Introduction

Culicoides biting midges play an important role in the spread of vector-borne infectious diseases worldwide, transmitting disease agents to humans, domestic and wild animals. Among them, African horse sickness is a transboundary and non-contagious infectious disease of equids and is one of the most lethal equine virus infections known. The pathogenic virus is African horse sickness virus (AHSV), which belongs to the genusOrbivirus, the familyReoviridae (Howell, 1962). The World Organization for Animal Health classifies African horse sickness as a listed notifiable disease (OIE, 2021).

The epidemic area and seasonality of African horse sickness occurrence are related to vector epidemiology (Robin, 2019). At present,Meiswinkel & Paweska (2003) have shown thatCulicoides bolitinos is a proven vector of AHSV in South Africa.Culicoides imicola (C. imicola) is another important vector for field transmission of AHSV (Mellor & Boorman, 1995). It is widely distributed in most of the inhabited world, including Africa, southern Europe, and southern Asia, which existC. imicola around the years, and are also potential risk areas of African horse sickness occurrence (Guichard et al., 2014;Meiswinkel, 1989).

African horse sickness virus is endemic in sub-Saharan Africa, and it periodically invaded Europe and Asia (Carpenter et al., 2017). In 2020, Thailand and Malaysia successively reported the first incidence of African horse sickness, the first outbreak caused by AHSV-1 outside sub-Saharan Africa (King et al., 2020). The epidemic could pose a major threat to Southeast Asia and even other Asian countries. Currently, the effective way to prevent and control African horse sickness is to controlCulicoides population.

Under the increasingly severe situation of prevention and control ofCulicoides-borne diseases in the world, it is essential to better understand the possible geographical dynamics ofCulicoides vector. Species distribution models can assist in the targeted monitoring and the implementation of controlling programs for disease vector (Liu et al., 2019). Therefore, we used the ecological niche model to evaluate the current global distribution ofC. imicola, based on occurrence records and bioclimate variables. In addition, our modeling projected the future habitats in the 21st century based on global climate change.

Materials & methods

Culicoides imicola data and processing

We obtained theC. imicola presence points (n = 1,046) from the literature (Duan et al., 2021;Leta et al., 2019;Ye et al., 2019) and the Global Biodiversity Information Facility database (https://www.gbif.org/). All the occurrence data were spatially filtered at 10 km² grid cells (Radosavljevic & Anderson, 2014) to minimize the spatial autocorrelation using the Species Distribution Model Toolbox (Brown, Bennett & French, 2017). Thus, 729 spatially rarefied occurrence records ofC. imicola were included in the current and future model in this study (Fig. 1).

Results

The AUC value is 0.902 (±0.011) in the current model forC. imicola distribution (Fig. S1), indicating the excellent predictive power of the model. The contributions of the environmental variables were shown inTable S1, and the top three variables with greater contribution were important variables. Temperature seasonality was identified as the most important variable for model construction (30.3% contribution), followed by precipitation of coldest quarter (29.5% contribution) and mean temperature of wettest quarter (16.5% contribution).

The response curves of the variables reflect the environmental requirements ofC. imicola (Fig. 2). In this non-identifiability of prevalence, when the threshold of habitat suitability (i.e., logistic output) was greater than 0.5, the variation range represented by the horizontal axis was the optimal range (Elith et al., 2011). In the response curve of bio_4, temperature seasonality represented the standard deviation of the monthly mean temperature multiplied by 100. When temperature seasonality was about 564 (i.e., the standard deviation of the monthly mean temperature is 5.64), the predicted habitat suitability forC. imicola was better for values of optimal range from 403 to 631 (Fig. 2A). In the response curve of bio_19, the optimal range for precipitation of coldest quarter was from 115 to 462 mm. When precipitation exceeded 244 mm, the habitat suitability decreased rapidly, and the probability ofCulicoides distribution was 0 when precipitation exceeded 1,500 mm (Fig. 2B). For bio_8, the optimal range for mean temperature of wettest quarter was from 8.8 to 17 °C. In addition, the habitat suitability ofC. imicola was highest when the temperature was about 11.5 °C (Fig. 2C).

The potential habitat suitability ofC. imicola under current and future climate scenarios were shown inFig. 3. In the current climate scenario, southern North America, southern South America, southwestern Europe, most of Africa, the coastal areas of the Middle East, almost all regions of South Asia, southern China, a few countries in Southeast Asia, and the whole Australia were predicted as suitable habitats forC. imicola by the MaxEnt model. Among these areas, small parts of southern coastal Europe and northern coastal Africa were high suitability forC. imicola distribution. In the future climate scenarios, the predicted areas under different scenarios were generally similar, except for Australia under the RCP 4.5 scenario in the 2030s. However, suitability was expected to increase at high latitudes in the northern hemisphere, such as Norway, Sweden, Finland, and the Kola Peninsula. Moreover, in the low and middle latitudes of the southern hemisphere, the medium suitability areas in South America and high suitability areas in the western and southern coasts of Australia were predicted to decrease significantly.

The areas of different habitat suitability under current and future climate scenarios were shown inFig. 4, and the current suitable area ofC. imicola was estimated to be about 34,176,260 km². Except for the 2070s under RCP 8.5, the unsuitable areas forC. imicola will decrease continuously under all future climate scenarios. However, compared with the current climate scenario, the areas of low suitability will continue to increase in the future. The change of medium suitability areas under the RCP 2.6 scenario was higher, while the change of high suitability areas under the RCP 4.5 scenario in the 2030s and RCP 8.5 scenario in 2070s were higher (the increments were 27.70% and 21.61%, respectively). It is worth noting that the suitable areas under the RCP 8.5 scenario hardly increased in the 2070s, but decreased sharply.

Discussion

Temperature and rainfall are determining factors in the activity, abundance, and survival ofCulicoides (Mellor, Boorman & Baylis, 2000). Previous studies have demonstrated that temperature is positively correlated with theCulicoides activity, and adult vector activity was suppressed at low temperatures (Murray, 1987;Searle et al., 2014). AdultCulicoides activity was related to seasonality (Carpenter et al., 2011;Sanders et al., 2019). At the same time, high temperature favored larval development leading to faster population growth (Kitaoka, 1982;Vaughan & Turner, 1987). Furthermore, at high temperature, the vector virogenesis rates were high (Verhoef, Venter & Weldon, 2014;Wittmann, Mello & Baylis, 2002). However, laboratory studies have shown that individualCulicoides survived for a relatively short lifespan at very high temperature (Wellby, Matthew & Peter, 1996;Wittmann, Mello & Baylis, 2002).

On the other hand, rainfall can also influence the activity ofC. imicola, which will decrease at low moisture levels (Walker, 1977). In Australia, after rainfall, there was an increase in the feed time of manyCulicoides midges. The feeding frequency influences the host-biting rates; therefore, the population transformed to one capable of explosive transmission (Murray, 1986). At a suitable temperature, the abundance of the vector is often more closely related to rainfall. Diarra found that the largest amount ofC. imicola abundance appeared in the year of the greatest rainfall in Senegal (Diarra et al., 2014). In addition, rainfall influences the spread of African horse sickness by governing the availability of larval habitat and regulating the survival and dispersal of adultCulicoides (Purse et al., 2015). The water content can determine the suitable semiaquatic habitat for the larva (Meiswinkel, 1997), and more rain may produce more suitable habitats. However, if the habitats are flooded,C. imicola will drown (Nevill, 1967).

The predicted map showed that the neighboring countries, India, Myanmar, Southern China, Laos, Cambodia, and Vietnam were all in the suitable areas forC. imicola. Identification of such regions is related to biosecurity purposes. Bluetongue virus (BTV) belongs to the same genus as AHSV and mainly infects ruminants. Recently, cases of BTV infection have been confirmed inC. imicola in Yunnan Province of China (Duan et al., 2021), suggesting that introducing AHSV infection types into disease-free areas is possible, therefore suitable areas of the vector should be strictly monitored.

In the future scenario, the habitat suitability ofC. imicola will likely expand to higher latitudes, and the predicted map even showed that there were low suitability areas in high latitudes in Norway, Sweden, Finland, and the Kola Peninsula.Rawlings et al. (1997) also proposed that global climate change may causeC. imicola to expand northward. Moreover, it is worth noting that the Americas and Australasia have large areas of low suitability and medium suitability, and ifC. imicola were translocated to these areas, there are increasing risks forC. imicola to expand the activity range. This will lead to a wider geographical distribution of the AHSV, thereby increasing the risk of exposure to diseases.

The surveillance ofC. imicola is of great significance for the prevention and control of African horse sickness because AHSV was transmitted to domestic and wild populations throughC. imicola biting. In domestic populations, the horses of AHSV infection have obvious clinical signs, a short period of viremia, and high mortality (Robin, 2019;Wilson et al., 2009). In wild populations, zebra played a crucial role in the persistence of AHSV in Africa and was considered the natural virus reservoir (Mellor & Hamblin, 2004). Zebra rarely showed clinical signs, but the period of viremia can extend approximately 40 days (Barnard et al., 1994). In 2020, African horse sickness was reported in Southeast Asia, and vaccination was the most effective way to control African horse sickness. Before the vaccine development, the priority measures were to prevent vector-host interaction. Stabling of horses overnight can protect horses from African horse sickness becauseC. imicola andCulicoides bolitinos were less reluctant to enter enclosed space (Wilson et al., 2009). Stabling was also a practical way as part of quarantine measures related to the intercontinental transport of horses (Page et al., 2015). Therefore, multiple integrated measures (such as staling of horses, clearance of larval habitat, insecticides, etc.) may have a greater impact on AHSV transmission (Carpenter et al., 2017).

This study still has a limitation, for most ofC. imicola occurrence data were obtained from the published literature, which might suffer from underreporting. However, the results could be related to differences in reporting rather than the true ecological habits ofCulicoides. In the future, it will be interesting to model the impact of the geographical expansion ofCulicoides on disease distribution. Simultaneously, it is of great significance if the prediction model includesCulicoides survivability under high temperatures.

Conclusions

Based onC. imicola occurrence records and bioclimatic variables, the current and future suitable habitat ofC. imicola all over the world was modeled using MaxEnt model. Three bioclimatic variables were revealed to have important effects onC. imicola distribution. In the 21st century, the habitat suitability ofC. imicola may be different with climate change. The prediction of this study is of strategic significance for vector surveillance and the prevention of vector-borne diseases.

Supplemental Information