxdoiorg101111j1472-4642200800536x

Gibbons JW Scott DE Ryan TJ Buhlmann KA Tuberville TD Metts BSGreene JL Mills T et al 2000 The global decline of reptiles Deja Vu Am-phibians BioScience 50 (8) 653e666 httpdxdoiorg1016410006-3568(2000)050[0653TGDORD]20CO2

Gilbert F Zalat S 2008 The Butterflies of Egypt Atlas Red Data Listing amp Con-servation BioMAP EEAA Cairo available at httpecologynottinghamacuk~plzfg

Golicher DJ Cayuela L Newton AC 2012 Effects of climate change on the po-tential species richness of Mesoamerican forests Biotropica 44 284e293httpdxdoiorg101111j1744-7429201100815x

Graham CH Elith J Hijmans RJ Guisan A Peterson AT et al 2008 Theinfluence of spatial errors in species occurrence data used in distribution

models J Appl Ecol 45 239e247 httpdxdoiorg101111j1365-2664200701408x

Guisan A Tingley R Baumgartner JB Naujokaitis-Lewis I Sutcliffe PRTulloch AIT Regan TJ Brotons L et al 2013 Predicting species distributionsfor conservation decisions Ecol Lett 16 (12) 1424e1435 httpdxdoiorg101111ele12189

Hannah L 2010 A global conservation system for climate-change adaptationConserv Biol 24 70e77 httpdxdoiorg101111j1523-1739200901405x

Hannah L 2011 Climate Change Biology Elsevier Ltd ISBN 9780123741820Hannah L Lovejoy TE Schneider SH 2005 Biodiversity and climate change in

context In Lovejoy TE Hannah L (Eds) Climate Change and BiodiversityYale University Press New Haven CT USA

Hannah L Midgley G Andelman S Arauacutejo M Hughes G et al 2007 Protectedarea needs in a changing climate Front Ecol Environ 5 131e138 httpdxdoiorg1018901540-9295(2007)5[131PANIAC]20CO2

Hastie T Fithian W 2013 Inference from presence-only data the ongoing con-troversy Ecography 36 864e867 httpdxdoiorg101111j1600-0587201300321x

Hernandez PA Graham CH Master LL Albert DL 2006 The effect of samplesize and species characteristics on performance of different species distributionmodeling methods Ecography 29 773e785 httpdxdoiorg101111j0906-7590200604700x

Hijmans RJ Cameron SE Parra JL Jones PG Jarvis A 2005 Very high reso-lution interpolated climate surfaces for global land areas Int J Climatol 251965e1978 httpdxdoiorg101002Joc1276

Holt AC Salkeld DJ Fritz CL Tucker JR Gong P 2009 Spatial analysis ofplague in California niche modeling predictions of the current distribution andpotential response to climate change Int J Health Geogr 8 38 httpdxdoiorg1011861476-072X-8-38

Hoyle M James M 2005 Global warming human population pressure andviability of the worlds smallest butterfly Conserv Biol 19 1113e1124 httpdxdoiorg101111j1523-1739200500166x

Ihlow F Dambach J Engler JO Flecks M Hartmann T et al 2012 On the brinkof extinction How climate change may affect global chelonian species richnessand distribution Glob Change Biol 18 1520e1530 httpdxdoiorg101111j1365-2486201102623x

IPCC 2007 Climate Change 2007 Synthesis Report Contribution of WorkingGroups I II and III to the Fourth Assessment Report of the IntergovernmentalPanel on Climate Change IPCC Geneva Switzerland

Kgosiesele E 2010 Predictive Distribution Modelling of Timon Lepida in SpainUniversity of Twente (ITC) the Netherlands p 87 MSc thesis

Klorvuttimontara S McClean CJ Hill JK 2011 Evaluating the effectiveness ofprotected areas for conserving tropical forest butterflies of Thailand BiolConserv 144 2534e2540 httpdxdoiorg101016jbiocon201107012

Leach K Zalat S Gilbert F 2013 Egypts protected area network under futureclimate change Biol Conserv 159 490e500 httpdxdoiorg101016jbiocon201211025

Lobo JM Jimenez-Valverde A Real R 2008 AUC a misleading measure of theperformance of predictive distribution models Glob Ecol Biogeogr 17145e151 httpdxdoiorg101111j1466-8238200700358x

Maiorano L Falcucci A Zimmermann NE Psomas A Pottier J et al 2011 Thefuture of terrestrial mammals in the Mediterranean basin under climatechange Philosophical Trans R Soc Lond Ser B Biol Sci 366 2681e2692httpdxdoiorg101098rstb20110121

Marini MA Barbet-Massin M Lopes LE Jiguet F 2009 Predicted climate-driven bird distribution changes and forecasted conservation conflicts in aneotropical savanna Conserv Biol 23 1558e1567 httpdxdoiorg101111j1523-1739200901258x

Merow C Smith MJ Silander JA 2013 A practical guide to MaxEnt for modelingspecies distributions what it does and why inputs and settings matterEcography 36 (10) 1058e1069 httpdxdoiorg101111j1600-0587201307872x

Moilanen A Meller L Leppeuroanen J Pouzols FM Arponen A et al 2012 Zona-tion Spatial Conservation Planning Framework and Software v31 User Manualwwwhelsinkifibioscienceconsplan

Newbold T 2009 The Value of Species Distribution Models as a Tool for Conser-vation and Ecology in Egypt and Britain University of Nottingham UK p 312PhD thesis

Newbold T Gilbert F Zalat S El-Gabbas A Reader T 2009a Climate-basedmodels of spatial patterns of species richness in Egypts butterfly and mammalfauna J Biogeogr 36 2085e2095 httpdxdoiorg101111j1365-2699200902140x

Newbold T Reader T Zalat S El-Gabbas A Gilbert F 2009b Effect of charac-teristics of butterfly species on the accuracy of distribution models in an aridenvironment Biodivers Conserv 18 3629e3641 httpdxdoiorg101007s10531-009-9668-5

Pandit SN Kolasa J Cottenie K 2009 Contrasts between habitat generalists andspecialists an empirical extension to the basic metacommunity frameworkEcology 90 2253e2262 httpdxdoiorg10189008-08511

Pearson RG Dawson TP 2003 Predicting the impacts of climate change on thedistribution of species are bioclimate envelope models useful Glob EcolBiogeogr 12 361e371 httpdxdoiorg101046j1466-822X200300042x

Pearson RG Raxworthy CJ Nakamura M Peterson AT 2007 Predicting speciesdistributions from small numbers of occurrence records a test case using

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 221

cryptic geckos in Madagascar J Biogeogr 34 102e117 httpdxdoiorg101111j1365-2699200601594x

Phillips SJ ATampT Research 2011 A Brief Tutorial on Maxent Available at httpwwwcsprincetonedu~schapiremaxenttutorialtutorialdoc (Last accessed290512)

Phillips SJ Anderson RP Schapire RE 2006 Maximum entropy modeling ofspecies geographic distributions Ecol Model 190 231e259 httpdxdoiorg101016jecolmodel200503026

Phillips SJ Dudiacutek M Elith J Graham CH Lehmann A Leathwick J Ferrier S2009 Sample selection bias and presence-only distribution models implica-tions for background and pseudo-absence data Ecol Appl 19 181e197 httpdxdoiorg10189007-21531

R Development Core Team 2012 R a Language and Environment for StatisticalComputing R Foundation for Statistical Computing Vienna Austria ISBN 3-900051-07-0 httpwwwR-projectorg

Radosavljevic A Anderson RP 2014 Making better Maxent models of speciesdistributions complexity overfitting and evaluation J Biogeogr 41 (4)629e643 httpdxdoiorg101111jbi12227

Royle JA Chandler RB Yackulic C Nichols JD 2012 Likelihood analysis ofspecies occurrence probability from presence-only data for modelling speciesdistributions Methods Ecol Evol 3 545e554 httpdxdoiorg101111j2041-210X201100182x

Sauer J Domisch S Nowak C Haase P 2011 Lowmountain ranges summit trapsfor montane freshwater species under climate change Biodivers Conserv 203133e3146 httpdxdoiorg101007s10531-011-0140-y

Songer M Delion M Biggs A Huang Q 2012 Modeling impacts of climatechange on Giant Panda Habitat Int J Ecol 2012 1e12 httpdxdoiorg1011552012108752

Soultan AE 2011 Antelope distribution in Egypt Modeling the Impact of ClimateChange on Antelopes Distribution in Egypt Lambert Academic Publishing

Standards IUCN Subcommittee Petitions 2010 Guidelines for Using the IUCNRed List Categories and Criteria Version 81

Stockwell DRB Peterson AT 2002 Effects of sample size on accuracy of speciesdistribution models Ecol Model 148 1e13 httpdxdoiorg101016S0304-3800(01)00388-X

Suarez AV Tsutsui ND 2004 The value of museum collections for research andsociety Bioscience 54 66 httpdxdoiorg1016410006-3568(2004)054[0066tvomcf]20co2

Taheri S 2010 Hyper Temporal NDVI Images for Modelling and Prediction theHabitat Distribution of Balkan Green Lizard (Lacerta Trilineata) e Case Study

Crete (Greece) University of Twente (ITC) the Netherlands p 54 MScthesis

Taubmann J Theissinger K Feldheim KA Laube I Graf W et al 2011Modelling range shifts and assessing genetic diversity distribution of themontane aquatic mayfly Ameletus inopinatus in Europe under climate changescenarios Conserv Genet 12 503e515 httpdxdoiorg101007s10592-010-0157-x

Teodoro AC Sillero N Alves S Duarte L 2013 Correlation between the habitatsproductivity and species richness (amphibians and reptiles) in Portugal throughremote sensed data In Proc SPIE 8887 Remote Sensing for Agriculture Eco-systems and Hydrology httpdxdoiorg101117122028502

Thomas CD Cameron A Green RE Bakkenes M Beaumont LJ et al 2004Extinction risk from climate change Nature 427 145e148 httpdxdoiorg101038nature02121

Thuiller W 2004 Patterns and uncertainties of species range shifts under climatechange Glob Change Biol 10 2020e2027 httpdxdoiorg101111j1365-2486200400859x

Thuiller W 2007 Biodiversity climate change and the ecologist Nature 448550e552 httpdxdoiorg101038448550a

Thuiller W Lavorel S Arauacutejo MB Sykes MT Prentice IC 2005 Climate changethreats to plant diversity in Europe Proc Natl Acad Sci U S A 1028245e8250 httpdxdoiorg101073pnas0409902102

Thuiller W Lafourcade B Engler R Arauacutejo MB 2009 BIOMOD e a platform forensemble forecasting of species distributions Ecography 32 (3) 369e373httpdxdoiorg101111j1600-0587200805742x

Tolba MK Saab NW 2009 Arab environment Climate Change - Impact ofClimate Change on Arab Countries Arab Forum for Environment andDvelopment

Trotta-Moreu N Lobo JM 2010 Deriving the species richness distribution ofGeotrupinae (Coleoptera Scarabaeoidea) in Mexico from the overlap of indi-vidual model predictions Environ Entomol 39 42e49 httpdxdoiorg101603EN08179

Warren DL Seifert SN 2011 Ecological niche modeling in Maxent the impor-tance of model complexity and the performance of model selection criteriaEcol Appl 21 (2) 335e342 httpdxdoiorg10189010-11711

Wilson KA Cabeza M Klein CJ 2009 Fundamental concepts of spatial con-servation prioritisation In Moilanen A Wilson KA Possingham HP (Eds)Spatial Conservation Prioritization Quantitative Methods and ComputationalTools Oxford University Press pp 16e27 ISBN 9780199547760

Page 9: Journal of Arid Environments - elgabbas.github.io · e Faculty of Sciences & Arts, Taibah University, Al-Ula Campus, Saudi Arabia f Department of Zoology, Suez Canal University, Ismailia,

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 219

between eight (unlimited dispersal) and 21 (no dispersal) specieswere expected to become extinct in at least one future projectionwith two more losing more than 80 of their currently suitablehabitats This is a much more damaging scenario than the expectedresponses of Egyptian reptiles in our study There are no othercomparable studies from North Africa or the Middle East

44 Area prioritization for conservation

In our study Protected Areas (Fig S3) were more valuable inconserving reptiles than areas outside them both currently and inthe future although the difference is expected to get smaller withtime (Fig 5) Although this is encouraging the Protected Areanetwork appears inadequate to conserve Egyptian reptiles NewProtected Areas that have already been proposed would go someway towards reducing this deficiency in coverage although otherswill be needed in the middle to north Sinai (especially aroundGebel El-Hallal) the Suez Canal area both sides of the Suez Gulf thenorthern Mediterranean coast coastal wadis of the Red Sea be-tween Hurghada and Mersa Alam and the Gebel El-Gallala areaMore effective protection is required in the established ProtectedAreas including those in South Sinai Gebel Elba and Siwa oasis dueto their relatively high expected future suitability loss The Pro-tected Areas of Siwa oasis and El-Gilf El-Kebir plus the areas ofWadiEl-Natrun the Gebel El-Hallal area and wadis of the Red Sea areexpected to have particularly high turnover in species composition(Fig S6) and hence probably need more conservation efforts Somecomparable information is available for Egypts butterflies andmammals (Gilbert and Zalat 2008 Basuony et al 2010 Leach et al2013) Egypts network of Protected Areas appears adequate toconserve Egyptian butterfly hotspots except for the Mediterraneancoast between Alexandria and Sallum (Gilbert and Zalat 2008) butinadequate to conserve important mammal areas with new Pro-tected Areas needed to be constructed in the lower Nile Valleyalong the north coast between Alexandria and Sallum top part ofthe Suez Gulf and perhaps the Qattara Depression (Basuony et al2010) Suggestions for conserving mammals are consistent to someextent with our results which also suggest new Protected Areasalong the Suez Gulf Mediterranean coast from Mersa Matruh toSallum and the Qattara Depression Leach et al (2013) concur withour results in suggesting priority areas along the western Medi-terranean coast around Greater Cairo central and north Sinai andthe Red Sea coast

45 Limitations

There are some limitations of the results shown in this study 1)Egyptian climate data in WorldClim are interpolated from ratherfew non-randomly placed weather stations in Egypt and adjacentcountries Although making the climate data less reliable thanelsewhere there are no alternative climate data available 2) Anensemble approach (eg using BIOMOD Thuiller et al 2009) isoften suggested for SDM modelling to avoid dependence on oneparticular algorithm This is not possible without a greater numberof records since other algorithms are not as flexible or robust asMaxent Despite all the caveats with using SDMs and criticisms ofMaxent in particular (Royle et al 2012 but see Hastie and Fithian2013 Elith et al 2006 Hernandez et al 2006) in practice coun-tries such as Egypt have little else to guide them in attempts atclimate change adaptation 3) There were signs of sampling bias inthe available records Although this violates the assumptions of allSDMs it is a problem common to nearly all datasets worldwideNew techniques may allow for the correction of sampling bias egusing target-group approach (Phillips et al 2009) although thedata requirements are currently impossible for Egypt and most

other developing countries) 4) Using Maxents default settings iscriticized by some authors they need to be modified dependent onthe data (Warren and Seifert 2011 Radosavljevic and Anderson2014 Merow et al 2013) but we think using the default settingsis enough in this study

5 Conclusion

Although Egyptian datasets exist knowledge of Egyptian spe-cies and conservation priorities are limited (with consequencesexplored in Leach et al 2013) nevertheless our results and those ofLeach et al (2013) encourage us to believe that useful insights canbe gained for conservation priorities by using the combination ofspecies distribution modelling plus spatial conservation prioriti-zation techniques (Hannah 2010 Klorvuttimontara et al 2011) Itis crucial for the Egyptian Environmental Affairs Agency to establishan accurate and reliable database of records assessed by experts tohelp improve the basic information for future conservation as-sessments and analyses

Although many SDM studies have been done on a variety oforganisms in recent decades aiming to solve conservation prob-lems evidence of their practical utility in real-world conservation issparse with very few studies clearly targeting conservation de-cisions (Guisan et al 2013) If implemented correctly SDMs canplay a critical role in supporting spatial conservation decision-making (Addison et al 2013) For reptiles there is a shortage ofstudies estimating their response to expected climate changes(Arauacutejo et al 2006) Although some suggest that reptiles (and to agreater extent amphibians) will be strongly affected by climatechange (eg Gibbons et al 2000 Arauacutejo et al 2006 Carvalho et al2010) very few of themwere concerned with arid environments orthe Middle East We hope that this and other similar studies thatestimate the fate of reptiles under climate change help decision-making authorities to take the measures necessary to conservethem

Acknowledgements

We thank the Italian Cooperation (Debt Swap) in Egypt forfunding the BioMAP project that allowed the collation of the da-tabases used in this work Dr Tim Newbold for allowing the use ofthe data from his field trips Dr Mustafa Fouda (Former Director ofthe Nature Conservation Sector EEAA) for facilities offered duringBioMAP project and all the BioMAP staff This work was supportedby the award of a Chevening Scholarship Masters grant to AEl-GWe are especially grateful to two anonymous referees whogreatly improved an earlier manuscript

Appendix A Supplementary data

Supplementary data related to this article can be found at httpdxdoiorg101016jjaridenv201512007

References

Addison PFE Rumpff L Bau SS Carey JM Chee YE Jarrad FC McBride MFBurgman MA 2013 Practical solutions for making models indispensable inconservation decision-making Divers Distrib 19 (5e6) 490e502 httpdxdoiorg101111ddi12054

Anderson RP Peterson AT Egbert SL 2006 Vegetation-index models predictareas vulnerable to purple loosestrife (Lythrum salicaria) invasion in KansasSouthwest Nat 51 471e480 httpdxdoiorg1018940038-4909(2006)51[471VMPAVT]20CO2

Arauacutejo MB 2009 Climate change and spatial conservation planning InMoilanen A Wilson KA Possingham HP (Eds) Spatial Conservation Prior-itization Quantitative Methods and Computational Tools Oxford UniversityPress pp 172e184 ISBN 9780199547760

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221220

Arauacutejo MB Rahbek C 2006 How does climate change affect biodiversity Sci-ence 313 1396e1397 httpdxdoiorg101126science1131758

Arauacutejo MB Thuiller W Pearson RG 2006 Climate warming and the decline ofamphibians and reptiles in Europe J Biogeogr 33 1712e1728 httpdxdoiorg101111j1365-2699200601482x

Baha El Din SM 2001 The Herpetofauna of Egypt Species Communities andAssemblages University of Nottingham p 435 Unpublished PhD thesis

Baha El Din SM 2006 A Guide to the Reptiles and Amphibians of Egypt TheAmerican University in Cairo Press ISBN 9789774249792

Baha El Din SM 2007 A new lizard of the Acanthodactylus scutellatus group(Squamata Lacertidae) from Egypt Zool Middle East 40 21e32 httpdxdoiorg10108009397140200710638200

Baldwin RA 2009 Use of maximum entropy modeling in wildlife research En-tropy 11 854e866 httpdxdoiorg103390e11040854

Ball IR Possingham HP Watts M 2009 Marxan and relatives software forspatial conservation prioritisation In Moilanen A Wilson KAPossingham HP (Eds) Spatial Conservation Prioritization QuantitativeMethods and Computational Tools Oxford University Press pp 185e195 ISBN9780199547760

Basuony MI Gilbert F Zalat S 2010 The Mammals of Egypt Atlas Red DataListing amp Conservation BioMAP amp CultNat EEAA amp Bibliotheca AlexandrinaCairo available at httpecologynottinghamacuk~plzfg

Bombi P Salvi D Bologna MA Pettorelli N 2012 Cross-scale predictions allowthe identification of local conservation priorities from atlas data Anim Conserv15 378e387 httpdxdoiorg101111j1469-1795201200526x

Brotons L Thuiller W Arauacutejo MB Hirzel AH 2004 Presence-absence versuspresence-only modelling methods for predicting bird habitat suitability Ecog-raphy 27 437e448 httpdxdoiorg101111j0906-7590200403764x

Cadby CD While GM Hobday AJ Uller T Wapstra E 2010 Multi-scaleapproach to understanding climate effects on offspring size at birth and date ofbirth in a reptile Integr Zool 5 164e175 httpdxdoiorg101111j1749-4877201000201x

Carvalho SB Brito JC Crespo EJ Possingham HP 2010 From climate changepredictions to actions e conserving vulnerable animal groups in hotspots at aregional scale Glob Change Biol 16 (12) 3257e3270 httpdxdoiorg101111j1365-2486201002212x

Currie DJ 2001 Projected effects of climate change on patterns of vertebrate andtree species richness in the conterminous United States Ecosystems 4216e225 httpdxdoiorg101007s10021-001-0005-4

de Pous P 2011 Climate Change and European Reptiles from Large-scale SpatialRange Shift Projections towards a More Realistic Assessment of Species Re-sponses University of East Anglia UK p 64 MSc thesis

de Pous P Beukema W Weterings M Duumlmmer I Geniez P 2011 Area priori-tization and performance evaluation of the conservation area network for theMoroccan herpetofauna a preliminary assessment Biodivers Conserv 2089e118 httpdxdoiorg101007s10531-010-9948-0

Edgar P Foster J Baker J 2010 Reptile Habitat Management HandbookAmphibian and Reptile Conservation Bournemouth

Egbert SL Martiacutenez-Meyer E Ortega-Huerta M Peterson AT 2002 Use ofdatasets derived from time-series AVHRR imagery as surrogates for land covermaps in predicting species distributions In Proceedings IEEE 2002 Interna-tional Geoscience and Remote Sensing Symposium (IGARSS) vol 4pp 2337e2339 httpdxdoiorg101109IGARSS20021026537

El Alqamy H Ismael A Abdelhameed A Nagy A Hamada A et al 2010 Pre-dicting the status and distribution of the Nubian Ibex (Capra nubiana) in thehigh-altitude mountains of south Sinai (Egypt) In Galemys Newsletter of theSpanish Society for the Preservation and Study of Mammals vol 22pp 517e530

Elith J Graham CH Anderson RP Dudiacutek M Ferrier S et al 2006 Novelmethods improve prediction of species distributions from occurrence dataEcography 29 129e151 httpdxdoiorg101111j20060906-759004596x

Elith J Phillips SJ Hastie T Dudiacutek M Chee YE Yates CJ 2011 A statisticalexplanation of MaxEnt for ecologists Divers Distrib 17 (1) 43e57 httpdxdoiorg101111j1472-4642201000725x

FAO 2012 Food and Agriculture Organisation World Development IndicatorsAverage Precipitation in Depth (mm per year) httpdataworldbankorgindicatorAGLNDPRCPMM

Franklin J 2009 Mapping Species Distributions Spatial Inference and PredictionCambridge University Press Cambridge UK ISBN 9780521700023

Franklin J Wejnert KE Hathaway SA Rochester CJ Fisher RN 2009 Effect ofspecies rarity on the accuracy of species distribution models for reptiles andamphibians in southern California Divers Distrib 15 167e177 httpdxdoiorg101111j1472-4642200800536x

Gibbons JW Scott DE Ryan TJ Buhlmann KA Tuberville TD Metts BSGreene JL Mills T et al 2000 The global decline of reptiles Deja Vu Am-phibians BioScience 50 (8) 653e666 httpdxdoiorg1016410006-3568(2000)050[0653TGDORD]20CO2

Gilbert F Zalat S 2008 The Butterflies of Egypt Atlas Red Data Listing amp Con-servation BioMAP EEAA Cairo available at httpecologynottinghamacuk~plzfg

Golicher DJ Cayuela L Newton AC 2012 Effects of climate change on the po-tential species richness of Mesoamerican forests Biotropica 44 284e293httpdxdoiorg101111j1744-7429201100815x

Graham CH Elith J Hijmans RJ Guisan A Peterson AT et al 2008 Theinfluence of spatial errors in species occurrence data used in distribution

models J Appl Ecol 45 239e247 httpdxdoiorg101111j1365-2664200701408x

Guisan A Tingley R Baumgartner JB Naujokaitis-Lewis I Sutcliffe PRTulloch AIT Regan TJ Brotons L et al 2013 Predicting species distributionsfor conservation decisions Ecol Lett 16 (12) 1424e1435 httpdxdoiorg101111ele12189

Hannah L 2010 A global conservation system for climate-change adaptationConserv Biol 24 70e77 httpdxdoiorg101111j1523-1739200901405x

Hannah L 2011 Climate Change Biology Elsevier Ltd ISBN 9780123741820Hannah L Lovejoy TE Schneider SH 2005 Biodiversity and climate change in

context In Lovejoy TE Hannah L (Eds) Climate Change and BiodiversityYale University Press New Haven CT USA

Hannah L Midgley G Andelman S Arauacutejo M Hughes G et al 2007 Protectedarea needs in a changing climate Front Ecol Environ 5 131e138 httpdxdoiorg1018901540-9295(2007)5[131PANIAC]20CO2

Hastie T Fithian W 2013 Inference from presence-only data the ongoing con-troversy Ecography 36 864e867 httpdxdoiorg101111j1600-0587201300321x

Hernandez PA Graham CH Master LL Albert DL 2006 The effect of samplesize and species characteristics on performance of different species distributionmodeling methods Ecography 29 773e785 httpdxdoiorg101111j0906-7590200604700x

Hijmans RJ Cameron SE Parra JL Jones PG Jarvis A 2005 Very high reso-lution interpolated climate surfaces for global land areas Int J Climatol 251965e1978 httpdxdoiorg101002Joc1276

Holt AC Salkeld DJ Fritz CL Tucker JR Gong P 2009 Spatial analysis ofplague in California niche modeling predictions of the current distribution andpotential response to climate change Int J Health Geogr 8 38 httpdxdoiorg1011861476-072X-8-38

Hoyle M James M 2005 Global warming human population pressure andviability of the worlds smallest butterfly Conserv Biol 19 1113e1124 httpdxdoiorg101111j1523-1739200500166x

Ihlow F Dambach J Engler JO Flecks M Hartmann T et al 2012 On the brinkof extinction How climate change may affect global chelonian species richnessand distribution Glob Change Biol 18 1520e1530 httpdxdoiorg101111j1365-2486201102623x

IPCC 2007 Climate Change 2007 Synthesis Report Contribution of WorkingGroups I II and III to the Fourth Assessment Report of the IntergovernmentalPanel on Climate Change IPCC Geneva Switzerland

Kgosiesele E 2010 Predictive Distribution Modelling of Timon Lepida in SpainUniversity of Twente (ITC) the Netherlands p 87 MSc thesis

Klorvuttimontara S McClean CJ Hill JK 2011 Evaluating the effectiveness ofprotected areas for conserving tropical forest butterflies of Thailand BiolConserv 144 2534e2540 httpdxdoiorg101016jbiocon201107012

Leach K Zalat S Gilbert F 2013 Egypts protected area network under futureclimate change Biol Conserv 159 490e500 httpdxdoiorg101016jbiocon201211025

Lobo JM Jimenez-Valverde A Real R 2008 AUC a misleading measure of theperformance of predictive distribution models Glob Ecol Biogeogr 17145e151 httpdxdoiorg101111j1466-8238200700358x

Maiorano L Falcucci A Zimmermann NE Psomas A Pottier J et al 2011 Thefuture of terrestrial mammals in the Mediterranean basin under climatechange Philosophical Trans R Soc Lond Ser B Biol Sci 366 2681e2692httpdxdoiorg101098rstb20110121

Marini MA Barbet-Massin M Lopes LE Jiguet F 2009 Predicted climate-driven bird distribution changes and forecasted conservation conflicts in aneotropical savanna Conserv Biol 23 1558e1567 httpdxdoiorg101111j1523-1739200901258x

Merow C Smith MJ Silander JA 2013 A practical guide to MaxEnt for modelingspecies distributions what it does and why inputs and settings matterEcography 36 (10) 1058e1069 httpdxdoiorg101111j1600-0587201307872x

Moilanen A Meller L Leppeuroanen J Pouzols FM Arponen A et al 2012 Zona-tion Spatial Conservation Planning Framework and Software v31 User Manualwwwhelsinkifibioscienceconsplan

Newbold T 2009 The Value of Species Distribution Models as a Tool for Conser-vation and Ecology in Egypt and Britain University of Nottingham UK p 312PhD thesis

Newbold T Gilbert F Zalat S El-Gabbas A Reader T 2009a Climate-basedmodels of spatial patterns of species richness in Egypts butterfly and mammalfauna J Biogeogr 36 2085e2095 httpdxdoiorg101111j1365-2699200902140x

Newbold T Reader T Zalat S El-Gabbas A Gilbert F 2009b Effect of charac-teristics of butterfly species on the accuracy of distribution models in an aridenvironment Biodivers Conserv 18 3629e3641 httpdxdoiorg101007s10531-009-9668-5

Pandit SN Kolasa J Cottenie K 2009 Contrasts between habitat generalists andspecialists an empirical extension to the basic metacommunity frameworkEcology 90 2253e2262 httpdxdoiorg10189008-08511

Pearson RG Dawson TP 2003 Predicting the impacts of climate change on thedistribution of species are bioclimate envelope models useful Glob EcolBiogeogr 12 361e371 httpdxdoiorg101046j1466-822X200300042x

Pearson RG Raxworthy CJ Nakamura M Peterson AT 2007 Predicting speciesdistributions from small numbers of occurrence records a test case using

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 221

cryptic geckos in Madagascar J Biogeogr 34 102e117 httpdxdoiorg101111j1365-2699200601594x

Phillips SJ ATampT Research 2011 A Brief Tutorial on Maxent Available at httpwwwcsprincetonedu~schapiremaxenttutorialtutorialdoc (Last accessed290512)

Phillips SJ Anderson RP Schapire RE 2006 Maximum entropy modeling ofspecies geographic distributions Ecol Model 190 231e259 httpdxdoiorg101016jecolmodel200503026

Phillips SJ Dudiacutek M Elith J Graham CH Lehmann A Leathwick J Ferrier S2009 Sample selection bias and presence-only distribution models implica-tions for background and pseudo-absence data Ecol Appl 19 181e197 httpdxdoiorg10189007-21531

R Development Core Team 2012 R a Language and Environment for StatisticalComputing R Foundation for Statistical Computing Vienna Austria ISBN 3-900051-07-0 httpwwwR-projectorg

Radosavljevic A Anderson RP 2014 Making better Maxent models of speciesdistributions complexity overfitting and evaluation J Biogeogr 41 (4)629e643 httpdxdoiorg101111jbi12227

Royle JA Chandler RB Yackulic C Nichols JD 2012 Likelihood analysis ofspecies occurrence probability from presence-only data for modelling speciesdistributions Methods Ecol Evol 3 545e554 httpdxdoiorg101111j2041-210X201100182x

Sauer J Domisch S Nowak C Haase P 2011 Lowmountain ranges summit trapsfor montane freshwater species under climate change Biodivers Conserv 203133e3146 httpdxdoiorg101007s10531-011-0140-y

Songer M Delion M Biggs A Huang Q 2012 Modeling impacts of climatechange on Giant Panda Habitat Int J Ecol 2012 1e12 httpdxdoiorg1011552012108752

Soultan AE 2011 Antelope distribution in Egypt Modeling the Impact of ClimateChange on Antelopes Distribution in Egypt Lambert Academic Publishing

Standards IUCN Subcommittee Petitions 2010 Guidelines for Using the IUCNRed List Categories and Criteria Version 81

Stockwell DRB Peterson AT 2002 Effects of sample size on accuracy of speciesdistribution models Ecol Model 148 1e13 httpdxdoiorg101016S0304-3800(01)00388-X

Suarez AV Tsutsui ND 2004 The value of museum collections for research andsociety Bioscience 54 66 httpdxdoiorg1016410006-3568(2004)054[0066tvomcf]20co2

Taheri S 2010 Hyper Temporal NDVI Images for Modelling and Prediction theHabitat Distribution of Balkan Green Lizard (Lacerta Trilineata) e Case Study

Crete (Greece) University of Twente (ITC) the Netherlands p 54 MScthesis

Taubmann J Theissinger K Feldheim KA Laube I Graf W et al 2011Modelling range shifts and assessing genetic diversity distribution of themontane aquatic mayfly Ameletus inopinatus in Europe under climate changescenarios Conserv Genet 12 503e515 httpdxdoiorg101007s10592-010-0157-x

Teodoro AC Sillero N Alves S Duarte L 2013 Correlation between the habitatsproductivity and species richness (amphibians and reptiles) in Portugal throughremote sensed data In Proc SPIE 8887 Remote Sensing for Agriculture Eco-systems and Hydrology httpdxdoiorg101117122028502

Thomas CD Cameron A Green RE Bakkenes M Beaumont LJ et al 2004Extinction risk from climate change Nature 427 145e148 httpdxdoiorg101038nature02121

Thuiller W 2004 Patterns and uncertainties of species range shifts under climatechange Glob Change Biol 10 2020e2027 httpdxdoiorg101111j1365-2486200400859x

Thuiller W 2007 Biodiversity climate change and the ecologist Nature 448550e552 httpdxdoiorg101038448550a

Thuiller W Lavorel S Arauacutejo MB Sykes MT Prentice IC 2005 Climate changethreats to plant diversity in Europe Proc Natl Acad Sci U S A 1028245e8250 httpdxdoiorg101073pnas0409902102

Thuiller W Lafourcade B Engler R Arauacutejo MB 2009 BIOMOD e a platform forensemble forecasting of species distributions Ecography 32 (3) 369e373httpdxdoiorg101111j1600-0587200805742x

Tolba MK Saab NW 2009 Arab environment Climate Change - Impact ofClimate Change on Arab Countries Arab Forum for Environment andDvelopment

Trotta-Moreu N Lobo JM 2010 Deriving the species richness distribution ofGeotrupinae (Coleoptera Scarabaeoidea) in Mexico from the overlap of indi-vidual model predictions Environ Entomol 39 42e49 httpdxdoiorg101603EN08179

Warren DL Seifert SN 2011 Ecological niche modeling in Maxent the impor-tance of model complexity and the performance of model selection criteriaEcol Appl 21 (2) 335e342 httpdxdoiorg10189010-11711

Wilson KA Cabeza M Klein CJ 2009 Fundamental concepts of spatial con-servation prioritisation In Moilanen A Wilson KA Possingham HP (Eds)Spatial Conservation Prioritization Quantitative Methods and ComputationalTools Oxford University Press pp 16e27 ISBN 9780199547760

Page 10: Journal of Arid Environments - elgabbas.github.io · e Faculty of Sciences & Arts, Taibah University, Al-Ula Campus, Saudi Arabia f Department of Zoology, Suez Canal University, Ismailia,

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221220

Arauacutejo MB Rahbek C 2006 How does climate change affect biodiversity Sci-ence 313 1396e1397 httpdxdoiorg101126science1131758

Arauacutejo MB Thuiller W Pearson RG 2006 Climate warming and the decline ofamphibians and reptiles in Europe J Biogeogr 33 1712e1728 httpdxdoiorg101111j1365-2699200601482x

Baha El Din SM 2001 The Herpetofauna of Egypt Species Communities andAssemblages University of Nottingham p 435 Unpublished PhD thesis

Baha El Din SM 2006 A Guide to the Reptiles and Amphibians of Egypt TheAmerican University in Cairo Press ISBN 9789774249792

Baha El Din SM 2007 A new lizard of the Acanthodactylus scutellatus group(Squamata Lacertidae) from Egypt Zool Middle East 40 21e32 httpdxdoiorg10108009397140200710638200

Baldwin RA 2009 Use of maximum entropy modeling in wildlife research En-tropy 11 854e866 httpdxdoiorg103390e11040854

Ball IR Possingham HP Watts M 2009 Marxan and relatives software forspatial conservation prioritisation In Moilanen A Wilson KAPossingham HP (Eds) Spatial Conservation Prioritization QuantitativeMethods and Computational Tools Oxford University Press pp 185e195 ISBN9780199547760

Basuony MI Gilbert F Zalat S 2010 The Mammals of Egypt Atlas Red DataListing amp Conservation BioMAP amp CultNat EEAA amp Bibliotheca AlexandrinaCairo available at httpecologynottinghamacuk~plzfg

Bombi P Salvi D Bologna MA Pettorelli N 2012 Cross-scale predictions allowthe identification of local conservation priorities from atlas data Anim Conserv15 378e387 httpdxdoiorg101111j1469-1795201200526x

Brotons L Thuiller W Arauacutejo MB Hirzel AH 2004 Presence-absence versuspresence-only modelling methods for predicting bird habitat suitability Ecog-raphy 27 437e448 httpdxdoiorg101111j0906-7590200403764x

Cadby CD While GM Hobday AJ Uller T Wapstra E 2010 Multi-scaleapproach to understanding climate effects on offspring size at birth and date ofbirth in a reptile Integr Zool 5 164e175 httpdxdoiorg101111j1749-4877201000201x

Carvalho SB Brito JC Crespo EJ Possingham HP 2010 From climate changepredictions to actions e conserving vulnerable animal groups in hotspots at aregional scale Glob Change Biol 16 (12) 3257e3270 httpdxdoiorg101111j1365-2486201002212x

Currie DJ 2001 Projected effects of climate change on patterns of vertebrate andtree species richness in the conterminous United States Ecosystems 4216e225 httpdxdoiorg101007s10021-001-0005-4

de Pous P 2011 Climate Change and European Reptiles from Large-scale SpatialRange Shift Projections towards a More Realistic Assessment of Species Re-sponses University of East Anglia UK p 64 MSc thesis

de Pous P Beukema W Weterings M Duumlmmer I Geniez P 2011 Area priori-tization and performance evaluation of the conservation area network for theMoroccan herpetofauna a preliminary assessment Biodivers Conserv 2089e118 httpdxdoiorg101007s10531-010-9948-0

Edgar P Foster J Baker J 2010 Reptile Habitat Management HandbookAmphibian and Reptile Conservation Bournemouth

Egbert SL Martiacutenez-Meyer E Ortega-Huerta M Peterson AT 2002 Use ofdatasets derived from time-series AVHRR imagery as surrogates for land covermaps in predicting species distributions In Proceedings IEEE 2002 Interna-tional Geoscience and Remote Sensing Symposium (IGARSS) vol 4pp 2337e2339 httpdxdoiorg101109IGARSS20021026537

El Alqamy H Ismael A Abdelhameed A Nagy A Hamada A et al 2010 Pre-dicting the status and distribution of the Nubian Ibex (Capra nubiana) in thehigh-altitude mountains of south Sinai (Egypt) In Galemys Newsletter of theSpanish Society for the Preservation and Study of Mammals vol 22pp 517e530

Elith J Graham CH Anderson RP Dudiacutek M Ferrier S et al 2006 Novelmethods improve prediction of species distributions from occurrence dataEcography 29 129e151 httpdxdoiorg101111j20060906-759004596x

Elith J Phillips SJ Hastie T Dudiacutek M Chee YE Yates CJ 2011 A statisticalexplanation of MaxEnt for ecologists Divers Distrib 17 (1) 43e57 httpdxdoiorg101111j1472-4642201000725x

FAO 2012 Food and Agriculture Organisation World Development IndicatorsAverage Precipitation in Depth (mm per year) httpdataworldbankorgindicatorAGLNDPRCPMM

Franklin J 2009 Mapping Species Distributions Spatial Inference and PredictionCambridge University Press Cambridge UK ISBN 9780521700023

Franklin J Wejnert KE Hathaway SA Rochester CJ Fisher RN 2009 Effect ofspecies rarity on the accuracy of species distribution models for reptiles andamphibians in southern California Divers Distrib 15 167e177 httpdxdoiorg101111j1472-4642200800536x

Gibbons JW Scott DE Ryan TJ Buhlmann KA Tuberville TD Metts BSGreene JL Mills T et al 2000 The global decline of reptiles Deja Vu Am-phibians BioScience 50 (8) 653e666 httpdxdoiorg1016410006-3568(2000)050[0653TGDORD]20CO2

Gilbert F Zalat S 2008 The Butterflies of Egypt Atlas Red Data Listing amp Con-servation BioMAP EEAA Cairo available at httpecologynottinghamacuk~plzfg

Golicher DJ Cayuela L Newton AC 2012 Effects of climate change on the po-tential species richness of Mesoamerican forests Biotropica 44 284e293httpdxdoiorg101111j1744-7429201100815x

Graham CH Elith J Hijmans RJ Guisan A Peterson AT et al 2008 Theinfluence of spatial errors in species occurrence data used in distribution

models J Appl Ecol 45 239e247 httpdxdoiorg101111j1365-2664200701408x

Guisan A Tingley R Baumgartner JB Naujokaitis-Lewis I Sutcliffe PRTulloch AIT Regan TJ Brotons L et al 2013 Predicting species distributionsfor conservation decisions Ecol Lett 16 (12) 1424e1435 httpdxdoiorg101111ele12189

Hannah L 2010 A global conservation system for climate-change adaptationConserv Biol 24 70e77 httpdxdoiorg101111j1523-1739200901405x

Hannah L 2011 Climate Change Biology Elsevier Ltd ISBN 9780123741820Hannah L Lovejoy TE Schneider SH 2005 Biodiversity and climate change in

context In Lovejoy TE Hannah L (Eds) Climate Change and BiodiversityYale University Press New Haven CT USA

Hannah L Midgley G Andelman S Arauacutejo M Hughes G et al 2007 Protectedarea needs in a changing climate Front Ecol Environ 5 131e138 httpdxdoiorg1018901540-9295(2007)5[131PANIAC]20CO2

Hastie T Fithian W 2013 Inference from presence-only data the ongoing con-troversy Ecography 36 864e867 httpdxdoiorg101111j1600-0587201300321x

Hernandez PA Graham CH Master LL Albert DL 2006 The effect of samplesize and species characteristics on performance of different species distributionmodeling methods Ecography 29 773e785 httpdxdoiorg101111j0906-7590200604700x

Hijmans RJ Cameron SE Parra JL Jones PG Jarvis A 2005 Very high reso-lution interpolated climate surfaces for global land areas Int J Climatol 251965e1978 httpdxdoiorg101002Joc1276

Holt AC Salkeld DJ Fritz CL Tucker JR Gong P 2009 Spatial analysis ofplague in California niche modeling predictions of the current distribution andpotential response to climate change Int J Health Geogr 8 38 httpdxdoiorg1011861476-072X-8-38

Hoyle M James M 2005 Global warming human population pressure andviability of the worlds smallest butterfly Conserv Biol 19 1113e1124 httpdxdoiorg101111j1523-1739200500166x

Ihlow F Dambach J Engler JO Flecks M Hartmann T et al 2012 On the brinkof extinction How climate change may affect global chelonian species richnessand distribution Glob Change Biol 18 1520e1530 httpdxdoiorg101111j1365-2486201102623x

IPCC 2007 Climate Change 2007 Synthesis Report Contribution of WorkingGroups I II and III to the Fourth Assessment Report of the IntergovernmentalPanel on Climate Change IPCC Geneva Switzerland

Kgosiesele E 2010 Predictive Distribution Modelling of Timon Lepida in SpainUniversity of Twente (ITC) the Netherlands p 87 MSc thesis

Klorvuttimontara S McClean CJ Hill JK 2011 Evaluating the effectiveness ofprotected areas for conserving tropical forest butterflies of Thailand BiolConserv 144 2534e2540 httpdxdoiorg101016jbiocon201107012

Leach K Zalat S Gilbert F 2013 Egypts protected area network under futureclimate change Biol Conserv 159 490e500 httpdxdoiorg101016jbiocon201211025

Lobo JM Jimenez-Valverde A Real R 2008 AUC a misleading measure of theperformance of predictive distribution models Glob Ecol Biogeogr 17145e151 httpdxdoiorg101111j1466-8238200700358x

Maiorano L Falcucci A Zimmermann NE Psomas A Pottier J et al 2011 Thefuture of terrestrial mammals in the Mediterranean basin under climatechange Philosophical Trans R Soc Lond Ser B Biol Sci 366 2681e2692httpdxdoiorg101098rstb20110121

Marini MA Barbet-Massin M Lopes LE Jiguet F 2009 Predicted climate-driven bird distribution changes and forecasted conservation conflicts in aneotropical savanna Conserv Biol 23 1558e1567 httpdxdoiorg101111j1523-1739200901258x

Merow C Smith MJ Silander JA 2013 A practical guide to MaxEnt for modelingspecies distributions what it does and why inputs and settings matterEcography 36 (10) 1058e1069 httpdxdoiorg101111j1600-0587201307872x

Moilanen A Meller L Leppeuroanen J Pouzols FM Arponen A et al 2012 Zona-tion Spatial Conservation Planning Framework and Software v31 User Manualwwwhelsinkifibioscienceconsplan

Newbold T 2009 The Value of Species Distribution Models as a Tool for Conser-vation and Ecology in Egypt and Britain University of Nottingham UK p 312PhD thesis

Newbold T Gilbert F Zalat S El-Gabbas A Reader T 2009a Climate-basedmodels of spatial patterns of species richness in Egypts butterfly and mammalfauna J Biogeogr 36 2085e2095 httpdxdoiorg101111j1365-2699200902140x

Newbold T Reader T Zalat S El-Gabbas A Gilbert F 2009b Effect of charac-teristics of butterfly species on the accuracy of distribution models in an aridenvironment Biodivers Conserv 18 3629e3641 httpdxdoiorg101007s10531-009-9668-5

Pandit SN Kolasa J Cottenie K 2009 Contrasts between habitat generalists andspecialists an empirical extension to the basic metacommunity frameworkEcology 90 2253e2262 httpdxdoiorg10189008-08511

Pearson RG Dawson TP 2003 Predicting the impacts of climate change on thedistribution of species are bioclimate envelope models useful Glob EcolBiogeogr 12 361e371 httpdxdoiorg101046j1466-822X200300042x

Pearson RG Raxworthy CJ Nakamura M Peterson AT 2007 Predicting speciesdistributions from small numbers of occurrence records a test case using

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 221

cryptic geckos in Madagascar J Biogeogr 34 102e117 httpdxdoiorg101111j1365-2699200601594x

Phillips SJ ATampT Research 2011 A Brief Tutorial on Maxent Available at httpwwwcsprincetonedu~schapiremaxenttutorialtutorialdoc (Last accessed290512)

Phillips SJ Anderson RP Schapire RE 2006 Maximum entropy modeling ofspecies geographic distributions Ecol Model 190 231e259 httpdxdoiorg101016jecolmodel200503026

Phillips SJ Dudiacutek M Elith J Graham CH Lehmann A Leathwick J Ferrier S2009 Sample selection bias and presence-only distribution models implica-tions for background and pseudo-absence data Ecol Appl 19 181e197 httpdxdoiorg10189007-21531

R Development Core Team 2012 R a Language and Environment for StatisticalComputing R Foundation for Statistical Computing Vienna Austria ISBN 3-900051-07-0 httpwwwR-projectorg

Radosavljevic A Anderson RP 2014 Making better Maxent models of speciesdistributions complexity overfitting and evaluation J Biogeogr 41 (4)629e643 httpdxdoiorg101111jbi12227

Royle JA Chandler RB Yackulic C Nichols JD 2012 Likelihood analysis ofspecies occurrence probability from presence-only data for modelling speciesdistributions Methods Ecol Evol 3 545e554 httpdxdoiorg101111j2041-210X201100182x

Sauer J Domisch S Nowak C Haase P 2011 Lowmountain ranges summit trapsfor montane freshwater species under climate change Biodivers Conserv 203133e3146 httpdxdoiorg101007s10531-011-0140-y

Songer M Delion M Biggs A Huang Q 2012 Modeling impacts of climatechange on Giant Panda Habitat Int J Ecol 2012 1e12 httpdxdoiorg1011552012108752

Soultan AE 2011 Antelope distribution in Egypt Modeling the Impact of ClimateChange on Antelopes Distribution in Egypt Lambert Academic Publishing

Standards IUCN Subcommittee Petitions 2010 Guidelines for Using the IUCNRed List Categories and Criteria Version 81

Stockwell DRB Peterson AT 2002 Effects of sample size on accuracy of speciesdistribution models Ecol Model 148 1e13 httpdxdoiorg101016S0304-3800(01)00388-X

Suarez AV Tsutsui ND 2004 The value of museum collections for research andsociety Bioscience 54 66 httpdxdoiorg1016410006-3568(2004)054[0066tvomcf]20co2

Taheri S 2010 Hyper Temporal NDVI Images for Modelling and Prediction theHabitat Distribution of Balkan Green Lizard (Lacerta Trilineata) e Case Study

Crete (Greece) University of Twente (ITC) the Netherlands p 54 MScthesis

Taubmann J Theissinger K Feldheim KA Laube I Graf W et al 2011Modelling range shifts and assessing genetic diversity distribution of themontane aquatic mayfly Ameletus inopinatus in Europe under climate changescenarios Conserv Genet 12 503e515 httpdxdoiorg101007s10592-010-0157-x

Teodoro AC Sillero N Alves S Duarte L 2013 Correlation between the habitatsproductivity and species richness (amphibians and reptiles) in Portugal throughremote sensed data In Proc SPIE 8887 Remote Sensing for Agriculture Eco-systems and Hydrology httpdxdoiorg101117122028502

Thomas CD Cameron A Green RE Bakkenes M Beaumont LJ et al 2004Extinction risk from climate change Nature 427 145e148 httpdxdoiorg101038nature02121

Thuiller W 2004 Patterns and uncertainties of species range shifts under climatechange Glob Change Biol 10 2020e2027 httpdxdoiorg101111j1365-2486200400859x

Thuiller W 2007 Biodiversity climate change and the ecologist Nature 448550e552 httpdxdoiorg101038448550a

Thuiller W Lavorel S Arauacutejo MB Sykes MT Prentice IC 2005 Climate changethreats to plant diversity in Europe Proc Natl Acad Sci U S A 1028245e8250 httpdxdoiorg101073pnas0409902102

Thuiller W Lafourcade B Engler R Arauacutejo MB 2009 BIOMOD e a platform forensemble forecasting of species distributions Ecography 32 (3) 369e373httpdxdoiorg101111j1600-0587200805742x

Tolba MK Saab NW 2009 Arab environment Climate Change - Impact ofClimate Change on Arab Countries Arab Forum for Environment andDvelopment

Trotta-Moreu N Lobo JM 2010 Deriving the species richness distribution ofGeotrupinae (Coleoptera Scarabaeoidea) in Mexico from the overlap of indi-vidual model predictions Environ Entomol 39 42e49 httpdxdoiorg101603EN08179

Warren DL Seifert SN 2011 Ecological niche modeling in Maxent the impor-tance of model complexity and the performance of model selection criteriaEcol Appl 21 (2) 335e342 httpdxdoiorg10189010-11711

Wilson KA Cabeza M Klein CJ 2009 Fundamental concepts of spatial con-servation prioritisation In Moilanen A Wilson KA Possingham HP (Eds)Spatial Conservation Prioritization Quantitative Methods and ComputationalTools Oxford University Press pp 16e27 ISBN 9780199547760

Page 11: Journal of Arid Environments - elgabbas.github.io · e Faculty of Sciences & Arts, Taibah University, Al-Ula Campus, Saudi Arabia f Department of Zoology, Suez Canal University, Ismailia,

A El-Gabbas et al Journal of Arid Environments 127 (2016) 211e221 221

cryptic geckos in Madagascar J Biogeogr 34 102e117 httpdxdoiorg101111j1365-2699200601594x

Phillips SJ ATampT Research 2011 A Brief Tutorial on Maxent Available at httpwwwcsprincetonedu~schapiremaxenttutorialtutorialdoc (Last accessed290512)

Phillips SJ Anderson RP Schapire RE 2006 Maximum entropy modeling ofspecies geographic distributions Ecol Model 190 231e259 httpdxdoiorg101016jecolmodel200503026

Phillips SJ Dudiacutek M Elith J Graham CH Lehmann A Leathwick J Ferrier S2009 Sample selection bias and presence-only distribution models implica-tions for background and pseudo-absence data Ecol Appl 19 181e197 httpdxdoiorg10189007-21531

R Development Core Team 2012 R a Language and Environment for StatisticalComputing R Foundation for Statistical Computing Vienna Austria ISBN 3-900051-07-0 httpwwwR-projectorg

Radosavljevic A Anderson RP 2014 Making better Maxent models of speciesdistributions complexity overfitting and evaluation J Biogeogr 41 (4)629e643 httpdxdoiorg101111jbi12227

Royle JA Chandler RB Yackulic C Nichols JD 2012 Likelihood analysis ofspecies occurrence probability from presence-only data for modelling speciesdistributions Methods Ecol Evol 3 545e554 httpdxdoiorg101111j2041-210X201100182x

Sauer J Domisch S Nowak C Haase P 2011 Lowmountain ranges summit trapsfor montane freshwater species under climate change Biodivers Conserv 203133e3146 httpdxdoiorg101007s10531-011-0140-y

Songer M Delion M Biggs A Huang Q 2012 Modeling impacts of climatechange on Giant Panda Habitat Int J Ecol 2012 1e12 httpdxdoiorg1011552012108752

Soultan AE 2011 Antelope distribution in Egypt Modeling the Impact of ClimateChange on Antelopes Distribution in Egypt Lambert Academic Publishing

Standards IUCN Subcommittee Petitions 2010 Guidelines for Using the IUCNRed List Categories and Criteria Version 81

Stockwell DRB Peterson AT 2002 Effects of sample size on accuracy of speciesdistribution models Ecol Model 148 1e13 httpdxdoiorg101016S0304-3800(01)00388-X

Suarez AV Tsutsui ND 2004 The value of museum collections for research andsociety Bioscience 54 66 httpdxdoiorg1016410006-3568(2004)054[0066tvomcf]20co2

Taheri S 2010 Hyper Temporal NDVI Images for Modelling and Prediction theHabitat Distribution of Balkan Green Lizard (Lacerta Trilineata) e Case Study

Crete (Greece) University of Twente (ITC) the Netherlands p 54 MScthesis

Taubmann J Theissinger K Feldheim KA Laube I Graf W et al 2011Modelling range shifts and assessing genetic diversity distribution of themontane aquatic mayfly Ameletus inopinatus in Europe under climate changescenarios Conserv Genet 12 503e515 httpdxdoiorg101007s10592-010-0157-x

Teodoro AC Sillero N Alves S Duarte L 2013 Correlation between the habitatsproductivity and species richness (amphibians and reptiles) in Portugal throughremote sensed data In Proc SPIE 8887 Remote Sensing for Agriculture Eco-systems and Hydrology httpdxdoiorg101117122028502

Thomas CD Cameron A Green RE Bakkenes M Beaumont LJ et al 2004Extinction risk from climate change Nature 427 145e148 httpdxdoiorg101038nature02121

Thuiller W 2004 Patterns and uncertainties of species range shifts under climatechange Glob Change Biol 10 2020e2027 httpdxdoiorg101111j1365-2486200400859x

Thuiller W 2007 Biodiversity climate change and the ecologist Nature 448550e552 httpdxdoiorg101038448550a

Thuiller W Lavorel S Arauacutejo MB Sykes MT Prentice IC 2005 Climate changethreats to plant diversity in Europe Proc Natl Acad Sci U S A 1028245e8250 httpdxdoiorg101073pnas0409902102

Thuiller W Lafourcade B Engler R Arauacutejo MB 2009 BIOMOD e a platform forensemble forecasting of species distributions Ecography 32 (3) 369e373httpdxdoiorg101111j1600-0587200805742x

Tolba MK Saab NW 2009 Arab environment Climate Change - Impact ofClimate Change on Arab Countries Arab Forum for Environment andDvelopment

Trotta-Moreu N Lobo JM 2010 Deriving the species richness distribution ofGeotrupinae (Coleoptera Scarabaeoidea) in Mexico from the overlap of indi-vidual model predictions Environ Entomol 39 42e49 httpdxdoiorg101603EN08179

Warren DL Seifert SN 2011 Ecological niche modeling in Maxent the impor-tance of model complexity and the performance of model selection criteriaEcol Appl 21 (2) 335e342 httpdxdoiorg10189010-11711

Wilson KA Cabeza M Klein CJ 2009 Fundamental concepts of spatial con-servation prioritisation In Moilanen A Wilson KA Possingham HP (Eds)Spatial Conservation Prioritization Quantitative Methods and ComputationalTools Oxford University Press pp 16e27 ISBN 9780199547760