Initial invasion of glyphosate-resistant Amaranthus palmeri around grain-import ports in Japan

The dispersal of alien species is tightly coupled to human activities such as trade and transport. Trade is known to spread troublesome weeds from countries exporting, to those importing, grain. Glyphosate resistant (GR) Amaranthus palmeri is one of the most problematic weeds in the US, which is the largest grain exporter to Japan. We demonstrate that GR A. palmeri has become established in a Japanese port in less than 10 years from the first report of GR A.


| INTRODUC TI ON
The land for global production of genetically modified (GM) crops increased from 1.7 million hectares to over 191.7 million hectares between 1996 and 2018. The US continues to be the largest producer of GM crops in the world (ISAAA, 2018). More than 80% of the area planted to GM crops has been planted with herbicide-resistant crops including stacked herbicide/insect-resistant crops (ISAAA, 2018).
Among the herbicide-resistant crops, the glyphosate-resistant (GR) crops have been the most widely cultivated for the last 20 years (Bonny, 2016;Duke & Powles, 2008). The intensive use of glyphosate has resulted in the evolution of resistance to this herbicide in several problematic weeds (Heap & Duke, 2018). To date, resistance to glyphosate has been documented in 48 species (Heap, 2020).
Currently, one of the most problematic weeds is Amaranthus palmeri (Amaranthaceae), which has become a major GR weed of the US (Webster & Nichols, 2012). After the first report in 2004 in Georgia, US (Culpepper et al., 2006), GR A. palmeri was discovered in 29 states by 2018 (Heap, 2020). Despite the species never appearing as a problem until the early 1990s, A. palmeri is now ranked as the most troublesome weed among broadleaf crops in the US (Van Wychen, 2016) and the third most troublesome weed among graminaceous crops (Van Wychen, 2017). In particular, A. palmeri is listed among the most troublesome weeds of corn, sorghum, soybean, cotton, and peanuts. For corn, soybean, and cotton, herbicide-tolerant varieties account for more than 90% of the crops in the US (USDA, 2019). Exports from the US are the largest source of these crops to Japan. In 2017, Japan imported approximately 15.3, 3.2, and 0.1 million tonnes of corn, soybean, and cottonseed, respectively (Ministry of Finance, 2017), of which 78, 73, and 52%, respectively, was from the US (Ministry of Finance, 2017).
Amaranthus palmeri is an annual dioecious forb native to the area encompassing north-western Mexico and the south-western US (Ward, Webster, & Steckel, 2013). In the absence of competition, seed production for A. palmeri is over 600,000 seeds per female plant (Keeley, Carter, & Thullen, 1987). This species was reported in Japan for the first time in 1936 (Osada, 1972) and has spread throughout Japan except Hokkaido Prefecture after the 1960s according to records of herbarium specimens (GBIF; www.gbif.org). For example, the first records of the species are dated 1964 in Fukuoka Prefecture, southern Japan (Osada, 1972), and 1968 in Ibaraki and Chiba Prefectures, east-central Japan (Suzuki, 1981;Flora-Kanagawa Association, 2018). In Japan, the species has been relatively rare until now and there has been no report of any herbicide resistance or agricultural damage. However, because the weed has become prevalent in US croplands in recent years, GR A. palmeri introduction into Japan as a contaminant of imported GM commodities is inevitable.
Internationally traded grain commodities are recognized as a pathway for the introduction of weed seeds into new areas (Lehan, Murphy, Thorburn, & Bradley, 2013) because similarities in shape and size to crop seeds hinder removal of contaminant weed seeds (Michael, Owen, & Powles, 2010). Indeed, the seeds of major weeds in grain-exporting countries are found as contaminants in imported grain commodities (Asai, Kurokawa, Shimizu, & Enomoto, 2007;Norsworthy, Smith, Steckel, & Koger, 2009;Shimono & Konuma, 2008;Wilson, Castro, Thurston, & Sissons, 2016). The contaminants sometimes include herbicide-resistant seeds, which can later spill during the transport of grain commodities and become naturalized in importing countries (Shimono, Shimono, Oguma, Konuma, & Tominaga, 2015). In Japan, GM oilseed rape and GM soybean with glyphosate transgenes have been found at several major ports (Aono et al., 2006;MAFF, 2018;Saji et al., 2005). Because GM plants have not been commercially cultivated in Japan, their feral occurrence is evidence for spillage during transport of grain commodities (Saji et al., 2005). The recent appearance of GR A. palmeri in Brazil suggests the potential for intercontinental spread of GR individuals, mediated by seed transfer (Küpper et al., 2017).
In Japan, even though A. palmeri has been rare until now, GR A. palmeri has the potential to become a troublesome weed because glyphosate is one of the most widely used herbicides. Another concern is interspecific hybridization within the Amaranthus genus. In fact, the GR gene was transferred from A. palmeri to A. spinousus by pollen flow in the US (Gaines et al., 2012;Nandula et al., 2014).
In Japan, plants of the Amaranthus genus (A. spinousus, A. patulus, A. retroflexus etc.) have been common problematic weeds of roadsides, cultivated ground, docks, and riverbanks (Ecological Society of Japan, 2002). These ruderal populations in roadside habitats may serve as conduits for further inter-population spread of the GR gene.
The copy-number amplification of the herbicide target site gene 5-enolpyruvylshikimate-3-phosphate synthase (EPSPS) is the primary mechanism that confers glyphosate resistance to A. palmeri: the genomes of resistant plants contain from 5-to more than 100fold more copies of the EPSPS gene than the genomes of susceptible plants (Gaines et al., 2010). The amplified region comprises not only the EPSPS locus but also genomic sequences corresponding to 71 putative genes, tandem repeats, and regulatory elements (Molin, Wright, Lawton-Rauh, & Saski, 2017). The "EPSPS cassette" is a 297 kb region and is found as extrachromosomal circular DNA (ec-cDNA) (Koo et al., 2018). The eccDNA in GR populations of A. palmeri from geographically distant locations within the US showed high sequence similarity with no structural variation and few single nucleotide polymorphisms (SNPs), supporting the hypothesis of a single origin of the EPSPS cassette in A. palmeri that then rapidly spread across the US (Gaines, Patterson, & Neve, 2019;Molin et al., 2018).
On the other hand, Küpper et al. (2018) found distinct population genetic structure between GR A. palmeri populations from Georgia and Tennessee based on genome wide SNP analysis, suggesting multiple origins of GR populations. However, Gaines et al. (2019) speculated that the behavior of the eccDNA could be different from those of nuclear genomes because the eccDNA is not integrated with the genome. It would be unlikely for nearly identical EPSPS cassettes to independently evolve in any two individuals.
The EPSPS cassette can be a marker to detect initial plant invasions that occur through accidental introduction. Such a detection system is critical because there are substantial time lags between first arrival and subsequent spread of problematic alien species (Essl et al., 2011), during which time effective mitigation could be done. In this study, to elucidate the initial invasion of GR A. palmeri to Japan, we investigated whether GR A. palmeri was established in major grain-importing ports.

| Sampling of local populations in ports and plant materials
We visited 14 major Japanese ports of entry of international commodities in August-September of 2014 to 2017 ( Figure 1) and censused mainly roadsides of about 10 km 2 around each port. At five of those ports, A. palmeri was growing. At Kashima Port, more than 10, 000 individuals were growing thickly along 1.5 km of roadside and center divider. We randomly collected leaves from approximately 40-60 individuals per year at 2 points (Kashima 1 and 2) for 4 years (Table 1, Okayama, Japan. Ten to fifteen seeds per accession were germinated and grown in pots containing commercial nursery soil. At the 2-3 leaf stage, the DNA was extracted from seedlings.

| DNA extraction
Total DNA was extracted from leaves by using a modified hexadecyltrimethyl ammonium bromide (CTAB) method (Murray & Thompson, 1980). The samples were ground to a fine powder and mixed with 800 µl of CTAB extraction buffer and incubated at 60°C for 20 min. Chloroform, 200 µl, was added and emulsified by shaking. The mixture was centrifuged at 10,000 g for 10 min, and the aqueous phase was collected. The DNA was then precipitated by adding two-thirds of a volume of isopropanol and washing once with 70% ethanol. The DNA was dried and suspended in 100 µl of TE.

| Quantitative PCR
Glyphosate resistance in A. palmeri is conferred by copynumber amplification of the herbicide target site gene 5-enolpyruvylshikimate-3-phosphate synthase (EPSPS). Quantitative real-time PCR was used to measure the EPSPS genomic copy number relative to the acetolactate synthase (ALS) gene following the method of Gaines et al. (2010). The relative EPSPS copy number and glyphosate resistance are reported to be strongly associated and a threshold EPSPS copy number value for GR plants were set to 5 (Gaines et al., 2010). Primer efficiency curves were drawn for each primer set by using a 1×, 1/2×, 1/4×, 1/8×, and 1/16 × dilution series of susceptible genomic DNA. The primer sets EPSF1 and EPSR8 (195-bp product) for the EPSPS gene and ALSF2 and ALSR2 (118-bp product) for the ALS gene (Gaines et al., 2010) were used for quantitative PCR on genomic DNA. and finally a melt-curve analysis to check for primer-dimers.

| PCR analysis of EPSPS cassette
In a previous study, the length of the EPSPS amplicon was extended to 297 kb and termed the "EPSPS cassette" (Molin et al., 2017). The cassette comprises not only the EPSPS locus (10 kb) but also genomic sequences corresponding to 71 putative genes, tandem repeats, and regulatory elements, and 36% of the cassette region is estimated to be absent in susceptible plants (Molin et al., 2017). To clarify whether the entire EPSPS cassette was in resistant individuals, eight primer sets (AW550 × AW553,

| Microsatellite marker analysis
Whole genome sequencing data for A. palmeri were downloaded from the Sequence Read Archive (accession number SRR5012825) at NCBI (http://www.ncbi.nlm.nih.gov/sra). The reads were qualitytrimmed and assembled de novo with CLC Genomics Workbench ver. 10.1.1 (Qiagen, Aarhus, Denmark) using default parameters to construct the contigs. All the contig sequences were used as input for the CD-HIT-EST, MISA, ipcress, and BlastCLUST pipeline (Ueno et al., 2012) to obtain PCR primers for amplifying unique microsatellite sequences with the number of repeat units ≥ 9, 8, 7, 6, and 5 for di-, tri-, tetra-, penta-, and hexa-simple sequence repeats (SSRs) respectively. The contigs for which primer pairs were successfully designed were BLASTed against the NCBI nr database with an e-value of 1e -5 . We selected 48 primer pairs for SSRs with the number of repeat units ≥ 10 and BLAST hits, and the primers with tail sequences for forward primers (Blacket, Robin, Good, Lee, & Miller, 2012) and reverse primers with a "PIG-tailing" modification (Brownstein, Carpten, & Smith, 1996) were synthesized by Eurofins Genomics. Of the 48 primer pairs, 10 were successfully amplified in the multiplex PCR and exhibited polymorphisms.
The PCR was performed in a 6-µl reaction volume containing 1 × Multiplex PCR master mix (Qiagen), primer mix, and 10-50 ng of template DNA. The primer mix for each marker in the multiplex contained 0.1 µM universal fluorescent primer, 0.1 µM tailed forward primer, and 0.2 µM reverse primer. The PCR cycle conditions were 94°C for 15 min; 30 cycles of 94°C for 30 s, 60°C for 60 s, and 72°C for 30 s; and a final extension step at 72°C for 7 min. The PCR products were analysed in a 3,500 Genetic Analyzer with GeneMapper software (Thermo Fisher Scientific). Samples that were genotyped at more than 9 of the 10 microsatellite loci were used for analysis.
Although A. palmeri were growing at five ports, less than 10 individuals were detected at all ports except Kashima. Therefore, genetic diversity parameters were estimated for the Kashima populations by using GenAlEx 6.5 (Peakall & Smouse, 2012). The number of alleles per locus (A), observed heterozygosity (Ho), expected heterozygosity (He), and the fixation index (F IS ) were calculated. The significance of the deviation from Hardy-Weinberg equilibrium (HWE) within each locus, as evidenced by the deviations of F IS from zero, was tested by 10,000 randomizations of alleles among individuals within the population, with Bonferroni correction. These calculations were performed with the software FSTAT 2.9.3.2 (Goudet, 1995). The null allele frequencies for each locus were estimated using the software FREENA (Chapuis & Estoup, 2007). Principal coordinates analysis (PCoA) was conducted based on the pairwise genetic distance (Smouse & Peakall, 1999) by using the program GenAlEx version 6.5 (Peakall & Smouse, 2012).  Table 1).

| Genetic variation
We genotyped 112 A. palmeri individuals at 10 microsatellite loci (  Table 2). The frequencies of null alleles for these loci were estimated to be relatively high ( Table 2).
The results of PCoA showed genetic similarity among individuals from Kashima 1 and 2, Mizushima, and Hakata, where EPSPS gene amplifications were detected (Figure 3). Except for the samples of

| D ISCUSS I ON
In this study, A. palmeri resistant to glyphosate were detected at Kashima, Hakata, and Mizushima ports (Figure 1). To date no GR A. palmeri has been found in any Japanese agricultural land, and the independent evolution of the same EPSPS cassette in multiple Japanese ports seems difficult and unlikely. These populations were most likely introduced via contamination in internationally traded grain commodities. This hypothesis is supported by the PCoA and has lost some characteristics typical of weedy species (OECD, 2000). Therefore, because soybean cannot self-sustain without human interference, the detected GM soybeans were derived from spillage during transport of grain commodity rather than from seed produced by established plants. Thus, with the detection of GM GR soybeans at these ports, the evidence is strong that a species such as A. palmeri could be accidentally introduced as a seed contaminant with relatively high propagule pressure. Another study detected approximately 1,700 contaminant seeds in 10 kg of wheat imported from Canada to Japan (Shimono & Konuma, 2008).
In addition to Kashima, a few resistant individuals were growing at Hakata and Mizushima. In these areas, A. palmeri might be a casual plant. These alien plants may flourish and even reproduce occasionally in an area, but not form self-replacing populations. However, even if A. palmeri is currently a casual plant at these two ports, establishment in the near future is possible due to the high propagule pressure.
At Kashima, more than 10,000 individuals were growing, and  (Giacomini, Westra, & Ward, 2014;Vila-Aiub et al., 2014). Therefore, once resistance evolves in a population, the frequency of the resistance gene is expected to remain relatively constant over time, even though herbicide selection is absent.

F I G U R E 3
Principal coordinates analysis of Amaranthus palmeri. PCoA1 (8.05%) and PCoA2 (6.73%) refer to the first and second principal coordinates respectively The frequencies of individuals with the EPSPS cassette were approximately 24%-29% at Kashima 1 for 4 years. Some of these individuals had the cassette without EPSPS gene amplification.
The EPSPS cassette is carried by eccDNA, as described earlier.
The eccDNA is tethered to chromosomes by a structural protein, which enables transmission from cell to cell during mitosis and the copy number is variable among cells (Koo et al., 2018). Therefore, to explain the individuals without EPSPS gene amplification, the DNA may have been extracted from a leaf with low EPSPS gene amplification. Further study is needed to clarify whether individuals that possess the cassette without gene amplification are glyphosate resistant.
Glyphosate-resistant A. palmeri was first reported in Georgia in 2004 (Culpepper et al., 2006) and is now widespread in the southern US. Less than 10 years from that first report, GR A. palmeri has become established in Japan. Thus, this study provides evidence that plant invasions are being accelerated by global trade. Because the proportion of species introduced accidentally is expected to increase through time as trade increases (Lehan et al., 2013), we must expedite the process for initial detection of potentially problematic species.

ACK N OWLED G M ENTS
We thank Mr. Yoshiki Kawano for providing Oita samples. We thank Drs. Yoshiko Shimono and Takashi Enomoto for their assistance with field sampling. This research was financially supported by a Grantin-Aid for Scientific Research (C) (No. 15K07322) from the Japan Society for the Promotion of Science.

AUTH O R CO NTR I B UTI O N S
AS designed the research, conducted field sampling, analyzed the data, and wrote the paper. HK and SN conducted field sampling and genetic experiments. SU developed the microsatellite markers and extensively edited the paper. JY provided the seed materials. MA conducted field sampling and provided conceptual advice.