The African rhinoceros beetle, Oryctes monoceros (Olivier; Coleoptera: Scarabaeidae), as a serious pest of palms such as Phoenix dactylifera L. (Kabiru et al., 2014), Elaeis guineensis, Jacq. (Bedford, 2014), and Cocos nucifera L., is responsible for about 40% of coconut losses in tropical Africa (Allou et al., 2006). Apart from palms, O. monoceros attacks non-palm hosts, such as Saccharum officinarum L., Ananas comosus L., and Musa paradisiaca L. O. monoceros inflicts damage to palms in both direct and indirect ways. The beetle directly stunts the growth of the palms by boring and laying eggs in the soft growing points (Bedford, 2013). Indirectly, O. monoceros feeding exposes palms to secondary infestation by the African palm weevil, Rhynchophorus phoenicis L., which enters the palms via holes or galleries created by O. monoceros. At present, O. monoceros has been reported primarily in Africa (Philippe and Dery, 2004; Allou et al., 2006; Edijala et al., 2009; Bila et al., 2019; EPPO, 2021), and Middle East (Arrow, 1937; NHM, 1938; Lepesme, 1947; EPPO, 2021). However, predictions of potential suitable climate regions for preventive and quarantine measures are generally lacking for O. monoceros.

There are over 40 Oryctes species worldwide (Bedford, 1976). Despite the species’ similarities, morphometric features on different life stages, such as adults and larvae, have been critical in distinguishing these species. For example, in O. monoceros, the anteroventral border of the prothoracic shield is rectilinear, whereas in Oryctes rhinoceros Linnaeus, it is scalloped and more concave (Hurpin and Fresneau, 1970). Adults of O. rhinoceros are gregarious, however just a single O. monoceros is seen feeding on the growing point of palms in most occasions (Philippe and Dery, 2004). O. monoceros is found across Africa, whereas O. rhinoceros is widely distributed in South and Southeast Asia (EPPO, 2021). O. monoceros differs from other Oryctes species like Oryctes boas Fabricius and Oryctes gigas Verkrüzen by having a third tooth on the right mandible between the scissorial and molar sections. Furthermore, O. monoceros has raster teges with 78–120 small sharp setae and a lower anal lip with 60–80 similar setae (Bedford, 1979). On the other hand, each claw of O. gigas has three to five long setae, whereas each claw of O. boas has two long setae. The diameter of the O. boas prothoracic shield is almost equal to its length, according to Hurpin (1969), with rectilinear and parallel anterior and posterior edges.

Changes in the earth’s temperature due to an increase in greenhouse gases have been shown to alter the distribution of species (Dhaliwal et al., 2010; Netherer and Schopf, 2010). Insect life cycles and population dynamics will undoubtedly be affected by global warming, including their growth rate, voltinism, and dispersal patterns (Bale et al., 2002), which influence the distribution and seasonal activity of species (Halsch et al., 2021). It can also induce changes in the habitat suitability of agricultural pests, thereby increasing invasion or exacerbating outbreaks and eventually optimizing the likelihood of damage to food production and food security.

Environmental niche models (ENMs) combine species occurrence records and environmental datasets, build a model using a machine learning algorithm, simulate environmental requirements of species, and project the analysis at a different time and space to predict its habitat suitability (Warren and Seifert, 2011; Valencia-Rodríguez et al., 2021). Environmental niche modeling uses correlation and mechanistic or process-based models to estimate species’ habitat requirements. The mechanistic species distribution model (e.g., CLIMEX) considers how environmental conditions constrain the physiological characteristics of species at a given location (Louvrier et al., 2020). In contrast, correlative models establish mathematical correlations between observable species distributions (presence or absence) and environmental variables [e.g., Maximum Entropy, Boosted Regression Trees (BRT), and Random Forest (RF)] (Phillips et al., 2006; Yu et al., 2020). Each model is theoretically distinct, has distinct data requirements, and employs unique analytic techniques. However, the BRT model is considered as a more robust technique than the other ecological niche models (Yu et al., 2020). The model works well with large datasets or when there are a lot of environmental factors compared to the number of observations (Elith et al., 2008). No prior data modification or outlier removal is required to fit the BRT model’s complicated non-linear relationships and automatically manage predictor interaction effects (De’Ath, 2007). The main limitation of single tree models is its low predictive powers, which can be improved by using BRT to fit multiple trees (Elith and Leathwick, 2017). The BRT model has been used to identify habitat suitability for Aedes aegypti and Aedes albopictus (the arbovirus vectors), based on land-cover and environmental variables (Kraemer et al., 2015). The model has also been used to model a mountain pine beetle distribution (Ramazi et al., 2021). The output from such models provides baseline information for monitoring, surveillance, and development of ecologically friendly management strategies (Mu et al., 2013).

Several studies on O. monoceros have focused on its biology and ecology (Ukeh et al., 2003; Allou et al., 2006; Edijala et al., 2009; Kabiru et al., 2014; Gbangboche et al., 2016), and cultural management practices, such as farm sanitation by destroying felled wood and palms or covering them with cover crops to minimize larval breeding (Mariau and Calvez, 1973).

However, due to the high planting density of palms and the expensive expense of these management measures, they are rarely adopted (Allou et al., 2006). Attempts to manage the pests using biological control have been unsuccessful because none of the parasitoids or pathogens (Baculovirus oryctes) have provided a long-lasting solution to the threat posed by O. monoceros (Seguni et al., 1999). In addition, application of synthetic pesticides has not been recommended due to environmental and health risks (Parven et al., 2021). In Ghana, a new method for controlling O. monoceros in coconut plantations is extracting adults using a metal mini-harpoon from their feeding galleries (Philippe and Dery, 2004). This mechanical control strategy is labor-intensive and costly in high-density palm plantations. Also, since preventive measures are economically cheaper than post entry management (Arthur et al., 2015), creating risk maps would serve as an early warning against the future spread, especially in areas where the pest is currently absent.

The average global temperatures would rise ranging from 2 and 9.7°F (1.1 to 5.4°C) by 2100 in a variety of greenhouse gas emission scenarios (Herring, 2012). In recent decades, it is accepted that a rise in global temperature is considered as one of the causes of the drastic change in climate (Soytas and Sari, 2006; Karki, 2007). Climate change is an important factor restricting insects’ survival and population growth (Vanhanen et al., 2007). However, there is a scarcity of information on the effects of climate change on the global distribution of O. monoceros as well as identifying environmental factors driving the geographical distribution for surveillance, monitoring and preventive programs is generally lacking. To address this research gap, we simulated the potential distribution of O. monoceros under different climate change scenarios to identify climate suitable areas for biosecurity and risk assessment.

Our research included two key procedures: (i) obtaining occurrence records and climate datasets; and (ii) predicting the pest’s existing and future potential environmental niche shifts. The BRT modeling framework is a prominent technique to ecological studies that has been used to accurately forecast the potential geographic range of many species (Jiang et al., 2019; Kraemer et al., 2019; Chen et al., 2020; Ding et al., 2020). The BRT modeling approach combines the advantages of regression trees (Breiman et al., 1984) and gradient boosting (Friedman, 2001), and has a good ability in fitting complex non-linear response functions from outcomes based on given environmental covariates. To assess the effects of climate change on environmental suitability for O. monoceros, we used the following datasets as key input data for modeling analysis: (a) the assembled contemporary occurrence records of O. monoceros, and a set of pseudo-absence points of pest; (b) a set of environmental covariates reflecting a baseline scenario for the year 2020; and (c) a set of environmental covariates reflecting two representative concentration pathways (RCPs) (R 4.5 and 8.5) scenarios on future periods (the 2030s, 2050s, and 2080s).

The occurrence datasets used in the modeling were mostly taken from scientific papers. This is consistent with the open science movement’s definition, which encourages data reuse for further research and decision-making (Allen and Mehler, 2019). As a result, some studies have used historical species data to estimate invasion risk areas (Jiang et al., 2018; Kraemer et al., 2019; Guo et al., 2021; Zhou et al., 2021; Aidoo et al., 2022a,b). The model outputs from such studies have been utilized in developing surveillance, monitoring, and preventive measures against invasive pests (Hao et al., 2022).

The African rhinoceros beetle has been observed in Africa and Asia (Ukeh et al., 2003), showing an expanding trend in recent decades. O. monoceros, an important pest of palms in tropical Africa (Allou et al., 2006), causes server damage to these plants, thus leading to economic losses. Therefore, predicting O. monoceros suitable areas can help monitor the pest, and develop early warning, and management programmes wherever it occurs. In this study, we used the BRT model and ArcGIS to understand the potential climatic risk factors and estimate the global distribution of O. monoceros under climate change.

The environmental factors that influence the pest distribution were annual temperature variation, followed by the seasonality of temperature, the isothermality, precipitation of the hottest quarter months, and average variation of day temperature. This suggests that temperature factors contribute more to the global geographical distribution of the pest, which may explain why temperatures and relative humidity ranging from 27 to 29°C and 85 to 95%, respectively, are ideal for larval growth (Bedford, 1980).

The African rhinoceros beetle has been reported in two Asian countries: Yemen and Saudi Arabia. According to our predictions, extensive areas of Asia, though remaining uninfested at present, are highly favorable to the pest. For example, the southeastern parts of India, Vietnam, Cambodia, and Thailand are predicted to be suitable for O. monoceros. In the present study, our model predicts high habitat suitability in the world’s largest coconut-producing countries, such as Indonesia, Philippines, India, Sri Lanka, Brazil, Vietnam, Papua New Guinea, Mexico, Thailand, and Myanmar (FAOSTAT, 2019). These largest coconut-producing countries will remain suitable to O. monoceros in the future. In addition, Indonesia, Malaysia, Thailand, Colombia, Nigeria, Guatemala, Honduras, Papua New Guinea, Ecuador, and Brazil, as the leading producers of oil palm (FAOSTAT, 2019), were all predicted to have suitable areas for the pest. Our predictions show habitat suitability for O. monoceros in Sudan and Saudi Arabia, which is quite discouraging for the date palm industry in these regions since these countries are among the major date palm producing countries in the world. These regions should be alert to the possibility of O. monoceros invasion.

In this study, the BRT model predicted habitat suitability for O. monoceros in areas where the Asiatic rhinoceros beetle O. rhinoceros Linnaeus (Coleoptera: Scarabaeidae) is currently present. O. monoceros and O. rhinoceros are native to Africa and Asia, respectively (Hinckley, 1973). Oryctes rhinoceros has spread outside its native range inflicting damage to economically important palms in the Americas, Oceania and Africa (Giblin-Davis, 2001). The countries affected by O. rhinoceros include Mauritius, Nigeria, and Reunion in Africa; India, Indonesia, Philippines, Sri Lanka, and Vietnam in Asia; Texas and Hawaii in the Americas; Papua New Guinea and American Samoa in Oceania (Catley, 1969; Onyeike et al., 2005; Okaraonye and Ikewuchi, 2009; Olowu et al., 2012; EPPO, 2022). O. monoceros has not been reported from Oceania, Southeast Asia, and the Americas. As predicted in this study, parts of these areas have medium to high habitat suitability for O. monoceros. Given the substantial economic impacts of Oryctes species, the invasion of O. monoceros in countries where O. rhinoceros occurs could pose a severe threat to food security and biodiversity conservation. Therefore, our risk maps would guide regular monitoring and surveillance of O. monoceros, thereby controlling the pest in an ecologically friendly.

Global warming would cause changes in species distributions (Kiritani, 2006; Freeman et al., 2018; Chen et al., 2019). Our results indicated that future predicted suitable areas occurred outside the present distribution of O. monoceros. The new areas were mainly found in South America, the Caribbean, Asia, and Oceania. On the one hand, our predictions revealed that the distribution of O. monoceros would expand under future climate scenarios (present < 2030 < 2050 < 2080), indicating that there would be more suitable habitats for O. monoceros distribution in the future. The range shift was more noticeable in Central America than in any other region. The predicted range contraction for O. monoceros in this study could be useful for pest control and eradication. The range of habitat suitability will potentially contrast in some parts of the world, making this pest less competitive and possibly contributing to its eradication. On the other hand, a few areas are predicted to remain suitable and may require regular monitory and surveillance for early detection and control, especially in regions such as parts of Australia that currently have no occurrence records of O. monoceros. The predicted suitable areas are primarily in line with earlier pest occurrence records and show expansion of the pest’s geographical distribution.

Our study identifies the regions where O. monoceros would most likely occur, as well as the relationship between the environmental indicators of this pest and the risk of its invasion, especially in areas where it has not yet been observed. It is important to note that there are some limitations in this study. Firstly, the environmental suitability maps generated from the ensemble BRT models can be interpreter to predict the potential geographical distribution of the O. monoceros, rather than where the pest will be spread in the future. Secondly, factors such as the interaction of the invader’s characteristics (e.g., high propagule output), the abiotic environment (especially disturbance), and biotic interactions within novel environments (e.g., enemy-release) (Sakai et al., 2001; Keane and Crawley, 2002; Shea and Chesson, 2002) were not included in the BRT model because of unavailability of data. However, these biotic and abiotic factors are involved in the ability of a pest to establish in a new geographical area. Among them, competitors, the ability to survive in the absence of hosts, and phenotypic plasticity are considered important influencers of the establishment of species outside their native range (Lee and Lee, 2006; Aidoo et al., 2021). Thirdly, local policies, such as the inspection of plants, plant parts, farm machinery, and plant by-products at entry ports, can affect the establishment of this species outside of its natural range (McNeely, 2000; Pyšek et al., 2020). It is our recommendation that the implementation of plant protection and regulatory procedures, the development of programs to monitor the quality of plant products and byproducts, as well as an increase in academic awareness and the transfer of technology can help prevent the spread of this pest in areas where it has not yet been detected (Shine, 2007; Early et al., 2016).

In this study, we have combined the BRT models with O. monoceros occurrence and environmental data to predict the potential geographical distribution of the pest. The AUC value of the model was excellent, suggesting that our model predictions are accurate and can be used to inform policy formulation and preventive measures. From now until 2080, we found that the worldwide habitat suitability for O. monoceros increased, as predicted by the BRT model. Therefore, to prevent more O. monoceros invasions in the face of present and future climate change scenarios, many of the world’s largest oil palm, coconut, and date palm plantations must take preventative steps.

The original contributions presented in this study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.

OFA, MH, FD, TM, and YQ wrote the manuscript. FD, OFA, TM, NY, CB, ET, and KD conceived and designed the research. FD, DJ, TM, OFA, DW, MH, and TM analyzed the data. All authors reviewed the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.