A noteworthy step on a vast continent: new expansion records of the guava fruit fly, Bactrocera correcta (Bezzi, 1916) (Diptera: Tephritidae), in mainland China

Bactrocera correcta (Bezzi, 1916), commonly known as the guava fruit fly, is of concern as an invasive pest in tropical and subtropical countries. It was first recorded in China in 1982 in Yuanjiang, in southern Yunnan Province. We monitored the spread of B. correcta in the field during 2017 and 2018, and found that it had moved about 300 kilometers eastward from its known range in 2011 in Yunnan and has now entered the neighboring Guangxi Province. A species distribution model used to predict the potential distribution of B. correcta in China revealed that southern China provides the most suitable habitat for B. correcta, and that further eastward expansion threatens large areas of southeastern China. Results from this study suggest that the rapid eastward migration of B. correcta is related to improved highway and high-speed railway systems in China that transport fruits and vegetables to eastern markets. This paper also discusses methods to slow the expansion of B. correcta in China.


Introduction
The guava fruit fly, Bactrocera correcta (Bezzi, 1916) (Diptera: Tephritidae), is an important insect pest that infests more than 70 species of tropical and subtropical fruits and melons, representing 35 plant families, such as guava, mango, cashew, cherry, jujube, orange, banana, carambola, and wax apple (Allwood et al. 1999;Maynard et al. 2004;USDA 2014). In Thailand, B. correcta along with Bactrocera dorsalis (Hendel, 1912) are considered the main fruit pests, seriously affecting production of mango and guava (Kunprom et al. 2015). In India, damage rates to tropical fruits caused by B. correcta can reach 80%, ranking it as the country's most serious fruit fly pest (Jalaluddin et al. 1999).
Bactrocera correcta is currently a serious pest in tropical and subtropical areas of Asia, being first recorded in Bihar, India in 1916 (Bezzi 1916), followed by Nepal, Pakistan, Sri Lanka and Thailand (White and Elson-Harris 1992). It has since expanded to Bangladesh, Bhutan, China, Laos, Malaysia, Myanmar and Vietnam (Wang 1996;Drew and Raghu 2002;Liu et al. 2013;Qin et al. 2015). Possibly through trade of infested fruit, small numbers of B. correcta have been detected multiple times in California and Florida, USA, which were followed by active eradication programs (USDA 2014). Given the above, B. correcta is considered highly invasive, and is therefore targeted as a quarantine pest in many countries (White and Elson-Harris 1992;Qin et al. 2016).
Bactrocera correcta was first detected in Yunnan, China, in 1982 (Wang andZhao 1989). In the following decades, it gradually expanded inland, but was confined to various mountain valleys in southwestern Yunnan, displaying a patchy geographical distribution in Yunnan (Liu et al. 2013). Southern China, including Fujian, Guangdong, Guangxi, Hainan, and Taiwan, are considered the most suitable areas for B. correcta in China (Qin et al. 2015). In the past decade, the transportation systems connecting Yunnan and Guangxi in western China to Guangdong in eastern China have been dramatically improved (Liu et al. 2014), which have increased the frequency and volume of goods being transported eastward within China as well as the potential spread of pests. If B. correcta were to expand throughout southern China it would cause severe negative impacts on fruit and vegetable production in this region (Lu et al. 2010(Lu et al. , 2013Qin et al. 2015). Thus, understanding B. correcta dispersal and spread is a core component for developing prevention and control programs for this fruit fly in China.
During 2017 and 2018, we surveyed for B. correcta in areas around its known range in southwestern China to determine if there had been any spread eastward within China. Furthermore, we used the species distribution model to analyze future spread potential of B. correcta in China and thereby provide updated information for setting internal quarantines and management guidelines.

Study area and field monitoring
Within China, B. correcta was only known to occur in Yunnan prior to this study. However, since 2006, the distribution of B. correcta in China has expanded eastward to the Nanpanjiang River basin that borders Yunnan and the neighboring provinces of Guangxi to the south and Guizhou to the north (Liu et al. 2013). Therefore, our investigation was carried out in the three provinces of Yunnan, Guangxi, and Guizhou.
Field monitoring was conducted from March 2017 to July 2018 for a total of 17 months. At times, we expanded the monitoring area eastward in response to positive detections of B. correcta. We first established monitoring sites in Funing, Guangnan, and Qiubei in Yunnan Province, Baise in Guangxi Province, and Xingyi in Guizhou Province. Later in 2017, based on initial results of the investigation, we expanded the sampling area to include five additional counties in Guangxi Province, including Bama, Chongzuo, Longan, Pingguo, and Tianyang (Supplementary material Table S1). The sampling region covered a total area of nearly 90,000 square kilometers. All of these areas have been affected by other invasive fruit flies in recent years, with B. dorsalis being the dominant fruit fly species present; however, B. correcta was never detected in any earlier monitoring efforts.
At each survey site, one fruit orchard was selected, and three traps were deployed. We used Steiner traps baited with methyl eugenol to attract male fruit flies as described in Chen et al. (2010). The traps were suspended from branches in the fruit trees at a height of 1.5-2 m above the ground. All fruit fly specimens were collected at 15-day intervals, returned to the laboratory in Yunnan University, and identified to species using the morphological characters proposed by Drew and Romig (2013). All sampling sites were georeferenced by GPS (MG768W, UniStrong, Beijing, China) and the corresponding distribution map was drawn using GRASS GIS 7.2.2 (GRASS Development Team 2017).

Species distribution modeling
We used species distribution modeling to better understand the distribution pattern of B. correcta in China and to provide a theoretical basis when developing pest management strategies for this fruit fly. The occurrence data of B. correcta in Asia were obtained from the GBIF database (http://www.gbif.org/), genetic sample information (Kunprom et al. 2015;Qin et al. 2016), and field survey results (Satarkar et al. 2009;Liu et al. 2013) (Table S2).
Environmental data were obtained from the Bioclim Database (WorldClim Version2, http://www.worldclim.org/). We used seven bioclimatic variables (2.5' spatial resolution) to develop a B. correcta distribution model, which included Annual Mean Temperature (bio1), Temperature Seasonality (bio4), Max Temperature of Warmest Month (bio5), Min Temperature of Coldest Month (bio6), Temperature Annual Range (bio7), Mean Temperature of Warmest Quarter (bio10), and Mean Temperature of Coldest Quarter (bio11). All of the above selected variables have been regarded as important environmental factors affecting the distribution of B. correcta (Liu et al. 2013).
A Maxent model was used to predict the potential distribution of B. correcta in China because of its excellent performance in ecological niche modeling (Phillips et al. 2006). We randomly sampled 10,000 points in buffer areas around occurrences as the background samples. The buffer size was 0.5 degrees. When selecting the model parameters, we used models with regularization multiplier (RM) values ranging from 0.5 to 4.5 (with increments of 0.5) with six feature class (FC) combinations (L, LQ, H, LQH, LQHP, with L = linear, Q = quadratic, H = hinge, and P = product). According to the lowest delta AICc score, the FC = LQH and RM = 3 were used in the final modeling effort ( Figure S1). All analyses were conducted using the Wallace Package in R (R Core Team 2018) (Warren and Seifert 2011;Kass et al. 2018).

Results
We detected B. correcta in a mango orchard in March 2017 in Baise, Guangxi, which was the first time that this fruit fly was detected outside of Yunnan in China (Figure 1). The sampling site in Baise is about 300 km further east from where it had last been recorded in Yunnan in 2011 (Liu et al. 2013). As a result of the positive detection in Baise, we extended the monitoring eastward to Bama, Pingguo, and Tianyang in Guangxi Province in April 2017. Of these three new districts, we discovered B. correcta in Tianyang, which is located 38 km east of Baise.
In August 2017, we added two more monitoring sites to the south and east of Tianyang, one in Longan County about 100 km southeast of Tianyang, and the other in Chongzuo County about 140 km south of Tianyang. There were no B. correcta adults collected at these two sites during the subsequent monitoring period.
These data revealed that the eastern boundary of B. correcta range in China extended at least to Baise and Tianyang. However, neither of these two locations were contiguous with the known range of B. correcta in Yunnan. The species distribution model predicted highly suitable areas for establishment of B. correcta in the tropical and subtropical regions of China south of 25°N Latitude (Figure 3). The area suitable for B. correcta in Yunnan comprises about 20% of the total land area of Yunnan, resulting in a patchy distribution due to the many mountain chains in Yunnan. By contrast, the area suitable for B. correcta in southeast China is generally contiguous and extensive, indicating that southeastern China would be very suitable for invasion by B. correcta. The current distribution of B. correcta in southwestern China (Figure 3) suggests that the mountains of Yunnan have largely hindered natural spread from west to east, whereas the current disjunct populations found in Baise and Tianyang likely resulted from human-assisted movement and thus pose a higher risk for invasion of B. correcta to the east.

Discussion
Yunnan Province, in southwest China, borders Laos, Myanmar, and Vietnam and is believed to be the main entry site for invasive organisms entering China from the Indochina Peninsula, as well as an important transit  Wan et al. 2017). As for the exotic species of Bactrocera in China, many species, such as B. dorsalis and B. thailandica Hancock, 1994, entered China through Yunnan (Liu et al. 2009;Shi et al. 2012). This pattern of invasion has led to Yunnan becoming the richest area for fruit fly biodiversity in China. For example, Yunnan has 67% of the total number of Dacini fruit fly species found in all of China and of the 83 Dacini species known to occur in Yunnan, more than 50 were introduced (= 60%) (Wang 1996;Doorenweerd et al. 2018).
Although many exotics have entered Yunnan, the northern plateau and the complex series of longitudinal mountain ridges and valleys in southern Yunnan form a natural barrier for the eastward spread of many alien organisms into the interior of China (He et al. 2005;Gan and He 2006). Of all the Dacini fruit flies that have been reported in China, about one third are still confined to Yunnan, suggesting that the unique geography of Yunnan plays an important role in slowing natural spread both northward and eastward (Wang 1996;Chen et al. 2010).
As stated above, B. correcta was first reported in Yunnan in 1982 in the Yuanjiang River valley and was generally restricted to that region for more than 30 years (Wang and Zhao 1989;Liu et al. 2013). Several laboratory and field studies have confirmed that B. correcta is poorly adapted to low temperatures and long-range dispersal (Liu and Ye 2009;Liu et al. 2015), and therefore the Yunnan-Guizhou Plateau and Hengduan Mountains in northern Yunnan have effectively prevented natural spread of B. correcta northward toward Sichuan (Liu et al. 2013;Qin et al. 2016). By contrast, the geographic barriers that restrict natural eastward spread of B. correcta from Yunnan are relatively narrow mountain chains in Yunnan and therefore have been the main focus for pest surveys by regional plant health specialists.
In the present study, our field surveys revealed for the first time that B. correcta had expanded beyond Yunnan and entered Guangxi to the east. The newly invaded sites in Guangxi were more than 300 kilometers further east than the last historical distribution record in Yunnan made in 2011 (Liu et al. 2013). This discovery indicates that the geographic barriers in Yunnan to eastward spread of B. correcta have been breached, and now many agricultural crops in southeastern China are threatened.
What was responsible for the eastward jump of B. correcta to Guangxi? It was likely not natural spread given that there is no valley connection between the newly invaded sites in Guangxi and the historical known range in Yunnan, and moreover no B. correcta adults were detected in any of the four monitoring sites between the historical range in Yunnan and the newly invaded sites. In addition, although some studies have shown that B. correcta can adapt quickly to the new environments (Liu 2010;Liu et al. 2013), there is no evidence that B. correcta has adapted sufficiently to disperse naturally eastward over the high-altitude mountains in Yunnan.
The movement of people and goods is reasonably regarded as one of the main factors in the spread of exotic agricultural and forest organisms both within and between countries (Haack 2006;Colunga-Garcia et al. 2009Mazzi and Dorn 2012). Within a country, the rate at which invasive organisms can spread is often enhanced by transportation improvements (Ding et al. 2008;Hu et al. 2013), such as the new modern highway and high-speed railway links that were opened between Yunnan and Guangxi in 2012 and 2016, respectively, both of which run through Baise and Tianyang. These transportation improvements have shortened the travel time between Yunnan's historical B. correcta eastern outbreak areas and Guangxi's newly infested sites from 15 hours to 6 hours by highway and from 10 hours to 3 hours by high-speed rail. Similarly, the railway freight capacity between Yunnan and Guangxi has increased from 27 million tons to 110 million tons per year after completion of the high-speed rail system (Liu et al. 2014;Zhu 2015). These increases in the rate and volume of goods transported along these improved highways and high-speed railways could easily have served as the pathway for eastward movement of B. correcta into Guangxi.
Bactrocera dorsalis is one of the most damaging fruit flies worldwide and is similar to B. correcta in many life-history attributes, including sharing many of the same host plants (Gu et al. 2019). Evidence based on population genetics suggests that Yunnan was the main channel by which B. dorsalis entered China from more southern countries and expanded to the north and east within China, although this inference cannot be conclusively confirmed due to the lack of historical expansion records (Shi et al. 2012;Zeng et al. 2019). The invasion pattern so far demonstrated by B. correcta in China is very similar to that of B. dorsalis in that B. correcta entered China through Yunnan, and is now spreading eastward within China.
Given the potential for pests to move more readily within China as transportation networks improve indicates a need for China to adjust its monitoring efforts and internal quarantine regulations against pests like B. correcta. For example, a network of monitoring sites, using traps baited with attractant lures, could be established throughout southern China in major areas of fruit production and along major transportation routes. Second, given the apparent restricted range of B. correcta in Guangxi, it may be possible to attempt an eradication program to prevent further spread of this pest. And third, additional training on B. correcta identification and management could be provided to growers and inspectors throughout the known range of B. correcta and in neighboring provinces, with a special focus on vital transportation hubs and major fruit distribution centers. Such actions could enhance early detection of B. correcta populations, which would allow for rapid control efforts, and thereby slow the eastward invasion of B. correcta to southeastern China.