1 INTRODUCTION
Biotic invaders are a major threat, causing great destruction to the environment, economies, and biological safety, and with the development of transportation and commerce under globalization, few, if any, areas have escaped these invasions (Mack et al., 2000; Sun, Yuan, & Ou, 2002; Walker & Steffen, 1996). Pine wilt disease (PWD) is a major disease that invades forests and is highly destructive to its host pines. The disease spreads quickly, causes disease over a wide range, and is difficult to control and cure; thereby, PWD is also known as the “cancer” of pines (Naves, Camacho, Sousa, & Quartau, 2007; Wu et al., 2013). The disease, native to North America, was introduced into Japan in 1905, and then outbreaks occurred successively in China, South Korea, Mexico, Portugal, and Spain (Mota & Vieira, 2008; Zhao, Mota, Vieira, Butcher, & Sun, 2013). To date, PWD has caused serious negative effects on the economy and the environment of many countries. From 2001 to 2012, the European Union spent thirty million euros on the surveillance and control of the disease, with projected losses of up to twenty billion euros in Europe by 2030 (Soliman et al., 2012). In Japan, at least 700,000 m3 of pine trees are lost every year (Mamiya & Shoji, 2009). In China, the direct and indirect economic losses due to PWD have reached 100 billion yuan, and the area of Pinus massoniana , its main host, decreased by 8.07 million ha from 1994 to 2013, with 60 million ha of pine forests that remain greatly threatened by this disease (Ye, 2019). In addition, according to the hypothesis of an insect–fungus complex (Lu, Wingfield, Gillette, & Sun, 2011), the native ecosystems are at risk of reinvasion because pine wilt disease acquired greater adaptability in the invasion of new habitats (Maehara & Futai, 1997; Zhao et al., 2013).
The causal agent of PWD is the pinewood nematode (PWN,Bursaphelenchus xylophilus ), which is spread by vector beetles of the genus Monochamus (Mota & Vieira, 2008). The main vector beetles in North America include M. carolinensis , M. scutellatus , and M. mutator (Akbulut & Stamps, 2011). M. galloprovincialis is the only insect vector in Portugal and other European countries (Sousa et al., 2001); whereas in several Asian countries, including China, the main vector is M. alternatus . Therefore, the effective control of M. alternatus plays an important role in controlling the continued spread of PWD in Asia (Zhao et al., 2008). In addition to self-migration, human-mediated and environmental factors are also important in the spread of M. alternatus . Moreover, because of the host dominance and ecosystem dominance of forest pests (Lian & Zhang, 2005), it is necessary to better understand the influence of host and nonhost landscapes on the dispersal behavior of this species.
The landscape plays an important role in the ecological processes (Turner, 1989). Landscape diversity can have major effects on the dispersal of forest pests (Thibaud et al., 2014; Foley et al., 2005), by either promoting or inhibiting the gene flow of dispersing insects and their movement between different landscape types (Fontaine, Bergerot, Le Viol, & Julliard, 2016; Yadav, Stow, & Dudaniec, 2019). The abundance of insects generally increases with the density of host plants, and insects have a tendency to migrate to patches of host plants (Underwood, Inouye, & Hambäck, 2014). The evidence is also increasing that with high tree diversity associational resistance can form against pests and reduce their damage (Damien et al., 2016; Harri, Koricheva, & Kai, 2007; Jactel, Goulard, Menassieu, & Goujon, 2002; Jactel & Brockerhoff, 2007; Jacte et al., 2015). Especially in mixed stands with hosts and nonhosts, the nonhosts can reduce the proportion of hosts available to the pest, and the natural enemies of pests are also typically more abundant, both of which increase the resistance to forest pests (Barbosa et al., 2009; Quayle, Regniere, Cappuccino, & Dupont, 2003). In addition to the effects of tree species in the landscape, other landscape types (e.g., urban areas, farmlands, and roads) can also have different effects on the gene flow and dispersal of insects (Keller, Strien, & Holderegger, 2012; Ortego, Bonal, & Muñoz, 2010; Yadav et al., 2019). Therefore, in a heterogeneous landscape, it is particularly important to understand the relationships between landscape types and pest dispersal behavior, population genetic structure, and gene flow. With such an understanding, the mechanisms to explain pest occurrence can be revealed, and scientific and effective ecological control methods can be established.
Although the relationship between landscape and some longhorn beetles has been described, most studies used microsatellite or mitochondrial DNA markers to study the invasion history and the influence of environmental factors such as climate and altitude (David, Giffard, Piou, Roques, & Jactel, 2016; Haran, Roques, Barnard, Robinet, & Roux, 2015; Haran et al., 2017; Javal et al., 2017; Javal et al., 2019; Koutroumpa, Rougon, Bertheau, Lieutier, & Roux, 2013; Tsykun et al., 2019). However, because of the limited number of polymorphic sites and the lack of detailed information on tree species, the dispersal behavior and the genetic model of M. alternatus at a fine scale has not been studied. In particular, a comprehensive description of the relationships between this species and finely classified tree species and nonhost landscapes such as urban areas, farmlands, and roads is lacking. This information is particularly important to clarify the dispersal behavior of M. alternatus in a heterogeneous landscape, as well as to efficiently manage and control PWD. With the emergence of next-generation sequencing, millions of single nucleotide polymorphisms (SNPs) can be detected and compared with microsatellite markers. In addition, fewer samples are needed to obtain more accurate and better information on population structure. Thus, next-generation sequencing not only facilitates the development of landscape genetics but also is expected to become the standard in environmental association analysis in coming years (Rellstab et al., 2015; Jeffries et al., 2016). Currently, SNP analysis is becoming more widely accepted and applied in landscape research (Bay et al., 2018; Nancy et al., 2020; Zhen et al., 2017). Moreover, with the large numbers of SNPs, the analysis is more powerful, and the differences between populations can be detected at a fine scale (Liu et al., 2019; Rašić, Filipović, Weeks, & Hoffmann, 2014).
In this study, the whole genome of M. alternatus was resequenced at different landscape scales (intermediate and fine scales), and the tree species in the forested landscape were classified in detail based on a forest inventory of China. To analyze the influence of multiple scales, first, the population structure at intermediate and fine scales was compared, and then, the relations between gene flow and genetic diversity and forests with different tree species, urban areas, farmlands, roads, and water sources were studied at a fine scale. At the fine scale, the isolation-by-distance (IBD) and the least-cost distance (LCP) models(Adriaensen et al., 2003) were used, and the distances calculated by the models were correlated with genetic distance. To define the resistance surface more objectively and further study the effects of different landscape types on gene flow and dispersal ofM. alternatus , the least-cost transect analysis (LCTA) developed by Strien, Keller, & Holderegger. (2012) and maximum-likelihood population effects (MLPE) developed by Clarke, Rothery, & Raybould. (2002) were adopted. The distance-based redundancy analysis (db-RDA) (Legendre & Anderson, 1999) and the generalized additive model (GAM) (Lengyel & Podani, 2015) were also used to test the influence of the landscape composition around each sample point on genetic diversity.