One maternal lineage leads the expansion of Thaumastocoris peregrinus (Hemiptera: Thaumastocoridae) in the New and Old Worlds

The bronze bug, Thaumastocoris peregrinus, an Australian native insect, has become a nearly worldwide invasive pest in the last 16 years and has been causing significant damage to eucalypts (Myrtaceae), including Eucalyptus spp. and Corymbia spp. Its rapid expansion leads to new questions about pathways and routes that T. peregrinus used to invade other continents and countries. We used mtDNA to characterize specimens of T. peregrinus collected from 10 countries where this species has become established, including six recently invaded countries: Chile, Israel, Mexico, Paraguay, Portugal, and the United States of America. We then combined our mtDNA data with previous data available from South Africa, Australia, and Europe to construct a world mtDNA network of haplotypes. Haplotype A was the most common present in all specimens of sites sampled in the New World, Europe, and Israel, however from Australia second more frequently. Haplotype D was the most common one from native populations in Australia. Haplotype A differs from the two major haplotypes found in South Africa (D and G), confirming that at least two independent invasions occurred, one from Australia to South Africa, and the other one from Australia to South America (A). In conclusion, Haplotype A has an invasion success over many countries in the World. Additionally, analyzing data from our work and previous reports, it is possible to suggest some invasive routes of T. peregrinus to predict such events and support preventive control measures.


North America
Mexico 2015 76 United States of America 2016 77,78  DNA extraction, PCR amplification, and COI-gene sequencing. The identification of all T. peregrinus specimens was confirmed based on morphological characters 4 . DNA extraction was performed individually for each specimen using the DNeasy Blood and Tissue Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. Depending on the availability of insects and the proximity of sampling sites, we used 1 to 10 insects per location for DNA extraction ( Table 2). Each adult insect was removed from the vial with ethanol and   left to air dry on a paper towel. The entire body was individually macerated in a 1.5 mL tube containing 180 µL of buffer ATL and 20 µL of proteinase K and incubated at 56 °C for 12 hours. Subsequently, genomic DNA was purified in a silica-based matrix and eluted in 35 µL of buffer AE. The concentration of DNA was assessed in a spectrophotometer (NanoDrop ™ 1000, Thermo Scientific, Wilmington, DE, USA).
PCR amplification consisted of an initial denaturation step at 98 °C for 30 s, followed by 30 cycles at 95 °C for 30 s, 48 °C for 30 s and 72 °C for 1.5 min, and a final extension at 72 °C for 10 min 16 . Amplified products were resolved on 1.0% agarose electrophoresis gel, pre-stained with Nancy-520 DNA gel stain (Sigma-Aldrich) and visualized using a gel documentation system. Successfully amplified PCR products were sequenced by ACTGene Molecular Analyses (Alvorada, RS, BR), using the BigDye Terminator method on an ABI 3500 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). Data analysis. Quality assessment, trimming, editing, and analysis of each DNA sequence were performed using the software Pregap and Gap4 within the Staden package 19 . CLC Sequence Viewer (Version 7.8.1-QIAGEN Aarhus A/S) was used to retrieve and align sequences with a length of 468 bp. A pBLAST analysis (amino acid homology confirmation) with T. peregrinus partial mtDNA COI genes deposited on the NCBI (National Center for Biotechnology Information, USA) database was also performed.
The 116 COI sequences generated in this study were combined with 197 sequences of 16 populations from Australia 16,20 . We also used 32 sequences from South America, 62 from South Africa 16 , 1 from Italy (KF437485) 21 , 10 from Spain (MN401749) 22 , and 5 from Reunion Island (France) (GenBank − KT273623, KT273624, KT273625, KT273626, and KT273627) 23 . The inference and visualization of genetic relationships among intraspecific sequences used to generate a haplotype network were conducted using TCS network 24 within the program PopART 25 .

Results
Fragments of mtDNA from 116 individuals of T. peregrinus from 28 populations were amplified and sequenced, resulting in a trimmed sequence of 468 pb of the COI gene. Samples of T. peregrinus from six countries (Chile, Paraguay, Mexico, Portugal, Israel, and the USA) were sequenced for the first time. For all the sampled sites from South America (Chile, Paraguay, Argentina, Brazil, and Uruguay), North America (Mexico and the USA) and the Mediterranean basin (Italy, Portugal, and Israel), we confirmed the presence of a unique haplotype (haplotype A).
Combining our data with sequences not deposited in GenBank 20 and with others previously deposited in this database (Table 2), we obtained a total of 423 T. peregrinus COI sequences and 45 unique haplotypes were identified. The proportions of the different haplotypes among the countries are presented on a haplotype network (Fig. 2). This network reveals that the greatest diversity of haplotypes is from the native region of T. peregrinus in Australia. Haplotype D is the ancestral one and all the other haplotypes radiate from it (Fig. 2). Haplotype D was the most frequent haplotype in Australia and found in 10 sites 20 . Haplotype A was present in three of 10 sites where haplotype D was also reported: Sydney, Coonabarabran, and Cootamundra 20 . The difference between the two haplotypes is only one base pair in the partial mtDNA COI fragment (468 bp) 16 and possibly represents a silent mutation.
Outside of Australia, only four haplotypes were found: D and G in South Africa 16 , IT in Italy 21 , and A in South America 16 , North America, Europe, and Asia. From our work, haplotype A was the most frequent and widespread, found at 28 sites (116 sequences), in three populations from Australia (48 sequences) 20 , six populations in South America (32 sequences) 16 and one population in Spain (10 sequences) 22 , comprising a total of 206 sequences (Fig. 2). Based on the combined sequences, haplotype D was found in Australia (67 sequences) 16,20 , Reunion Island (five sequences) 23 , and in nearly all the sites sampled in South Africa (46 sequences) 16 , making a total of 118 sequences. The third and fourth most frequent haplotypes were G and H, 31 sequences (16 for South Africa and 15 for Australia) of G and 14 sequences (all in Australia) of H, totaling 45 sequences 16,20 . All the remaining haplotypes were represented by three or fewer individuals. We note that all haplotypes found in the invaded regions were also found in their naturally occurring regions in Australia, except for the IT haplotype, which was found only in Italy (Genbank KF437485) 21 (Fig. 2). To understand and to have a wide view of these results properly, the Fig. 3 shows the haplotype A distribution at our sampling sites in Americas, Europe and Israel (10 countries). In addition, the Fig. 4 represent a world haplotype distribution from 14 countries and 423 sequences.

Discussion
Genetic diversity. Our study updates information on the genetic diversity and mtDNA haplotype distribution of T. peregrinus worldwide. Samples from several locations in seven countries, including newly invaded countries in South America (Chile and Paraguay), North America (the USA and Mexico) and the Mediterranean Basin (Portugal, Italy, and Israel) were collected and analyzed. We also combined our mtDNA data with previously available sequences of T. peregrinus 16,[20][21][22][23] (Table 1). Our results demonstrate that haplotype A is predominant in South America, North America, Europe, and Israel, indicating a low genetic diversity of T. peregrinus outside its native region.
In contrast with the low genetic diversity found in the invaded areas, the Australian populations of T. peregrinus (the native range of the species) revealed a high mtDNA COI haplotype diversity 16,20 . Genetic diversity is expected to be higher in a species native range in comparison with invasive populations 26 , possibly because of bottleneck effects during the invasion process. The newly introduced species might be reduced to a few individuals because of environmental events (e.g. hurricane, forest fires) and anthropic actions, resulting in low genetic diversity 27 . Furthermore, genetic drift and the founder effect might also account for the low genetic diversity of the new invasions 28 . Selection pressure on a given population of the invasive species, that allow the survival of the fittest insects, might result in low genetic diversity during the first decades of species establishment 29 . These evolutionary forces associated with the adaptability of haplotype A to different environments seem to have driven the expansion of T. peregrinus worldwide, compared to the haplotypes D and G.
Low genetic diversity has also been reported for many other forest invasive pests. For example, populations of Anoplophora glabripennis (Motschulsky) (Coleoptera: Cerambycidae) 30 35 demonstrated low genetic diversity in areas where these species are considered invasive. In a new environment, invasive populations can take several years to increase their genetic diversity 36,37 .

Possible invasion routes of T. peregrinus in South America, Europe, North America, and Asia.
Knowledge of possible ways by which an insect can invade a new area and about its biology is essential to identify routes of invasion and to develop and adopt effective strategies to prevent and/or manage additional introductions 1,38 . Possible routes of invasion of eucalypts pests include active dispersion, passive transport by biotic and abiotic agents, or accidental transport 39 . Even though our data are limited to a partial mtDNA COI fragment (468 bp) from T. peregrinus, an invasive range can be projected using mtDNA which possesses a unique haplotype signature of a population as stated by Goldstein et al. 40 . The mtDNA is inherited most often only from the mother and does not undergo recombination 41 , enabling to assess the number of genetically unique founder females [42][43][44][45] . Studies of invasive species used similar methodologies, for example of T. peregrinus 16 , wood-boring beetles 46 , Melanagromyza sojae (Zehntner, 1900) (Diptera: Agromyzidae) 47 and Helicoverpa armigera (Hübner, 1808) (Lepidoptera: Noctuidae) 48 .
Using data collected in our work and from the literature, we suggest nine possible invasion routes of T. peregrinus (Fig. 5). These conclusions are based on (1) spatio-temporal events; (2) previous report of invasion routes of each country; (3) movement of people, goods, and timber through motorways and waterways between countries; and (4) our findings that analyzed 116 sequences from specimens collected in countries in South and North Americas, Europe, and Asia.
Routes 1 and 2: In 2003, T. peregrinus 4 was firstly reported in South Africa, erroneously identified as Thaumastocoris australicus Kirkaldy 9 . Years later, a study using DNA barcoding in specimens of T. peregrinus from South Africa revealed the presence of two haplotypes, D (route 1) and G (route 2), indicating that the www.nature.com/scientificreports www.nature.com/scientificreports/  www.nature.com/scientificreports www.nature.com/scientificreports/ introduction happened in two distinct invasion events, both aided by transport of people and goods primarily through airline with flights from Australia (Sydney) 16 . Further studies should update the T. peregrinus haplotypes present in Africa because its occurrence has increased throughout the continent, possibly as a consequence of new invasions or further dispersal of populations already present in the invaded area.
Route 3: T. peregrinus was first reported in the Americas in Argentina in 2005 4 . This invasion was an independent event of haplotype A directly from Australia and possibly from Sydney since the haplotype was different from that of the South African invasion 16 . Geographical proximity, land border, surface transport of people and goods, and the presence of extensive eucalypt plantations alongside highways margins in Argentina, Brazil, Paraguay, and Uruguay facilitated the spread of this invasive pest through South America. The monitoring data of T. peregrinus in Brazil suggest its entrance via airport 8 . In Chile, the invasion of T. peregrinus was thought to be passive and was spread by different means, including trucks, buses and ships, and other vectors, mainly birds 49,50 . Natural dispersion by flight is also possible due to Argentina's proximity to other countries.
The geographical position of a country can influence and help to predict the number of invasive insects, which is related to the length and nature of borders with neighboring countries and correlations among land area, population, and gross domestic product 1,51 . However, some factors may be more likely than others, as is the case of eucalypt pest introductions into New Zealand from Australia, probably due to the proximity and frequent trade between the two countries 1,52 .
Route 4: In Europe, the first reported occurrence of T. peregrinus was in Italy in 2011, possibly introduced through the transport of timber from South America and South Africa 53 . However, this hypothesis was not based on the population genetic signatures. In 2014, mtDNA molecular characterization of T. peregrinus specimens from Italy was performed, and a haplotype called IT was identified 21 . Strangely, haplotype IT has not been reported from Australia and several factors might offer an explanation, including a not yet detected haplotype with unusual low frequency in Australia, or a new haplotype restricted to Italy or from an undetermined population along the invasive routes between Australia and Italy. Additionally, our results confirm the presence of haplotype A in Italy, suggesting an invasive pathway from South America because of wood transport between the two countries. This scenario is more likely than an introduction from South Africa as previously reported 53 . Nonetheless, based on the presence of haplotype A in Portugal and recently in Spain 22 , we also hypothesize dispersion and incursions from many of these countries into Italy because of the short geographical distance, the land border, and the motorway connections between countries (Portugal, Spain, France, and Italy), and intense movement of tourists and goods.
Route 5: The second country in Europe to report an invasion of T. peregrinus was Portugal in 2012 54 . Based on South America's status as the main supplier of eucalypts wood to Europe 55,56 , South America has been suggested as a route of invasion of T. peregrinus into Europe 54 . Also, the genetic signature of the sampled population in Portugal revealed that it is most likely an arrival from South America. The latest European countries to report T. peregrinus were Spain (2014) 57 , Albania (2016) 58 , and Greece (2017) 59 . The genetic diversity in these countries remains unknown, except for Spain, where haplotype A was reported in Madrid, with two main invasion routes being suggested: either an invasion from South America to Portugal and then an expansion to Spain; or that T. peregrinus invaded Spain from South America 22 .
For other European countries, we suggest that the bronze bug spread within the Mediterranean countries similarly to other invasive species that feed on eucalyptus trees. Possible invasion routes of G. brimblecombei into www.nature.com/scientificreports www.nature.com/scientificreports/ Europe were suggested from South or North America, considering the Spatio-temporal events and the presence of this insect in the New World since the 20 th century 60 . Subsequent events demonstrated the spread of G. brimblecombei to Mediterranean countries 61 . The connections between countries, mainly motorways and waterways linking Portugal, Spain, France, Italy, Greece, Algeria, and Morocco, were suggested as the main invasion routes of this pest 62 .
Route 6: The invasion of T. peregrinus to Reunion Island (2013) 23 may have resulted from the flow of people and goods and the island's proximity to South Africa, where haplotype D is predominant 16 . However, a direct introduction from Australia is also possible based on distance and the presence of haplotype D in the site of origin of this species.
Route 7: The invasion of T. peregrinus into Israel possibly occurred from a country with the bordering to the Mediterranean Sea, where this species was already present, based on connections of motorways and waterways and the intense movement of people and goods between countries 60,62 . The recent expansion of eucalypts forestry in Asia could also influence the occurrence and detection of pests 1 , including the spread of T. peregrinus soon into new countries in Asia.
Routes 8 and 9: In North America, the first report of T. peregrinus was in Mexico in 2015 63 , ten years after its detection in Argentina. In 2016, this pest was found in the USA, in California in the greater Los Angeles area, specifically north Hollywood in the San Fernando Valley 64,65 , and Heartwell Park in Long Beach 65 . The first detection event might not be the actual first incursion and the establishment of a pest. Therefore, because of the short distance and the short period of this pest's detection between Mexico and California (USA), it is impossible to know definitely in which country of North America T. peregrinus first entered. However, the presence of the same haplotype A in all areas of the Americas, suggests that a single invasion starting from South America and spreading northward to California might have occurred. Nevertheless, as before, Australia should also be considered as a source of T. peregrinus in North America because of the presence of haplotype A.
In general, in the USA, two invasion routes are considered dominant in the introduction of insects and diseases. Most harmful, non-native species (62%) likely entered North America with live plants and 30% probably arrived with wood packaging material (WPM) or other wood products 66 . In 2017, several specimens of T. peregrinus were found on Eucalyptus twigs in a shipment of cut flowers from Mexico that was intercepted at the port of Brownsville (Texas). Because T. peregrinus had quarantine status, the cut flowers returned to Mexico as a safety measure 67 . The twin ports of Los Angeles and Long Beach are the busiest in the USA and the international airports of San Francisco and Los Angeles are main entry points for passengers and air cargo 68 ; all serve as gateways for pest introductions.
Our mtDNA analysis, including temporal and geographical sequence data, provides information that suggested scenarios of invasion routes of T. peregrinus in South America, North America, Europe, and Asia. These scenarios are an important source of information that assists in the planning and management of biosafety measures to prevent pests like T. peregrinus, from invading and colonizing new sites. Further studies using nuclear markers (e.g. microsatellites) or other mtDNA regions can aid to postulate new scenarios.

Factors involved in the successful invasion and establishment of T. peregrinus. Several insect
pests of eucalypts have undergone dispersal to two or more continents, for example, L. invasa, Gonipterus spp., and Glycaspis brimblecombei (Moore, 1964) (Hemiptera: Aphalaridae). Evidence exists that these invasive species possess and exploit mechanisms to aid in their dispersal 1 . In the specific case of T. peregrinus, adults do not show a strong natural ability to disperse except when aided by human activities 20 . However, nymphs are particularly mobile 8 and could be more easily dispersed. Besides, biological factors account for the success of many invasive insects into a new location 69 , and we believe that certain biological characteristics of T. peregrinus represent an important portion of its invasion success throughout more than 20 countries. The life cycle of T. peregrinus is about 60 days and has a rather high potential for breeding (fecundity of 60 eggs/female) conditioning to reproduce more than twice a year and increase its population density 50,70 . Once this pest invades a new location, the lack of natural enemies facilitates its establishment and dispersion 71 into the wide range of host trees of Eucalyptus and Corymbia (over 52 host species) found in many countries in the world 72 . T. peregrinus also thrives under a wide range of temperatures, varying from 4 °C to 34 °C, which explains its wide distribution and possible invasion into new areas in the future 73 . Its invasion success is also related to the small body size (± 3-4 mm) 9 that facilitates its dispersal over long distances, whether actively or passively by the wind (draft) 74 or by the movement of people or any kind of goods, especially timber, where is very difficult to be detected due to its small size.
Successful invaders are those featuring a small body size, several generations per year, long flight displacement and low incidence of diapause 69 . All these traits account for the rapid adaptation to new environmental conditions 75 . For example, the successful invasion of L. invasa into new areas was suggested to be a consequence of this pest's high dispersal capabilities 32 , strong resistance to low temperatures, the presence of eucalypt plantations worldwide 76 , and a lack of natural enemies in the invaded areas 77 . Moreover, synergic interactions among invasive species have been reported as a "meltdown" process that accelerates the impact of these species in the invaded area 78 . In line with this theory, both the eucalyptus lerp psyllid G. brimblecombei and T. peregrinus display a positive interaction that benefits from the preference of bronze bug females to oviposit on leaves bearing lerps 79 .
Considering the invasion success of haplotype A worldwide reported in this work and that the most common maternal lineages of T. peregrinus sampled in Australia, specifically in Sydney from 2001 to 2009, were haplotypes A, D, G and H 16,20 , the most likely source of T. peregrinus seems to be Sydney 16,20 or Chelmsford 20 for haplotypes A and D. The most commons haplotypes in a given population are more likely to be sampled and eventually to spread 16,43,80,81 , as it seems to have happened to haplotype A of T. peregrinus. On the other hand, haplotypes D, G and H were found restricted to some areas or not found outside its place of origin, although individuals might have dispersed to new areas but did not adapted and successfully establish as haplotype A.