Predicting the potential geographical distribution of parasitic natural enemies of the Dubas bug (Ommatissus lybicus de Bergevin) using geographic information systems

Abstract The Dubas bug (Ommatissus lybicus de Bergevin) is a pest species whose entire life cycle occurs on date palms, Phoenix dactylifera L, causing serious damage and reducing date palm growth and yield. Pseudoligosita babylonica Viggiani, Aprostocetus nr. Beatus, and Bocchus hyalinus Olmi are very important parasitic natural enemies of Ommatissus lybicus in northern Oman. In this study, random farms were selected to (a) model the link between occurrences of the Pseudoligosita babylonica, Aprostocetus nr beatus, and Bocchus hyalinus (dependent variables) with environmental, climatological, and Dubas bug infestation levels (the independent variables), and (b) produce distribution and predictive maps of these natural enemies in northern Oman. The multiple R 2 values showed the model explained 63%, 89%, and 94% of the presence of P. babylonica, A. nr beatus, and Bocchus hyalinus, respectively. However, the distribution of each species appears to be influenced by distinct and geographically associated climatological and environmental factors, as well as habitat characteristics. This study reveals that spatial analysis and modeling can be highly useful for studying the distribution, the presence or absence of Dubas bugs, and their natural enemies. It is anticipated to help contribute to the reduction in the extent and costs of aerial and ground insecticidal spraying needed in date palm plantations.

( Figure 1). The direct damage arises when the nymphs and adults feed by sucking sap from leaflets and rachis during the spring and autumn generations (Aldryhim, 2008;El-Shafie, Peña, & Khalaf, 2015).
Direct damage also results from injury to tissue due to egg laying activity (Bagheri, Fathipour, Askari-Seyahooei, & Zeinolabedini, 2016;Fathipour, Bagheri, Askari-Seyahooei, & Zeinolabedini, 2018;Khalaf & Khudhair, 2015). Indirect damage comes from the presence of honeydew (sticky liquid excretion), which allows dust accumulation and sooty mould growth, causing the possible deterioration of the palm fruits and other crops that are cultivated underneath the palm trees. The copious feeding of DBs weakens the tree, and the honeydew's accumulation of dust and mould reduces photosynthesis and other physiological processes on the fronds' surfaces, which then become chlorotic after several months (Bagheri, Fathipour, Askari-Seyahooei, & Zeinalabedini, 2017;Klein & Venezian, 1985;Mokhtar & Al Nabhani, 2010). A high DB population can cause a date palm tree to lose its vitality, leading to a decline in its productivity (Thacker, Al-Mahmooli, & Deadman, 2003).
Controlling DBs requires the effort and money of several countries worldwide, including the Sultanate of Oman (Thacker et al., 2003). Since DBs were first recorded in Oman in 1962, activities aiming to control DB infestations have focused on the use of insecticides via annual ground and aerial sprayings. However, the control of DBs using insecticides has led to many problems. The authorities have mainly and increasingly relied on them to manage this pest, a process that costs the government large sums of money every year (Al-Zadjali, F I G U R E 1 Dubas bug, Ommatissus lybicus life cycle (Al-Khatri, 2011) Abd-Allah, & El-Haidari, 2006). From 1993 to 2011, about 523 tonnes of insecticides were used in aerial applications to control DB infestations, at an estimated cost of $18.5 million (Al-Khatri, 2011). In 2016, the Omani government spent $1,550,000 and $395,630 for the spring and autumn generations, respectively, to control DB infestations (direct communication with the Plant Protection Department, Ministry of Agriculture and Fisheries, Oman).
Insecticides have been banned in many countries including Oman because of their perceived negative environmental effects, such as the pollution of water resources; deterioration of human health; and the resulting reductions in populations of nontarget species, particularly the natural enemies of DB and because of the relatively high application cost (Ansari, Moraiet, & Ahmad, 2014). Studies have shown that some insecticide residues can persist on date palm fruit for up to 2 months after application (Khan, Azam, & Razvi, 2001), which poses a high risk to both humans and animals. Moreover, aerial spraying is difficult to conduct on farms located within mountains, owing to high elevations and valley features (Al-Kindi, Kwan, Andrew, & Welch, 2017a). Thus, chemical control measures have resulted in limited success in Oman, where DBs continue to pose a major challenge to the date palm industry.
The natural enemies of DB consist of a range of predatory, parasitic, and pathogenic species (Howard, 2001). Of the parasitic species, there are three relatively specific parasitoid wasp species that are present in Oman. Due to their specificity, they may be highly important factors for regulating the population of the DB. Two of these species attack the egg stage, while the third attacks the nymphal and adult stages. Relatively few studies exist on these species, and most of these have focused on the internal egg parasitoid Pseudoligosita babylonica Viggiani (Hymenoptera: Trichogrammatidae). This species has been recorded in Iraq, Oman, Iran, Saudi Arabia, and Yemen (Al-Khatri, 2011;El-Shafie, 2012;Hassan, Al-Rubeai, Al-Jboory, & Viggiani, 2004;Hubaishan & Bagwaigo, 2010). It may also be present in other countries where the DB exists. It is a small, stout, yellowish wasp measuring 0.7 mm in length that parasitises and kills DB eggs (Hubaishan & Bagwaigo, 2010). Each parasitoid wasp develops inside a single DB egg. A significantly higher proportion of date palm leaflets, compared to interleaflet areas (the frond parts between leaflets) showed signs of egg parasitism by this wasp (AKA, unpublished data). According to Al-Khatri (2011), P. babylonica has three generations during each of the spring and autumn generations of the DB: a pre-DB egg hatching generation, a generation coinciding with DB egg hatching, and a post-DB egg hatching generation. Al-Khatri (2011) studied the biology and ecology of P. babylonica in Oman and stated that this egg parasitoid could be considered a potential biological control agent of DB.
F I G U R E 2 The study areas: (a) location and topography, (b) nine governorates in the Sultanate of Oman, with date palm plantations sampled for DB and its natural enemies highlighted in red The second parasitoid associated with the eggs of DB is Aprostocetus nr. beatus (Eulophidae: Hymenoptera) (Kinawi, 2005).
The mature larva of this wasp has been observed to be an external feeder on one or more eggs of the DB within the midrib area of date palm leaflets (AKA, unpublished data), but younger larvae could act as internal egg parasitoids, as demonstrated in the case of A. . beatus attacking eggs of other planthoppers and leafhoppers in other parts of the world (Carnegie, 1975(Carnegie, , 1980Kosheleva & Kostjukov, 2014).
The nymph-adult parasitoid Bocchus hyalinus Olmi (Hymenoptera: Dryinidae) has been recorded as present in Oman, the United Arab Emirates (UAE), and Kenya (Olmi, Copeland, & Guglielmino, 2015). Olmi (1998) described this wasp more than a decade ago, based on a female specimen collected in Oman. This small wasp species develops in the nymphs and adults of the DB, producing a dark pouch or sac that protrudes from the abdomen of its host. Although both sexes are winged, the female's body is generally yellowish-orange, measuring 2.8-3.8 mm in length, whereas the male is black to brown and measures 1.5-2.25 mm (Olmi, 2008(Olmi, , 2009. Although a few studies exist on the above parasitic natural enemies of DB (e.g., P. babylonica, A. nr. beatus and B. hyalinus) (Al-Khatri, 2011;Hassan et al., 2004;Hubaishan & Bagwaigo, 2010), none have applied geographic information system (GIS) and spatial analyses to these species in the Sultanate of Oman. In addition, the effective biological control of a pest requires biological and ecological knowledge: host plant(s), host insects, biological control agent(s), and the locations (areas) where the biological control agent is going to be used.
No focused research exists on the spatial and temporal distributions of these natural enemies and their relationships with environmental, climatological, and resource availability conditions. Practical tools and approaches are needed for mapping, analyzing and, more importantly, predicting the distribution of these important DB natural enemies and for understanding the interactions between these enemies and environmental and climatological factors, so that decision makers may improve the management of DB infestations on a large scale.
The main objective of this study was to investigate how the environmental, climatological, and DB infestation level variables impact populations of P. babylonica, A. nr. Beatus, and B. hyalinus in northern Oman. We considered environmental (e.g., elevation, slope, aspect or direction of slope, soil, water type and distance to streams), climatological (e.g., temperature, humidity, rainfall, and wind direction), and DB infestation level variables and incorporated them into a GIS platform to evaluate which combinations of variables are associated with these natural enemies of DBs. This study will also allow for the prediction of the future distribution of populations of these three potential biological control agents of DB in Oman.

| Study area
Oman covers an area of 309,500 square kilometers and extends from 16°40′N to 26°20′N, and 51°50′E to 59°40′E. It occupies the south-eastern corner of the Arabian Peninsula. It has 3,165 km of coastline, extending from the Strait of Hormuz in the north to its border with the Republic of Yemen in the south. The coastline faces three bodies of water, namely the Arabian Sea, the Persian Gulf (also known as Arabian Gulf), and the Gulf of Oman (Figure 2).
To the west, Oman is bordered by the UAE and the Kingdom of Saudi Arabia. Mountainous areas account for 15% of all land area, while desert plains, sandy areas, and coastal zones cover 77% of the land area. The remaining land area is covered by agricultural land. The elevation ranges in the study zone from 0 to 2,994 m above sea level, and the soil contains five soil-type categories: clay, gravel, loam, rock, and sand. The water in the area can be categorized into three classes based on total dissolved content: brackish water, freshwater, and saltwater (Al-Kindi et al. 2017a). Oman has an arid climate, receiving <100 mm of rain per year; however, the mountainous parts of the country enjoy higher precipitation levels. As one of the independent variables, DB infestations occur where palm trees are concentrated; therefore, in the present study, we focused on northern Oman, the area from latitude 26°50′N to 22°26′N and from longitude 55°50′E to 59°50′E (see Figure     were chosen at random. They included, for the most part, farm sites (cultivated or neglected) in villages and towns or, more rarely, aggregations of wildly growing date palms in wadis. Sites were marked geographically using GPS device (Garmin) to give positional and elevation data, and site names were recorded.

| Sampling of DB eggs
Three to five trees were selected randomly at each site. In most cases, samples were taken from shorter trees, those up to 2 m in height (growing point height). Fronds of these trees were reachable from the ground, making them easy to examine and cut. One middleaged green frond was cut from each tree. All fronds were examined for the presence of new DB eggs. At farm sites with tall trees, a farm attendant/laborer was asked to climb each tree to collect one green frond. Each collected frond was then cut into 3-6 pieces, which were placed in large trash bags. In some cases, leaflets were excised from the frond (using pruning scissors) and then placed in bags. Trash bags with frond material were placed in a shaded area in the field and then moved inside a large cool box. Finally, they were transported to the entomology laboratory at College of Agricultural and Marine Sciences, Sultan Qaboos University.

| Sampling of DB nymphs and adults
Sampling was performed on relatively short date palms of up to 2 m in height. At least five trees were sampled at each site, but when the DB populations were low, up to ten trees were examined. Nymphs and adult DBs were sampled using three methods. When populations were relatively large, suction was applied using handheld vacuum machines (Black & Decker car vacuum, and Bio-quip custom-made vacuum). DBs collected via a vacuum machine from a particular site were pooled inside a large jar. When populations were relatively low, one of the two following methods was used: Removal of leaflets infested with nymphs and adults using pruning scissors, and then, leaflets were placed inside a large jar; or shaking of the leaflets to dislodge the insects into a large jar. The field samples were taken roughly between 8 a.m. and 11 a.m.
as this is when the insects are feeding and not moving off the leaflets due to excessive heat exposure.

| Processing egg samples in the laboratory
If not already performed in the field, leaflets were separated from the rest of the fronds in the lab. All leaflets and frond pieces were then checked for new eggs. Leaflets and frond pieces with new eggs were retained, and the rest were discarded. The apical parts of leaflets were cutoff, and the leaflets were placed into large, 5-litre jars with a small amount of water at the bottom. Some space was left in the jar between leaflets, jar to allow for sufficient aeration. Leaflets from each farm site were combined together, but in cases with a large number of leaflets, they were distributed in more than one jar.
Frond pieces were placed in 5-litre jars in a similar manner to that described above for leaflets. All jars were labeled with site infor- Notes.An asterisk next to a number indicates a statistically significant pvalue (p < 0.01).
Interleaflet frond areas and leaflets with observed adult parasitoids were marked with red ink and kept separate from the rest of the frond material, to make it easier to follow up on the progress of parasitism of eggs. The material was also checked for the emergence holes of parasitoids in or around DB eggs. Emerged or observed stages of parasitoids were photographed.

| Parasitoid data
Although the study samples came from many governorates, data used

| Data and spatial analysis
To understand the factors behind observed spatial patterns or to predict spatial distributions, regression analysis methods are useful for modeling, examining, and exploring these relationships. In this study, ordinary least square (OLS) and geographically weighted regression

| Natural enemies mapping
The most prevalent parasitoid species was P. babylonica, which was found at 66 sites, followed by Aprostocetus nr. beatus at 35 sites and B. hyalinus at 33 sites (see Figure 3). Of the P. babylonica, Aprostocetus, and B. hyalinus data, a total of 168 collection sites returned information. The collection sites were located predominately in nine geographical governorates in northern Oman. The distribution of P. babylonica occurred in all nine governorates but varied across the study area. Aprostocetus nr. beatus distributions were found primarily in seven governorates, being absent in Al-Dhahirah and Ash-Sharqiyah South. By contrast, B. hyalinus distributions were found in seven governorates, but not in Musandam or Ash-Sharqiyah North (Figure 3).

| Nearest neighbour statistical analysis
The results of our nearest neighbour statistical analysis, in which the nearest neighbour ratios were 0.520, 0.636, and 0.689, respectively, showed that the expected mean (EM) distance (or spacing) of Aprostocetus nr. beatus, B. hyalinus, and P. babylonica distributions were, respectively, higher than the observed mean (OM), with the difference less than zero (a negative number).
These results indicated that the distributions of Aprostocetus nr.

| Spatial relationships modeling analysis
The model variables that best explain the occurrence of P. babylonica, A. nr. Beatus, and B. hyalinus in the study area, along with their variance inflection factors (VIF), are shown in Table 3. VIFs are based on tests designed to measure whether two or more factors are telling the same story (O'brien, 2007). The idea is that any factor that has a value of higher than 7.5 should be removed from consideration.
The preliminary steps in the OLS analysis, multicollinearity testing by mean square deviation and VIF were run separately (once per species) on the pool of 13 independent factors, which were selected to correspond to the conceptual model (see Table 2). The average VIF value in Model 1 was 1.49, followed by Model 2, at 2.11 and 1.96 in Model 3 (see Table 3). The VIF values in the three models are all well under the ESRI-defined threshold of 7.5, confirming that these factors are not redundant (see Table 2).
OLS and multiple regression revealed that the P. babylonica distribution had significant positive relationships p-value (p < 0.01) with DB infestation level (R 2 = 0.80), elevation (R 2 = 0.83), wind direction (R 2 = 0.64), humidity (R 2 = 0.45), and water type (R 2 = 0.48); however, significant negative correlations were found between temperature (R 2 = 0.53) and P. babylonica presence (Table 3). Table 3 shows significant positive association p-values (p < 0.01) were found between A. nr. beatus and elevation (R 2 = 0.90), DB population (R 2 = 0.88), and wind direction (R 2 = 0.71); nevertheless, significant negative associations were found between the temperature (R 2 = 0.93), humidity (R 2 = 0.94), and wind speed (R 2 = 0.60) factors and A. nr. beatus presence. In addition, significant F I G U R E 7 Frequency distributions of P. babylonica, A. nr. Beatus, and B. hyalinus with respect to elevation in the study area positive relationships were found between B. hyalinus and DB population (R 2 = 0.81), elevation (R 2 = 0.72), humidity (R 2 = 0.46), rainfall (R 2 = 0.74), and water type (R 2 0.39); but significant negative relationships were found between temperature (R 2 = 0.94) and wind speed (R 2 = 0.37) on the one hand and B. hyalinus presence on the other hand (see Table 3). Table 3 were significant at a confidence level of >95%, which indicates strong relationships between individual exploratory factors and the dependent variable. The coefficient values displayed in Figure 4 shows the impact of each of the variable that has the strongest correlation with P. babylonica, A. nr, and B. hyalinus, and that other variables still predicted a strong correlation with P. babylonica, A. nr. Beatus, and B. hyalinus distributions in the study area.

Several factors in
The significance of the models' predictions is evident in the mapping of residual standard deviations. The models produced under predictions, although it is likely that other variables could also predict strong correlation with P. babylonica, A. nr, and B. hyalinus in the study area as shown in ( Figure 5). The three parasitoid models explained 63%, 89%, and 94% of the impact of environmental, climatological and DB infestation levels on P. babylonica, A. nr. Beatus, and B. hyalinus, respectively.

| D ISCUSS I ON
Results shown in Table 1 indicated that the expected mean distances of the species' distributions were greater than the observed mean distance. These results indicated the presence of clustered distributions of P. babylonica, A. nr. Beatus, and B. hyalinus in the study area.
The results of the OLS regression method revealed models that confidently predict 63%, 89%, and 94% of the influence of DB infestation levels, and climatological and environmental variables on the P. babylonica, Aprostocetus nr. beatus and B. hyalinus presence in the study area, respectively.
Although P. babylonica is by far the most important natural enemy of DB, the result shows that the ability of the studied independent variables to predict the distribution of this natural enemy is relatively low (63%) compared with the two other parasitic. One possible explanation for this contrast between P. babylonica and the other two parasitoids involves differences in life history between the three species. P. babylonica is an internal egg parasitoid of DB where factors such as DB infestation, temperature, humidity, rainfall, and water type were important predictors in calculating the probability of occurrence for B. hyalinus (see Figure 6). SE. The southwest wind, which is commonly known as the "kous" in the Gulf countries, is warm and moist. This wind may bring with it viable conditions for these two species to survive in northern Oman. In contrast, we found no influence of wind direction on the B. hyalinus.  Haslett, 1990) of DBs and their enemies. In general, there are two approaches of identifying patterns in geographical data. The first is to display features on a map without conducting any statistical analyses, as showing the data in a spatial format can be valuable endeavor, even without detailed analysis (see Figure 3).
The second approach is to use spatial statistics to measure the extent to which features are clustered, dispersed, or random (Getis & Ord, 1992). Each of these measures is important when comparing the patterns for different sets of features or when comparing patterns across a given area (Kozak, Graham, & Wiens, 2008).
Geostatistical analyses allow us to generate optimal surfaces from sample data and to evaluate predictions, leading to better decision-making. GIS techniques are helpful in many data analysis models, such as environmental, precision agricultural, wildlife, and ecological studies (Al-Kindi et al., 2017d). However, the results of the present study are in the form of first approximation models. More data and ecological information on P. babylonica, A. nr.
Beatus, and B. hyalinus could produce better results in the future.
Hence, more surveys are needed to determine the distribution and density of these natural enemies for controlling DBs and for recording other enemies that can be used successfully against DBs throughout Oman. With this in mind, the relationship between natural enemies and DB infestation should be investigated prior to any planning for chemical control application. Farmers and the responsible authorities in Oman should take care of these natural enemies of DBs using alternative methods of insecticide application, to minimize impact on nontarget insects including beneficial ones such as honeybees and other pollinating insects as well as natural enemies of pest insects.

CO N FLI C T O F I NTE R E S T
None declared.