Introduction

Over the last few years, the studies of species niche modeling have substantially increased (Boucher et al. 2012; Schnitzler et al. 2012; Nyári and Reddy 2013; Sallam et al. 2013; Rato et al. 2015; Alahmed et al. 2015; Algar and Mahler 2015; Naeem et al. 2016; Mulieri and Patitucci 2019). The huge accumulative data of species records, availability of geographical, ecological and climatological information, in addition to the development of several statistical modeling tools helped in such revolution for studying biological units (Elith et al. 2011; Salariato and Zuloaga 2016). This is synchronizing with the raise of the new term ‘biodiversity informatics’. The term which apply to the science that benefit from our biological heritage records by manipulating these data with information techniques and geographical information systems (GIS) for improving our understanding of species spatial distributions (Maldonado et al. 2015; Asase and Peterson 2016).

The study of insects’ niche modeling received little attention than that of other creatures such as vertebrates. Only the highly medical important insect groups were studied using these techniques (Sallam et al. 2013; Alahmed et al. 2015; Naeem et al. 2016), neglecting the majority of insects’ fauna which represents 80% of earth’s biodiversity (Hochkirch 2016). One of these neglected groups is the family Bombyliidae, or bee flies as they are commonly called. Bee flies are one of the largest families in the order Diptera, with about 4800 species known worldwide and virtually cosmopolitan in distribution (Evenhuis and Greathead 2015). They are found primarily in warm arid and semi-arid habitats (Greathead and Evenhuis 1997). In Egypt, as predominantly having arid and semi-arid habitats, bee flies are very well represented, and the diversity of their appearance can scarcely fail to excite interest (El-Moursy et al. 1999).

El-Hawagry and Gilbert (2014) analyzed the distribution of 229 bee fly species based on their occurrence in the different Egyptian ecological zones showing a high species richness in Eastern Desert, Sinai and Coastal Strip. Species richness in Lower Nile Valley & Delta and Gebel Elba was moderate, while species richness in the Upper Nile Valley, Fayoum Basin and Western Desert were the lowest. The bee fly species richness, according to the previous study and El-Hawagry’s (2015) study, was positively correlated with plant diversity, and probably reflecting varying climatic factors and vegetation patterns in different Egyptian ecological zones.

El-Hawagry (2015) merely listed the localities from which Egyptian bee fly species have been recorded based on all previous studies and on the preserved material in the main Egyptian insect collections. These records provide information on only a subset of localities occupied by a species and do not provide information on localities that have not been surveyed, or that may be colonized in the future following climate changes (Rondinini et al. 2006; Hoegh-Guldberg et al. 2008). However, this information is very important as it can be employed for making powerful conservation management decisions and can be exploited for the predictions of species occurrences derived from environmental suitability models that combine biological records with spatial environmental data (Guisan et al. 2013).

The genus Anthrax Scopoli (family: Bombyliidae, subfamily: Anthracinae, tribe: Anthracini) is a well-known genus of bee flies, with 16 species recorded from Egypt, out of about 250 species known worldwide. It is characterized by the following combination of characters: small to large species, up to 15 mm; very short proboscis; pedicel globular or disc-shaped, not moulded to flagellum; flagellum onion-shaped; wings usually with a more or less extensive brown or black infuscation, seldom almost completely hyaline, never completely infuscated, second longitudinal vein in the wing is rather long; body with dense scales, lying flat, predominantly black with patches of white or silver scales on abdomen. Larvae are usually parasites of wild bees (Coquillett 1887; Steyskal and El-Bialy 1967; El-Hawagry and Evenhuis 2008).

Six of these species seem to be endangered or threatened (at least in parts of their distribution range) as they disappeared from most of their Egyptian localities which faced severe changes in the floral and faunal composition during the last 50 years. These changes may be attributed to urbanization or modern agricultural activities that caused severe degradation of natural habitats of these species. These endangered or threatened species are: Anthrax aethiopsbezzii (Paramonov 1957), Anthrax chionanthrax (Bezzi 1926), Anthrax dentata (Becker 1906), Anthrax lucidus (Becker 1902), Anthrax melanista (Bezzi 1925) and Anthrax trifasciatus Meigen 1804. The present study aims to propose environmental niche models, and to locate the sites of the highest habitat suitability for the studied species, based on climatological and topographical data. Also, to evaluate their present status in St. Catherine protected area and predict the climate change effects on the habitat suitability of Anthrax trifasciatus under four different climate change scenarios on 2070.

Materials and methods

This study was conducted in Egypt, the country that covers approximately 1001,449 square kilometers of land area. Egypt is divided into eight ecological zones: Coastal Strip, Lower Nile Valley & Delta, Upper Nile Valley, Fayoum Basin, Eastern Desert, Western Desert, Sinai and Gebel Elba (Larsen 1971, 1990; Boulos 1999–2005; El-Hawagry 2002, 2011, 2017; El-Hawagry and Gilbert 2014). It is characterized by a warm and almost rainless climate, and it is one of the driest countries in the world with a long-term average precipitation of 51 mm per year (Hoath 1992; FAO 2012).

A total of 80 records for the six Anthrax spp. (Fig. 1) were gathered from literature (Efflatoun 1945, 2002, 2015), museum collections, in addition to material collected by the second author (M. El-Hawagry) during a large series of collecting trips to various localities in the eight Egyptian ecological zones from 1991 to 2015.

Fig. 1
figure 1

The map shows the distribution of the occurrence records of the six species of genus Anthrax spp. Through the Egyptian territories

Full size image

Climatic data were downloaded from the WorldClim database with spatial resolution 30 arc-s (~ 1 km) (http://www.worldclim.org/). In this study, a combination of 19 climatic and three topographical variables was initially used to assess current climatic conditions (Table 1). We remove bioclimatic variables 8–9 and 18–19 from the analysis, due to spatial artifacts in those four variables (Samy et al. 2016). We first used the jackknifing function in Maxent to estimate the most important set of variables, also this analysis produce jackknife that obtain the variables that had no effect (Zero effect) for species distribution. The variables that showed any contribution for each species were undergoing the test in SDMTools in ArcGIS 10.3 (Universal tool; Remove highly correlated variables) to remove variables with high correlations and which consequentially have no biological effect on the species (Alkishe et al. 2017). Accordingly, we used a set of four most important variables that contribute to habitat suitability for each species (Table 2). All these bioclimatic layers were clipped to match the dimensions of Egypt and saved in ASCII grid format for using in Maxent, for clipping these layers ArcGIS software V. 10.3 was used (Alahmed et al. 2015). These variables are used in species distribution modeling (Kumar et al. 2009; Sanchez et al. 2011). The data were extracted for each species by using the “extract by mask” function in Arc Map to focus the results on the St. Catherine protected area. Climatic data projected to the year 2070 from the global climate models of the Meteorological Research Institute (MRI-CGCM3), Beijing Climate Center (BCC-CSM 1_1), National Center for Atmospheric Research (CCSM4) and Norwegian Climate Centre (NorESM 1-M). We used these four climate models to assess the effects of climate change on the species A. trifasciatus for RCP2.6 and RCP8.5. These data are among the recent GCM climate projections that are used in the fifth Assessment IPCC Report.

Table 1 Environmental and topographical variables used in Maxent to predict the current habitat suitability distribution of six species of the genus Anthrax in Egypt
Full size table
Table 2 Percent contributions of the bioclimatic and topographic variables in the Maxent Models for six target species; values shown are averages over ten replicate runs
Full size table

Maximum entropy model (MaxEnt version 3.3.3k http://www.cs.princeton.edu/wschapire/maxent/) were used in this study because Maxent perform better with small sample sizes compare to other modeling methods (Pearson et al. 2007). The program estimate the suitable habitat for species that varies from 0 (lowest suitability) to 1 (highest suitability) and generate response curve for each bioclimatic variable to estimate the relationships between the habitat suitability for a species and bioclimatic variables (Khanum et al. 2013). Maxent automatically removes the duplicate records in the same cell. In our models, we selected 75% of the occurrence records for training and 25% of the records for testing the model (Phillips 2008). To improve the model performance, tenfold cross-validation was used in which the model was run ten replicates for each species and the results were averaged. The area under curve (AUC) of the receiver operating characteristics (ROC) was used to estimate the model performance and its value varies from 0 (random discrimination) to 1 (perfect discrimination) (Peterson et al. 2008; Swets 1988).

Results

Models for six Anthrax spp. performed better than random, with averages test AUC values ranging from 0.83 to 0.99. The current distributions of the species as modeled are as follows: The Maxent model for A. aethiops bezzii performed very well with an average AUC value of 0.95. Current suitable habitat predicted that zones with high habitat suitability were North Sinai, some areas in South Sinai and Coastal Strip but the habitat suitability was low in the rest of the country (Fig. 2a). St. Catherine protected area showed high habitat suitability in some areas (Fig. 3a). Precipitation of wettest quarter (Bio 16) showed higher effects on the distribution of A. aethiops bezzii relative to other bioclimatic variables (Fig. 4). This variable was the top predictor in the Maxent model with contributions of 92.1% (Table 2). The probability of presence of A. aethiops bezzii increased with an increase in precipitation of wettest quarter (Bio 16) (Fig. 5).

Fig. 2
figure 2

Predicted current habitat suitability for six Anthrax species in Egypt; aA. aethiops, bA. chionanthrax, cA. dentata, dA. lucidus, eA. melanista and fA. trifasciatus

Full size image
Fig. 3
figure 3

Predicted current habitat suitability for six Anthrax species in St. Catherine protected area. aA. aethiops,bA. chionanthrax,cA. dentata,dA. lucidus,eA. melanista and fA. trifasciatus

Full size image
Fig. 4
figure 4

Jacknife for each species show the most biological effective environmental variables and their contribution on produced models

Full size image
Fig. 5
figure 5

Response curves showing the relationships between the probability of presence of a species and one top bioclimatic predictor of: values shown are average over ten replicate runs

Full size image

The model performance for A. chionanthrax was better than random with an AUC value of 0.83. The predicted distribution in Egypt showed areas with high suitability for this species in North Sinai, Eastern Desert especially in the Red Sea coast, and Coastal Strip. The habitat suitability was medium in the rest of Sinai, some areas in Lower Nile Valley & Delta and Western Desert. However, the habitat suitability was low in the rest of the country (Fig. 2b). The habitat suitability was high in almost all areas in St. Catherine protected area, but some parts showed very low habitat suitability (Fig. 3b). Temperature seasonality (Bio 4) provided the most useful and unique information for A. chionanthrax, that was not present in other variables (Fig. 4). Temperature seasonality (Bio 4) had the top contribution of 60.1% (Table 2). The suitability of habitat for A. chionanthrax decreased sharply with the increasing of temperature seasonality (Bio 4) (Fig. 5).

The Maxent model for A. dentata gave satisfactory results with an AUC value of 0.98. Current suitable habitat predicted that zones with high habitat suitability were Lower Nile Valley & Delta, some areas in Eastern Desert, Western Dessert, and South Sinai. The rest of the country showed low suitable habitat for this species (Fig. 2c). Some areas in St. Catherine protected area showed high habitat suitability to the species (Fig. 3c). Altitude was the most important predictor of A. dentata’s habitat distribution (Fig. 4). This variable presented the higher contribution of 62.8% compared to other predictors (Table 2). The suitability of habitat for A. dentata increased with increasing altitude also the species preferred the low altitude (Fig. 5).

The Maxent model for A. lucidus had an AUC value of 0.96. The predicted distribution in Egypt showed areas with high suitability for this species in Lower Nile Valley & Delta, Fayoum Basin, some areas in Coastal Strip and Western Desert. The habitat suitability was low in the rest of the country (Fig. 2d). Almost all parts of St. Catherine protected area had low habitat suitability for A. lucidus (Fig. 3d). Altitude contributed most to the model (Fig. 4). The contribution of this factor is 69.6% (Table 2). The suitability of habitat for A. lucidus initially decreased sharply with the increasing of altitude (Fig. 5).

The model performance for A. melanista performed very well with an average AUC value of 0.86. The predicted distribution in Egypt showed areas with high suitability for this species as parts in South Sinai, Eastern Desert especially in the Red Sea coast and in the north-eastern part of the country, North Sinai, Gebel Elba and Coastal Strip. The habitat suitability was medium in some areas in Lower Nile Valley & Delta, Western Desert and Eastern Desert. The habitat suitability was low in the rest of the country (Fig. 2e). St. Catherine protected area is considered as a highly suitable habitat for this species in some parts (Fig. 3e). Temperature annual range (Bio 7) was the strongest predictor A. melanista distribution (Fig. 4). This factor gave the highest influence on the model with 47.6% contribution. The probability of presence A. melanista decreased with the increasing of the temperature annual range (Bio 7) (Fig. 5).

The current and future distribution for A. trifasciatus showed that Maxent model provided satisfactory results with an AUC value of 0.99. Current suitable habitat predicted that the only zone with high habitat suitability was South Sinai especially St. Catherine protected area (Fig. 2f, 3f), whereas the rest zones of the country showed a very low habitat suitability. Altitude contained more useful information than the other variables (Fig. 4). Altitude had the top contribution to the model with 98% (Table 2). The suitability of habitat for A. trifasciatus increased sharply with the increasing of altitude (Fig. 5). The habitat suitability in RCP 2.6 for the year 2070 in four selected GCM showed that A. trifasciatus decreased in their suitable habitat and this species suffered from future habitat loss in RCP 8.5 larger than in RCP2.6 (Fig. 6).

Fig. 6
figure 6

Predicted future habitat suitability of A. trifasciatus in St. Catherine protected area for 2070 year under four different climate change scenarios (GCM); a Beijing Climate Center (BCC-CSM 1_1), b National Center for Atmospheric Research (CCSM4), c Meteorological Research Institute (MRI-CGCM3) and d Norwegian Climate Centre (NorESM 1-M) [under RCP 2.6 emission scenario (1), under RCP 8.5 emission scenario (2)]

Full size image

Discussion

In response to global environmental changes, species ranges are fragmenting, contracting, expanding or shifting (Chen et al. 2011). The development of biodiversity databases in the last decade has provided new opportunities to analyze species occurrence data to support the efforts to conserve these species (Jetz et al. 2012), and many species distribution models (SDMs) or environmental niche models (ENMs) have increasingly proposed to support conservation decision making (Guisan et al. 2013).

About 100 years ago, late Efflatoun Bey and his co-workers started to collect flies from different localities all over Egypt. They recorded many fly species including the species treated herein from localities in the neighborhood of Cairo such as wadies south east of Cairo, Abu-Rawash, Kerdassa, Kafr Hakim, Kafr Eldems (= Mansheyet Radwan), Borgash, Mansouriah, Mazghouna, Sakkara, Helwan, Ezbet El-Nakhl, etc. (Efflatoun 1934, 1937, 1945). They also recorded many species from Mariout District on the Coastal Strip. However, many of these species have disappeared from these localities because of severe changes in the floral and faunal composition during the last 50 years. These changes may be attributed to urbanization or modern agricultural activities that caused severe degradation of natural habitats of these species (El-Hawagry et al. 2000; El-Hawagry 2015). However, the present study showed that the climatic and topographic variables in the Lower Nile Valley & Delta and Coastal Strip including these localities still have high suitability for most studied species.

Many of these species seem to be endangered or threatened, and conservation actions should be taken especially in localities or zones that still maintain their natural habitats. Sinai is one of these zones, and it has a merit that it includes seven natural protectorates, namely: Ras Muhammad, Azzaraniq, Ahrash, St. Catherine, Nabaq, Abu Galum and Taba (Leach et al. 2013). Undoubtedly, managing any conservation actions in an area already declared as a natural protectorate will be easier and more applicable than in other areas. The present study aims to contribute to the implementation of desired conservation actions, and to be used as a guide for any management decisions.

The results indicate that St. Catherine area has the highest habitat suitability for the six-studied species except A. lucidus. The species Anthrax trifasciatus, particularly, is recorded only from Sinai and not recorded elsewhere in Egypt. One of most interesting results of the six species through the area is that each species has its own niche (Fig. 3). By analyzing the six maps, it appears clearly that they have different habitat suitability. Such result shows how species of the same genus tend to have different ecological requirements to reduce competition even if they share the same habitat.

Fortunately, the St. Catherine protected area is expected to be maintained and will not face the severe changes in the floral and faunal composition as happened in the Lower Nile Valley & Delta and the Coastal Strip. However, future scenarios of climatic changes in St. Catherine for one of the species, Anthrax trifasciatus predict a great degradation in habitat suitability in the area.

St. Catherine area is about 1650 m above sea level. It also contains the highest mountains in Sinai (Gebel Catherine, 2642 m high, and Gebel Mussa, 2285 m high) (El-Moursy et al. 1999; Zalat et al. 2008). It has unusual climatic conditions as compared to the climate of the rest of Egypt. It is determined primarily by the altitude, the effect of which masks that of latitude. There is a wide difference in temperature between Summer and Winter. August is the hottest month, with a mean temperature of 25.9 °C, and January is the coldest month, with a mean temperature of 7.7 °C. The effect of the high mountains of St. Catherine is to make the climate continental, and to augment the difference between day and night. The maximum mean relative humidity occurs during January (64%), while the lowest mean occurs during May (Semida 1994; El-Rayes 1992).

The vegetation diversity in Sinai Peninsula including St. Catherine is exceptionally large and significant. The vegetation is characterized by the dominance of four large families, namely Asteraceae, Fabaceae, Lamiaceae, and Brassicaceae. One of the most common plants in St. Catherine is Alkanna orientalis (L.) Boiss. (Asclepiadaceae).

El-Moursy et al. (1999) studied the foraging behavior of anthracine bee flies in southern Sinai, especially in Wadi El-Arbaein, St. Catherine area, and they noticed that Anthrax spp. were present abundantly only at certain sites in the wadi, and they were foraging only on Alkanna orientalis in spite of the presence of many other wild plants in the same area. That’s why they thought that there might be a relation between the diversity and abundance of these anthracine flies and the plant cover of Alkanna orientalis (Gilbert et al. 1996). Many dipterans love to visit yellow flowers. The signal effect of the compact inflorescence of Alcanna flowers is striking and attracts insects even from a greater distance-especially those, which look for open flowers. Fabaceae and Lamiaceae have zygomorph flowers, where the nectar sources are hidden. Could it be that the Anthrax flies meet their special hosts on the flowers and follow them into their nest? (Marston 1971). The existence of Alcanna may act as meeting point for the solitaire bees and in the second row to Anthrax.

Recently, some works used environmental niche modeling to analyse the status of some higher Diptera especially of medical and veterinary importance (Mulieri and Patitucci 2019). While the present study represents the first environmental niche modeling study on bee flies, it sheds light on the status of six of the Egyptian species that are thought to be threatened. The results can be used for making powerful conservation management decisions in St. Catherine Protectorate to incorporate these Anthrax spp. in any present or future conservation programs.