1 Exploring potential establishment of marine rafting species after transoceanic 1 long-distance dispersal 2 Christina Simkanin 1*, James T. Carlton2, Brian Steves1, Paul Fofonoff1, Jocelyn C. Nelson3,4, Cathryn 3 Clarke Murray3,5, Gregory M. Ruiz1 4 5 1 Smithsonian Environmental Research Center, Edgewater, MD, USA 6 2 Williams College - Mystic Seaport, Mystic, CT, USA 7 3 North Pacific Marine Science Organization (PICES), Sidney, BC, Canada 8 4 Fisheries and Oceans Canada, Pacific Biological Station, Nanaimo, BC, Canada 9 5 Fisheries and Oceans Canada, Institute of Ocean Sciences, Sidney, BC, Canada 10 * Corresponding author - 647 Contees Wharf Rd, Edgewater, MD 21037-0028; csimkanin@gmail.com; 11 simkanic@si.edu 12 13 35 Acknowledgements 36 We are grateful to the large team of collaborators and taxonomists who contributed enormous time and 37 expertise to collecting tsunami-related debris and identifying organisms. We also gratefully 38 acknowledge those who assisted in the literature review and compilation of relevant references: 39 Meagan Abele, Reva Gillman, Shigeo Kawaguchi, Lauran Liggan, Kiyotaka Matsumura, Hiroshi Ogawa, 40 Michio Otani, Masaki Sakaguchi, Danielle Scriven, and Janson Wong. Research support was provided by 41 the Ministry of Environment of Japan through the North Pacific Marine Science Organization (PICES); 42 National Science Foundation (Division of Ocean Science, Biological Oceanography), NSF-OCE-1266417, 43 1266234, 12667, and 1266406; and the Smithsonian Institution. 44 45 Biosketch 46 Christina Simkanin is a marine ecologist interested in biogeography, conservation biology, and nearshore 47 ecosystems. The research team includes scientists from the United States and Canada who work on the 48 ecology of non-native species and the role of marine debris in species transport. 49 50 Short running title: Modelling establishment of marine rafting species 51 52 53 54 2 Abstract 55 Aim On March 11, 2011, the Great East Japan Earthquake triggered a massive tsunami that resulted in 56 the largest known rafting event in recorded history. By spring 2012, marine debris began washing 57 ashore along the Pacific Coast of the U.S. and Canada with a wide-range of Asian coastal species 58 attached. We used this unique dataset, where the source region, date of dislodgment, and landing 59 location are known, to assess the potential for species invasions by transoceanic rafting on marine 60 debris. 61 Location Northeast Pacific from 20 to 60°N 62 Time period Current 63 Major taxa studied Forty-eight invertebrate and algal species recorded on Japanese tsunami marine 64 debris. 65 Methods We developed Maximum Entropy (MaxEnt) species distribution models for 48 species 66 recorded on Japanese tsunami marine debris to predict establishment potential along the Pacific Coast 67 from 20-60°N. Models were compared within the context of historical marine introductions from Japan 68 to this region to validate the emergence of marine debris as a novel vector for species transfer. 69 Results Overall, 27% (13 species) landed with debris at locations with suitable environmental conditions 70 for establishment and survival, indicating that these species may be able to establish new populations or 71 introduce greater genetic diversity to already established non-native populations. A further 22 species 72 have environmental match in areas where tsunami debris likely landed, but was not extensively 73 sampled. Nearly 100 Japanese marine species previously invaded the northeastern Pacific, 74 demonstrating this region's environmental suitability for rafting Japanese biota. Historical invasions 75 from Asia are highest in California and largely known from bays and harbors. 76 Main conclusions Marine debris is a novel and growing vector for non-native species introduction. By 77 utilizing a unique dataset of Japanese tsunami marine debris species, our predictive models show 78 capacity for new transoceanic invasions and can focus monitoring priorities to detect successful long-79 distance dispersal across the world’s oceans. 80 81 Keywords: biological invasions, introduced species, Japanese tsunami, long-distance dispersal, marine 82 debris, marine rafting, MaxEnt, species distribution modelling 83 84 3 85 1. Introduction 86 Accidental and episodic long-distance dispersal is considered a critical process for population 87 expansion (Darwin, 1859; Carlquist, 1981). However, the unpredictable and random nature of its 88 occurrence poses a challenge for quantifying its role in shaping species distributions (Nathan, 2006; 89 Gillespie et al., 2012). In some cases, extreme or catastrophic events may trigger long-distance 90 dispersal, as was shown for the 2011 Great Japan Tsunami (Carlton et al., 2017). On March 11, 2011 the 91 Great East Japan Earthquake, with a magnitude of 9.0, struck the Tōhoku region and triggered a massive 92 tsunami. Waves up to 40.5 m high inundated more than 550 km2 of land, causing large-scale 93 devastation and washing immense amounts of material and debris into the ocean. While non-buoyant 94 materials sank close to shore, buoyant material created a massive, floating debris field (Goto & Shibata, 95 2015). The dominant currents in the region, the Kuroshio and Oyashio, converge off the coast of Tōhoku 96 and curve eastward to join the North Pacific Current. This predominantly eastward flow meant that by 97 spring 2012, debris associated with the tsunami had transited the Pacific Ocean and started washing 98 ashore on beaches along the United States (U.S.) and Canadian coasts (Carlton et al., 2017). This debris 99 harbored hundreds of living Japanese species, having survived the journey, with potential to establish 100 populations in North America. 101 In the oceans, species have rafted long distances while attached to natural substrates, such as wood, 102 for millennia. However, most contemporary floating debris has shifted to being anthropogenic in origin 103 (Thiel & Gutow, 2005a). Since the 1950s, when reports of oceanic plastic pollution first appeared (Ryan, 104 2015), this non-biodegradable substrate has increased in prevalence throughout the world’s seas 105 (Thompson et al., 2004; Law, 2017). Plastic, of all shapes and sizes, is now ubiquitous in marine 106 environments and occurs as shoreline debris on all continents and islands (Barnes, 2002; Lavers & Bond, 107 2017). This debris can host and transport living marine species (Winston et al., 1997; Thiel & Gutow, 108 2005b). Critically, plastic and other human-made debris persists much longer than natural materials 109 (e.g., logs), such that previously rare transoceanic dispersal events may now be more common, but little 110 is known about the likelihood or prevalence of species spread associated with this growing transport 111 mechanism. 112 Biofouling, or the accumulation of sessile assemblages and associated mobile taxa on surfaces of 113 solid substrates, is a well-known and potent vector for global species introductions (Hewitt & Campbell, 114 2010; Davidson et al., 2018). Most examples of species introductions resulting from biofouling transfers 115 are associated with commercial ships, recreational boats, or as epibionts on algae or shellfish associated 116 4 with live bait and aquaculture shipments (Ruiz et al., 2000; Cohen et al., 2001; Floerl et al., 2005). There 117 are far fewer documented cases of species introductions through rafting on ocean debris (but see 118 Censky et al., 1998 for a documented example of a lizard introduction). The longevity of plastic and its 119 growing ubiquity in marine environments (Cozar et al., 2014; Jambeck et al., 2015) suggests that marine 120 debris may play an increasingly important role as a vector of potentially nuisance non-native species. 121 Ship-based surveys of debris in the open ocean have found a diversity of coastal marine invertebrates 122 including hydroids, barnacles, amphipods, bryozoans, echinoderms and bivalves (Goldstein et al., 2014; 123 Gil & Pfaller, 2016), which could establish new populations on arrival to foreign coastlines. 124 Quantifying invasion risks associated with species rafting on marine debris is challenging, as items 125 are typically devoid of any identifying characteristics to determine origin or duration at sea, and the 126 frequency of species arrivals on drift material is difficult to enumerate over large scales (Rech et al., 127 2016). Furthermore, surviving the dispersal event is only the first step to population establishment 128 (Blackburn et al., 2011); upon arrival, species face varying abiotic and biotic conditions which can 129 preclude survival and reproduction. Predictive models to investigate the suitability of a receiving 130 environment for species persistence may help infer invasion risks. Specifically, species distribution 131 models (also commonly known as Environmental Niche Models), which utilize global occurrence records 132 and broad scale environmental data, can be used to predict species distributions in novel environments 133 and conditions (Peterson & Vieglais, 2001; Guisan et al., 2014) and can provide valuable information for 134 targeted monitoring and surveillance efforts (Guisan et al., 2013). 135 A large scale survey of the organisms found attached to debris associated with the 2011 Japanese 136 tsunami identified over 289 species from 16 phyla of marine species on a variety of anthropogenic items 137 (Carlton et al., 2017). Here, we use this unique dataset, where the source region and the date of 138 dislodgement into the ocean are known, to evaluate the potential for species introduction through 139 transoceanic rafting. To do this, we used two complementary methods: (1) species distribution 140 modelling to investigate the environmental suitability of the Pacific coast of the U.S. and Canada for 141 Japanese species recorded on tsunami debris – many of which are not known to be previously 142 introduced or established; and (2) an assessment of historical records of marine species introductions 143 from Japan to the western U.S. and Canada, to place the recent rafting event within the context of 144 historical anthropogenic biotic exchange between these regions. This analysis provides predictive 145 evidence for the potential spatial distribution of species introduced through rafting, an approach which 146 can inform management and monitoring efforts while providing data to assess a rarely quantified 147 biogeographical process – long-distance dispersal. 148 5 149 2. Materials and Methods 150 2.1 Distribution modelling of species rafting on Japanese tsunami marine debris 151 In June 2012, confirmed items of Japanese tsunami marine debris (JTMD) began washing ashore 152 along the Oregon coast (Carlton et al., 2017). Shortly after, an extensive network of local, state, 153 provincial and federal agencies, private citizens, and environmental groups from California to Alaska and 154 Hawaii was established to recognize, record, and quantify tsunami debris items. To distinguish tsunami 155 objects from other types of ocean debris, multiple lines of evidence were combined, including (a) formal 156 identification through the Japanese Consulate using registration numbers or other identifying 157 characteristics from floating objects; (b) the use of oceanographic models and tracking databases to 158 predict debris fields and pulses; and (c) bioforensics using the known non-random diversity of coastal 159 marine species from shores of the Tōhoku region to indicate location of origin (see Carlton et al., 2017). 160 Recorded JTMD items included vessels, plastic totes, buoys, fishing gear, floating docks, post-and-beam 161 wood, and many other human-made items associated with coastal communities. Overall, more than 630 162 pieces of confirmed JTMD objects that washed ashore on the Pacific Coast of the U.S. and Canada were 163 sampled and found to carry over 289 species (Carlton et al., 2017). The majority of the sampled debris 164 was collected along the Oregon and Washington coasts, spanning from 42 to 48°N, but oceanographic 165 models show that high amounts of tsunami generated debris likely landed within a much larger region 166 spanning from Northern California to British Columbia and Alaska (Clarke Murray et al., 2018). 167 We used MaxEnt (ver. 3.4.0) species distribution modelling (Phillips et al., 2006) to locate areas 168 along the Pacific Coast of the U.S. and Canada that are most suitable to the establishment of species 169 identified from tsunami associated marine debris. We gathered geographically referenced global 170 occurrence records from species native and non-native distributions using the Global Biodiversity 171 Information System (GBIF.org 2016); the U.S. National Exotic Marine and Estuarine Species Information 172 System (Fofonoff et al., 2016); and detailed searches of literature from Japan, Russia, Korea, and China 173 to contribute additional records from species’ native ranges. Our aim was to collate data across 174 resources to reduce the presence of sampling bias in occurrence records and produce the most accurate 175 map of species current global distributions. For modeling, each species reported from tsunami debris 176 (see species list in Carlton et al., 2017) had to meet a series of criteria: (1) be identified to species level; 177 (2) not be suspected to be part of a species complex; (3) be reported from at least one object that 178 landed along the Pacific Coast of the U.S. or Canada (this excluded species reported only from debris 179 landing in Hawaii); (4) have georeferenced occurrence records from Japan; and (5) have at least 20 180 6 global occurrence records at 5 arcminute (c. 9.2 km or 0.083°) grid cell resolution. We endeavored to 181 only use occurrence records from established populations in our modelling, therefore species records 182 from debris landings were not included in this analysis. Previous research shows that MaxEnt performs 183 relatively well for species with low numbers of occurrence records (e.g. 20-30 spatially seperated points; 184 Hernandez et al., 2006; Pearson et al., 2007), but simpler models are necessary for species with few 185 occurrences to avoid over-fitting (Merow et al., 2014). After applying these criteria, 48 species from 10 186 phyla of marine algae and invertebrates remained for modelling. A majority of species were sessile 187 invertebrates (n=30), while 16 were mobile invertebrates and two were algae. Carlton et al. (2017) used 188 body size as a way to infer ages of debris rafting species and excluded species which appeared to be 189 newly settled and were therefore likely to be acquired as floating debris neared the North American 190 coastline. Therefore, this subset of species can be linked directly to marine debris which transited across 191 the Pacific from Japan to the Pacific Coast of North America. 192 Over 17,000 georeferenced occurrence points for the 48 species were collated and quality assessed 193 using the software ModestR 2.0 (http://www.ipez.es/ModestR/) and “spThin” (Aiello-Lammens et al., 194 2015) in R 3.4.2 (R Core Team, 2017). We conducted a systematic procedure to check the accuracy of 195 species occurrences, which included: removing non-contemporary (i.e. fossil) records, removing 196 duplicate points, removing coordinates with low spatial resolution (<2 decimal places), removing 197 terrestrial records, reducing the number of points to match the resolution of the environmental data 198 layers, and removing climatic outliers using jackknife resampling. This exhaustive data cleaning resulted 199 in 3,215 occurrence records remaining, ranging from 20 to 211 georeferenced points per species (Table 200 1). 201 Sixteen candidate environmental data layers at 5-arcminute resolution were selected from Bio-202 ORACLE (Tyberghein et al., 2012; Assis et al., 2017) to represent the abiotic factors essential for marine 203 invertebrate and algae distributions. Collinearity amongst environmental variables can decrease model 204 performance and overall predictive accuracy (Dormann et al., 2013); therefore we followed a variable 205 selection protocol, using the R packages “sdmpredictors” (Bosch, 2017) and “usdm” (Naimi et al., 2014), 206 to remove correlated and multi-collinear variables. We used a pairwise Pearson-moment correlation 207 matrix to identify highly correlated (r>0.70) variables and selected the most ecologically relevant data 208 layers for modelling (Supporting Information TableS1 and S2). Variance inflation factors (VIFs) were also 209 quantified, using a cut-off of 5, to determine how much variance of an estimated regression coefficient 210 is increased because of multi-collinearity (Guisan et al., 2017). In the end, seven suitable environmental 211 predictor variables were chosen for marine invertebrate models: annual mean calcite concentration 212 7 (mol/m-3), annual mean chlorophyll A (mg/m3), annual mean pH, mean primary productivity at minimum 213 depth (g m-2 yr-1), mean seawater salinity (psu), and annual maximum and range of sea surface 214 temperature (°C). For algal species models, six variables were selected, including: mean cloud fraction 215 (%), mean diffuse attenuation (m-1), mean pH, mean seawater salinity (psu), and annual maximum and 216 range of sea surface temperature (°C). 217 MaxEnt is a machine learning algorithm that uses presence only data, a random subset of 10,000 218 pseudo-absence background points, and a group of environmental data layers to determine the optimal 219 probability distribution of a species in alternate space or time (Phillips et al., 2006). There is increasing 220 evidence that use of MaxEnt’s default parameters does not always generate the best possible model 221 output (Anderson & Gonzalez, 2011; Morales et al., 2017). Therefore we used the R package “ENMeval” 222 (Muscarella et al., 2014) to reduce unnecessary model complexity and carry out model tuning for each 223 species. ENMeval runs automated analyses to determine the most optimal feature class and 224 regularization parameters using AICc scores, a version of the Akaike Information Criterion optimized for 225 smaller sample sizes. Background points were generated by using the bathymetry layer from MARSPEC 226 (Sbrocco & Barber, 2013) to create a global raster clipped to depths of 0-200m along the coastline. This 227 ‘depth raster’ was randomly sampled for 10,000 points to be used as pseudo-absences in the model. 228 This depth raster was also used as a mask in our raster stack of environmental layers to limit model 229 predictions to nearshore (<200 m) water depth. Data partitioning for testing and training was carried out 230 using jackknife cross-validation for species with >25 occurrence records, whereas block structured cross-231 validation was used for species with 25 or more records (Muscarella et al., 2014). Block cross-validation 232 is known to be most suitable for models including species transfers across space and time, such as native 233 versus non-native regions (Wenger & Olden, 2012). The feature classes and regularization multiplier 234 parameters selected to represent the simplest MaxEnt model with best fit for each of the 48 tsunami 235 debris species are shown in Supporting Information Table S3. For MaxEnt runs using the R package 236 ‘dismo’ (Hijmans et al., 2017), parameters leading to the simplest model and best fit were set for each 237 species individually, and global occurrence data were split into two parts – 70% of data were used for 238 model training and 30% were used for model testing. Jackknife resampling was used to test the 239 importance of each environmental variable, alone and relative to the other variables, to species 240 distributions. Modeling output used complementary log-log (cloglog) (Phillips et al., 2017) to produce an 241 estimate of occurrence probability for each species for the Pacific Coast of the U.S. and Canada 242 (Supporting Information Figures S1-48). The 48 suitability maps were transformed using MaxSST or the 243 ‘maximum test sensitivity plus specificity cloglog threshold’ (Liu et al., 2013) from the MaxEnt modelling 244 8 output and combined into a single binary heat map reflecting the risk of invasion of 1-48 tsunami debris 245 species. To determine which species landed attached to debris in areas of high predicted environmental 246 suitability for survival and growth, we obtained the predicted cloglog probability score for each species 247 within each 5 arcmin grid cell along the Pacific coast. MaxEnt modelling outputs are shown in Supporting 248 Information Table S4 and debris landing locations for each species were garnered from the 249 Supplementary Information in Carlton et al. (2017). 250 2.2 Prior anthropogenic biotic exchange from Japan to the Pacific coast of the U.S. and Canada 251 To understand earlier episodes of anthropogenic biotic exchange across the Pacific, we documented 252 the number of marine invertebrates and algae (i.e. excluding vascular plants and vertebrates) that are 253 native to Japan and have non-native populations on the Pacific Coast of the U.S. and Canada. This list 254 was compiled using records from published literature, museum collections, and field-based surveys 255 (Fofonoff et al., 2016; G.M. Ruiz unpublished data). We restricted our analyses to species with 256 established non-native populations, which we considered to exist if: (1) there were multiple occurrence 257 records over multiple years for a given location; (2) local populations were reported to be numerous and 258 successfully reproducing; or (3) the species was reported as established. Thus, we excluded some non-259 native species that were not known to be established or were considered to have diminished or failed 260 populations. We measured latitudinal extent of species non-native ranges on the Pacific coast of North 261 America by determining the most northerly and most southerly record of species established occurrence 262 and assessed whether currently known non-native populations are centered in bays, harbors and 263 estuaries, and/or open coastal locations. 264 265 3. Results 266 3.1 Distribution modelling of species rafting on Japanese tsunami marine debris 267 The hotspot of environmental suitability for the marine debris species modelled was centered on 268 warm waters surrounding Baja California, Mexico, with an additional peak in San Francisco Bay (Figure 269 1). As most of the sampling for debris objects occurred in Oregon and Washington (Figure 1 inset), most 270 species were recorded from debris that washed ashore in areas without suitable abiotic conditions for 271 survival and growth. Still, of the 48 species, 13 have high (>0.50) occurrence probability rates in 272 locations where tsunami debris was actively retrieved and sampled (Figure 2), indicating that these 273 species may be able to establish new populations or introduce greater genetic diversity to already 274 established non-native populations. Two of these species are not currently introduced or established on 275 the Pacific Coast of the U.S. or Canada – the bryozoan Exochella tricuspis, and the worm Hydroides 276 9 ezoensis; two species are already established in areas along this coast, but are not currently introduced 277 in Oregon or Washington where marine debris was sampled – the gastropod Crepidula onyx (which is 278 native to California) and the algae Undaria pinnatifida (which is introduced in California); and the other 279 nine species are introduced and established near debris landing locations already – the amphipods 280 Ampithoe valida and Caprella mutica, the oyster Crassostrea gigas, the bryozoan Cryptosula pallasiana, 281 the sea anemone Diadumene lineata, the sea-squirt Didemnum vexillum, the isopod Ianiropsis 282 serricaudis, the mussel Mytilus galloprovincialis, and the bryozoan Schizoporella japonica. Interestingly, 283 an additional 22 species have high occurrence probability (>0.50) for some areas of British Columbia and 284 Alaska (Figure 2 and Supporting Information Figures S1-48) where tsunami debris likely arrived (Clarke 285 Murray et al., 2018), but was not able to be retrieved and sampled. Nineteen of these species are not 286 currently known to be introduced and established anywhere on the Pacific Coast of the U.S. or Canada – 287 including the barnacles Chthamalus challengeri, Megabalanus rosa and Tetraclita japonica; the bivalve 288 mollusks Arca boucardi, Bankia carinata, Hiatella orientalis, Mizuhopecten yessoensis, Modiolus 289 nipponicus, Mytilisepta virgata, and Mytilus coruscus; the crab Hemigrapsus sanguineus; the gastropods 290 Lottia dorsuosa, Lottia kogamogai, Reishia bronni, Siphonaria japonica, and Siphonaria sirius; the 291 parasitic hydroid Eutima japonica, and the sea stars Aphelasterias japonica and Patiria pectinifera. 292 3.2 Prior anthropogenic biotic exchange from Japan to the Pacific coast of the U.S. and Canada 293 There are 319 known marine and estuarine invertebrate and algal species introduced and 294 established on the Pacific Coast of the U.S. and Canada, and 99 (or 31%) of these are native to Japan. 295 This represents 19 algae and 80 marine invertebrates, spanning 13 marine phyla (Supporting 296 Information Table S5). Although this number is a comprehensive assessment based on current 297 knowledge, it is also likely a conservative estimate, as the taxonomic and biogeographic history of many 298 marine species is still unknown or developing. The hotspot of invasion for non-native species originally 299 from Japan (Figure 3) is centered on San Francisco Bay and central California, extending from 36 to 38°N. 300 Many of these species have introduced ranges which extend further south into southern California, 301 while far fewer species of Japanese origin have ranges known to extend north into Oregon, Washington, 302 British Columbia and Alaska. A majority of these species have non-native populations in bays, harbors 303 and estuaries (n=96), while fewer (n=24) have known non-native populations on more open and 304 exposed coastlines (Supporting Information Table S5). 305 306 307 308 10 4. Discussion 309 The role of long-distance dispersal in population expansion is historically difficult to enumerate, but 310 evidence from the Japanese tsunami of 2011 shows that a wide variety of marine organisms are able to 311 survive for years adrift on debris across more than 7,000 km of ocean (Carlton et al., 2017). As the 312 amount of oceanic debris continues to increase (Eriksen et al., 2014; Lebreton et al., 2018), this dispersal 313 mechanism is likely to grow in prevalence and may lead to increased likelihood of species invasions. 314 Dispersal is a critical first step to population establishment, but on arrival, individuals surviving oceanic 315 transport on debris face several barriers to population growth (Carlquist, 1966; Blackburn et al., 2011). 316 The first such barrier is suitable abiotic conditions for survival and reproduction. Our species distribution 317 models show that 13 of the 48 marine debris species have abiotic tolerances which match the 318 environmental conditions in areas where they were sampled from tsunami debris, and a further 22 have 319 some environmental match to conditions in portions of British Columbia and Alaska where large 320 amounts of tsunami debris may have landed, but was not sampled. Twenty one of the species we 321 modelled are known to be introduced to other locations around the globe (Table 1). Past invasion 322 history is often considered a predictor of future invasion success (Kolar & Lodge, 2001) indicating that a 323 species is capable of establishing populations under novel environmental conditions in other non-native 324 ecosystems. New research, however, has highlighted the increasing proportion of non-native species 325 that have no prior invasion history and this is driven by links between new regions and species pools 326 (Seebens et al., 2018), for which, the emergence of anthropogenic floating debris as a vector for species 327 dispersal is a good example. These novel and emerging species and vectors pose a particular challenge 328 for introduced species management, but species distribution models (when possible) can provide early 329 predictions to aid targeted monitoring and quick responses to new invaders. 330 The 48 species modelled here are a small subset of the total species pool rafting on ocean debris. 331 For Japanese tsunami debris alone, at least 289 species (Carlton et al., 2017) have been recorded and 332 rarefaction curves show that this number is only a portion of the total taxa that arrived (Carlton et al., 333 2017). Furthermore, a large amount of debris likely washed up along the coastline without detection or 334 without identifying characteristics to connect it to the 2011 Japanese tsunami event and this debris may 335 have carried varied species richness. Ongoing field surveys show that tsunami debris from this event is 336 still washing ashore, although the frequency has decreased steadily since summer 2015 (Carlton et al., 337 2018). A greater number of species likely pose invasion risks for the Pacific coast of North America, but 338 without more knowledge of their arrival, along with the limited availability of global georeferenced 339 11 distribution records for most species, our ability to provide an assessment of the invasion risks posed by 340 species rafting on marine debris is hindered. Our data show, however, that 27% of species modeled 341 here have high environmental match with known debris landing locations. Further, most (n=45) of the 342 48 species have a planktonic life phase ranging from 130 days in the case of the sea star Asterias 343 amurensis to the minutes to hours of most of the ascidian and bryozoan species, highlighting that 344 natural dispersal across the Pacific Ocean is not possible (Table 1). 345 Nine of the marine debris species with high environmental suitability in the region of tsunami debris 346 landings are already established with large introduced ranges (from California to Alaska) on the Pacific 347 Coast of the U.S. and Canada. These species are clearly capable of survival, growth and reproduction in 348 this introduced range and the arrival of new propagules may increase genetic diversity of currently 349 established populations (Hanyuda et al., 2018). The Japanese tsunami debris dataset highlights the 350 ability of these species to raft across large swaths of ocean, and although these long-distance dispersal 351 events are historically rare, they may play a crucial role in the maintenance of genetic connectivity 352 within meta-populations (Trakhtenbrot et al., 2005). Increased levels of genetic diversity and mixing can 353 enhance spread and persistence of non-native populations by increasing the likelihood of local 354 adaptation, increasing fitness levels through greater mixing of genotypes, and producing novel hybrids 355 that are able to exploit new environments and habitats (Roman & Darling, 2007). In time, genetic tools, 356 used in combination with data from species sampling campaigns, could be used to determine genetic 357 differentiation between populations and estimate historical genetic admixture and linkages, providing 358 evidence for the role of long-distance dispersal in population connectivity. 359 A total of 99 invertebrates and algae from Japan have previously established populations on the 360 Pacific Coast of the U.S. and Canada, further supporting the environmental suitability between these 361 regions. The area of highest prevalence is centered on San Francisco Bay and the north-central part of 362 California. This region has historically been connected to Japan through varied human-mediated vectors, 363 including commercial shipping and Japanese oyster aquaculture (Carlton, 1979; Andrews, 1980). Most 364 debris from the Japanese tsunami is thought to have landed further north along the coastline, from 365 Northern California through Alaska (Clarke Murray et al., 2018). Marine debris represents a novel vector 366 for non-native species spread, in that it differs from other contemporary transport mechanisms, like 367 commercial shipping and aquaculture shipments, by connecting locations that are not typically linked 368 (i.e. outside ports and harbors). Furthermore, the origin and endpoint locations of marine debris are 369 difficult to predict and generally spatially haphazard. Debris, therefore, may act as a species transport 370 12 mechanism to areas rarely associated with traditional vector hubs (shipping terminals, marinas, and 371 aquaculture facilities) such as outer coastlines, which contemporary data show are currently less 372 impacted by introduced species (Zabin et al., 2018). Also, human-made debris dislodged during extreme 373 events often originates from populated and modified coastal ecosystems, with high amounts of coastal 374 infrastructure, which are hubs for non-native species (Ruiz et al., 2009). Landing locations, on the other 375 hand, are wide-ranging and random, and can include pristine, protected and remote areas of low human 376 density – which are typically less impacted by non-native species (Ardura et al., 2016; Gallardo et al., 377 2017). Of the 99 species previously introduced, only 24 are known to have established non-native 378 populations in locations beyond bays and harbors; although this pattern could also be driven by 379 sampling bias as a majority of surveys for non-native species are conducted in bays and harbors, 380 whereas the distribution of non-native species in open coastal sites is far less known (Ruiz et al. 2009). 381 The survival of species attached to debris washing ashore on the U.S. and Canadian Pacific coasts 382 shows that some coastal species can withstand oceanic conditions for years. These data challenge the 383 idea that coastal organisms experience low survival in open ocean conditions because of decreased food 384 availability in oligotrophic conditions (Polovina et al., 2008) and long-term exposure to harmful UV-B 385 (Smith & Baker, 1979). Not only did these species survive the long-distance ocean transit, but some are 386 known to have either undergone gametogenesis during the crossing, such as the mussel Mytilus 387 galloprovincialis (Miller et al., 2018a), or produced multiple filial populations, such as amphipods, 388 isopods, and marine insects. The reported debris species therefore represent a unique subset of 389 organisms which are hardy enough to withstand varying environmental conditions during transit. 390 Previous analyses indicate that some traits, like smaller adult size, a sessile lifestyle, and the ability to 391 reproduce both sexually and asexually may enhance survival of rafting species (Thiel & Gutow, 2005b). 392 An investigation of the traits associated with tsunami debris species more specifically, indicates that 393 having a greater occurrence on artificial substrates amongst biofouling communities and greater salinity 394 tolerance are consistent across species which have prior global invasion histories and are therefore 395 considered the most likely to establish in the U.S. and Canada (Miller et al., 2018b). 396 There are limitations to how accurate species distribution models can be for detecting newly 397 established species. For instance, despite having suitable environmental conditions at debris landing 398 sites, species must overcome a number of additional barriers for successful population growth and 399 establishment. These barriers include Allee effects, or the tendency for small populations to experience 400 limited growth due to low genetic diversity, inbreeding, and difficulty finding mates; the potentially 401 13 limited availability of suitable habitat in the receiving environment; and biotic interactions such as 402 predation and competition that can limit establishment success. The lack of consideration for these 403 additional factors is a more general criticism of species distribution models, and efforts are being made 404 to incorporate habitat mosaics, dispersal, and biotic information into mechanistic models to increase 405 predictive accuracy (e.g. Mellin et al., 2016). Habitat filtering, specifically, is likely to limit colonization 406 and establishment for debris species. For instance, species that need hard substrates for attachment 407 and growth are not likely to survive for long durations on sandy beaches, where a majority of the 408 tsunami debris was detected (Carlton et al. 2017). This detection could be an artifact of search effort, 409 however, as rocky shores, especially along high energy open coastlines such as the those in the 410 northeastern Pacific, are less likely to be visited by coastal resource rangers, beach walkers, and 411 beachcombers. Similarly, tsunami debris items washing into bays and estuaries were rarely detected 412 (per Carlton et al. 2017), but these areas have mostly inaccessible shorelines and are closer to urbanized 413 areas increasing the incidence of floating debris overall, making detection of tsunami related items more 414 challenging. 415 Furthermore, choices made during the modelling process itself, specifically in addressing sampling 416 biases in presence and pseudo-absence background points can have large effects on predicted species 417 distributions (Merow et al., 2013; Yuckulic et al., 2013). We attempted to reduce spatial bias in our 418 models by carrying out a randomized selection process for coastal pseudo-absence points, incorporating 419 data from a range of sources (e.g. online global repositories and detailed searches of the research 420 literature) for presence points, and running detailed quality analysis of known occurrence records 421 (described in the methods). We considered using a target list (as suggested by Phillips et al. 2009 and 422 Yuckulic et al., 2013), but this could introduce spatial bias in another way by assuming that researchers 423 investigating nearshore marine invertebrates have broad search images and data collection protocols 424 for the full suite of species modelled here, which is unlikely. This issue of spatial bias is an important 425 one, and greater knowledge of a species true presence and absence in a given location would certainly 426 improve predictive accuracy. Lastly, the occurrence probability estimates shown for each species are 427 not representative of predicted species density or habitat suitability in any way (Elith et al. 2011). The 428 density of an organism is likely to vary greatly, and this is especially true for newly establishing species 429 which are often at low-densities and therefore challenging to detect (Crooks, 2005). 430 Humans are notoriously successful dispersal agents and have been transporting species around the 431 globe, intentionally and unintentionally, for centuries. In modern times, these transits are often 432 14 recorded and can be enumerated over large scales (e.g. Verling et al., 2005; Seebens et al., 2013). The 433 global increase in floating human-made debris, however, changes this association, as these dispersal 434 events often go unnoticed or unaccounted for. The Japanese tsunami provided a unique case in which to 435 understand long-distance dispersal through rafting in the modern era (Carlton et al., 2017; Chown, 436 2017). It is clear that large numbers of coastal marine organisms can withstand varying durations of 437 open ocean transport attached to marine debris, although species-specific variability in growth, 438 reproduction, and survival while attached to oceanic debris is still unknown. The predictive maps and 439 models developed here can be used to set priorities for monitoring for establishment of tsunami debris 440 species, increasing the likelihood of early detection and successful eradication, and providing testable 441 hypotheses for investigating the frequency of successful long-distance dispersal across the world’s 442 oceans. 443 444 15 References 445 Aiello-Lammens, M.E., Boria, R.A., Radosavljevic, A., Vilela, B. & Anderson, R.P. (2015) spThin: an R 446 package for spatial thinning of species occurrence records for use in ecological niche models. 447 Ecography, 38, 541-545. 448 Anderson, R.P. & Gonzalez, I. (2011) Species-specific tuning increases robustness to sampling bias in 449 models of species distributions: An implementation with Maxent. Ecological Modelling, 222, 450 2796-2811. 451 Andrews, J.D. (1980) A review of introductions of exotic oysters and biological planning for new 452 importations. Marine Fisheries Review, 42, 1-11. 453 Ardura, A., Juanes, F., Planes, S. & Garcia-Vazquez, E. (2016) Rate of biological invasions is lower in 454 coastal marine protected areas. Scientific Reports, 6, 33013. 455 Assis, J., Tyberghein, L., Bosch, S., Verbruggen, H., Serrão, E.A. & De Clerck, O. (2017) Bio-ORACLE v2.0: 456 Extending marine data layers for bioclimatic modelling. Global Ecology and Biogeography, 27, 457 277-284. 458 Barnes, D.K.A. (2002) Invasions by marine life on plastic debris. Nature, 416, 808-809. 459 Blackburn, T.M., Pysek, P., Bacher, S., Carlton, J.T., Duncan, R.P., Jarosik, V., Wilson, J.R. & Richardson, 460 D.M. (2011) A proposed unified framework for biological invasions. Trends in Ecology & 461 Evolution, 26, 333-9. 462 Bosch, S. (2017) sdmpredictors: species distribution modelling predictor datasets. R package version 463 0.2.6. 464 Carlquist, S. (1966) The biota of long-distance dispersal, I: Principles of dispersal and evolution. The 465 Quarterly Review of Biology, 41, 247-270. 466 Carlquist, S. (1981) Chance dispersal: long-distance dispersal of organisms, widely accepted as a major 467 cause of distribution patterns, poses challenging problems of analysis. American Scientist, 69, 468 509-516. 469 Carlton, J., Chapman, J., Geller, J., Miller, J., Ruiz, G., Carlton, D., McCuller, M., Treneman, N., Steves, B., 470 Breitenstein, R., Lewis, R., Bilderback, D., Bilderback, D., Haga, T. & Harris, L. (2018) Ecological 471 and biological studies of ocean rafting: Japanese tsunami marine debris in North America and 472 the Hawaiian Islands. Aquatic Invasions, 13, 1-9. 473 Carlton , J.T. (1979) Introduced invertebrates of San Francisco Bay. Pacific Division of the American 474 Association for the Advancement of Science, San Francisco. 475 Carlton, J.T., Chapman, J.W., Geller, J.B., Miller, J.A., Carlton, D.A., McCuller, M.I., Treneman, N.C., 476 Steves, B.P. & Ruiz, G.M. (2017) Tsunami-driven rafting: Transoceanic species dispersal and 477 implications for marine biogeography. Science, 357, 1402-1406. 478 Censky, E.L., Hodge, K., Dudley, J. (1998) Over-water dispersal of lizards due to hurricans. Nature 395 479 (6702): 556. 480 Chown, S.L. (2017) Tsunami debris spells trouble. Science, 357, 1356. 481 Clarke Murray, C.C., Maximenko, N. & Lippiatt, S. (2018) The influx of marine debris from the Great 482 Japan Tsunami of 2011 to North American shorelines. Marine Pollution Bulletin, 132: 26-32. 483 Cohen, A.N., Weinstein, A., Emmett, M.A., Lau, W. & Carlton , J.T. (2001) Investigations into the 484 introduction of non-indigenous marine organisms via the cross-continental trade in marine 485 baitworms. In: A report for the U.S. Fish and Wildlife Service, p. 28, Sacramento, CA. 486 Cozar, A., Echevarria, F., Gonzalez-Gordillo, J.I., Irigoien, X., Ubeda, B., Hernandez-Leon, S., Palma, A.T., 487 Navarro, S., Garcia-de-Lomas, J., Ruiz, A., Fernandez-de-Puelles, M.L. & Duarte, C.M. (2014) 488 Plastic debris in the open ocean. Proceedings of the National Academy of Sciences of the U.S.A., 489 111, 10239-44. 490 16 Crooks, J.A. (2005) Lag-times and exotic species: The ecology and management of biological invasions in 491 slow motion. Ecoscience, 12, 316-329. 492 Darwin, C. (1859) On the orgin of species by means of natural selection. John Murray, London. 493 Davidson, I.C., Scianni, C., Minton, M.S., Ruiz, G.M. & Vamosi, S. (2018) A history of ship specialization 494 and consequences for marine invasions, management and policy. Journal of Applied Ecology, 55, 495 1799-1811. 496 Dormann, C.F., Elith, J., Bacher, S., Buchmann, C., Carl, G., Carré, G., Marquéz, J.R.G., Gruber, B., 497 Lafourcade, B., Leitão, P.J., Münkemüller, T., McClean, C., Osborne, P.E., Reineking, B., Schröder, 498 B., Skidmore, A.K., Zurell, D. & Lautenbach, S. (2013) Collinearity: a review of methods to deal 499 with it and a simulation study evaluating their performance. Ecography, 36, 27-46. 500 Elith, J., Phillips, S.J., Hastie, T., Dudík, M., Chee, Y.E., Yates, C.J. (2011) A statistical explanation of 501 MaxEnt for ecologists. Diversity and Distributions, 17, 43-57. 502 Eriksen, M., Lebreton, L.C., Carson, H.S., Thiel, M., Moore, C.J., Borerro, J.C., Galgani, F., Ryan, P.G. & 503 Reisser, J. (2014) Plastic Pollution in the World's Oceans: More than 5 Trillion Plastic Pieces 504 Weighing over 250,000 Tons Afloat at Sea. PLoS One, 9, e111913. 505 Floerl, O., Inglis, G.J. & Hayden, B.J. (2005) A risk-based predictive tool to prevent accidental 506 introductions of nonindigenous marine species. Environmental Management, 35, 765-78. 507 Fofonoff, P., Ruiz, G., Steves, B., Simkanin, C. & Carlton, J. (2016) National Exotic Marine and Estuarine 508 Species Information System. Available at: http://invasions.si.edu/nemesis/ (accessed November 509 1 2016) 510 Gallardo, B., Aldridge, D.C., Gonzalez-Moreno, P., Pergl, J., Pizarro, M., Pysek, P., Thuiller, W., Yesson, C. 511 & Vila, M. (2017) Protected areas offer refuge from invasive species spreading under climate 512 change. Global Change Biology, 23, 5331-5343. 513 GBIF.org (2016), GBIF Home Page. Available from: http://www.gbif.org (accessed 13 September 2016). 514 Gil, M.A. & Pfaller, J.B. (2016) Oceanic barnacles act as foundation species on plastic debris: implications 515 for marine dispersal. Scientific Reports, 6, 19987. 516 Gillespie, R.G., Baldwin, B.G., Waters, J.M., Fraser, C.I., Nikula, R. & Roderick, G.K. (2012) Long-distance 517 dispersal: a framework for hypothesis testing. Trends in Ecology & Evolution, 27, 47-56. 518 Goldstein, M.C., Carson, H.S. & Eriksen, M. (2014) Relationship of diversity and habitat area in North 519 Pacific plastic-associated rafting communities. Marine Biology, 161, 1441–1453. 520 Goto, T. & Shibata, H. (2015) Changes in abundance and composition of anthropogenic marine debris on 521 the continental slope off the Pacific coast of northern Japan, after the March 2011 Tohoku 522 earthquake. Marine Polution Bulletin, 95, 234-41. 523 Guisan, A., Thuiller, W. & Zimmermann, N.E. (2017) Habitat Suitability and Distribution Models: with 524 applications in R. Cambridge University Press, Cambridge, UK. 525 Guisan, A., Petitpierre, B., Broennimann, O., Daehler, C. & Kueffer, C. (2014) Unifying niche shift studies: 526 insights from biological invasions. Trends in Ecology and Evolution, 29, 260-9. 527 Guisan, A., Tingley, R., Baumgartner, J.B., Naujokaitis-Lewis, I., Sutcliffe, P.R., Tulloch, A.I., Regan, T.J., 528 Brotons, L., McDonald-Madden, E., Mantyka-Pringle, C., Martin, T.G., Rhodes, J.R., Maggini, R., 529 Setterfield, S.A., Elith, J., Schwartz, M.W., Wintle, B.A., Broennimann, O., Austin, M., Ferrier, S., 530 Kearney, M.R., Possingham, H.P. & Buckley, Y.M. (2013) Predicting species distributions for 531 conservation decisions. Ecology Letters, 16, 1424-35. 532 Hanyuda, T., Hansen, G.I. & Kawai, H. (2018) Genetic identification of macroalgal species on Japanese 533 tsunami marine debris and genetic comparisons with their wild populations. Marine Pollution 534 Bulletin, 132, 74-81. 535 Hernandez, P.A., Graham, C.H., Master, L.L. & Albert, D.L. (2006) The effect of sample size and species 536 characteristics on performance of different species distribtion modelling methods. Ecography, 537 29, 773-785. 538 17 Hewitt, C. & Campbell, M. (2010) The relative contribution of vectors to the introduction and 539 translocation of marine invasive species. In, p. 56. Department of Agriculture, Fisheries and 540 Forestry, Australia, Canberra, Australia. 541 Hijmans, R.J., Phillips, S., Leathwick, J. & Elith, J. (2017) dismo: Species Distribution Modeling. R package 542 version 1.1-4. 543 Jambeck, J.R., Geyer, R., Wilcox, C., Sieger, T.R., Perryman, M., Andrady, A., Narayan, R. & Law, K.L. 544 (2015) Plastic waste imputs from land into the ocean. Science, 347, 768-771. 545 Kolar, C.S. & Lodge, D.M. (2001) Progress in invasion biology: predicting invaders. Trends in Ecology and 546 Evolution, 16, 199-204. 547 Lavers, J.L. & Bond, A.L. (2017) Exceptional and rapid accumulation of anthropogenic debris on one of 548 the world's most remote and pristine islands. Proceedings of the National Academy of Sciences 549 of the U.S.A., 114, 6052-6055. 550 Law, K.L. (2017) Plastics in the Marine Environment. Annual Review of Marine Science, 9, 205-229. 551 Lebreton, L., Slat, B., Ferrari, F., Sainte-Rose, B., Aitken, J., Marthouse, R., Hajbane, S., Cunsolo, S., 552 Schwarz, A., Levivier, A., Noble, K., Debeljak, P., Maral, H., Schoeneich-Argent, R., Brambini, R. & 553 Reisser, J. (2018) Evidence that the Great Pacific Garbage Patch is rapidly accumulating plastic. 554 Scientific Reports, 8, 4666. 555 Liu, C., White, M., Newell, G. & Pearson, R. (2013) Selecting thresholds for the prediction of species 556 occurrence with presence-only data. Journal of Biogeography, 40, 778-789. 557 Mellin, C., Lurgi, M., Matthews, S., MacNeil, M.A., Caley, M.J., Bax, N., Przeslawski, R. & Fordham, D.A. 558 (2016) Forecasting marine invasions under climate change: Biotic interactions and demographic 559 processes matter. Biological Conservation, 204, 459-467. 560 Merow, C., Smith, M.J., Silander J.A. (2013) A pracitcal guide to MaxEnt for modeling species' 561 distributions: what it does, and why inputs and settings matter. Ecography, 36, 1058-1069. 562 Merow, C., Smith, M.J., Edwards, T.C., Guisan, A., McMahon, S.M., Normand, S., Thuiller, W., Wüest, 563 R.O., Zimmermann, N.E. & Elith, J. (2014) What do we gain from simplicity versus complexity in 564 species distribution models? Ecography, 37, 1267-1281. 565 Miller, J.A., Carlton, J.T., Chapman, J.W., Geller, J.B. & Ruiz, G.M. (2018a) Transoceanic dispersal of the 566 mussel Mytilus galloprovincialis on Japanese tsunami marine debris: An approach for evaluating 567 rafting of a coastal species at sea. Marine Pollution Bulletin, 132, 60-69. 568 Miller, J.A., Gillman, R., Carlton, J.T., Murray, C.C., Nelson, J.C., Otani, M. & Ruiz, G.M. (2018b) Trait-569 based characterization of species transported on Japanese tsunami marine debris: Effect of prior 570 invasion history on trait distribution. Marine Pollution Bulletin, 132, 90-101. 571 Morales, N.S., Fernandez, I.C. & Baca-Gonzalez, V. (2017) MaxEnt's parameter configuration and small 572 samples: are we paying attention to recommendations? A systematic review. PeerJ, 5, e3093. 573 Muscarella, R., Galante, P.J., Soley-Guardia, M., Boria, R.A., Kass, J.M., Uriarte, M., Anderson, R.P. & 574 McPherson, J. (2014) ENMeval: An R package for conducting spatially independent evaluations 575 and estimating optimal model complexity forMaxentecological niche models. Methods in 576 Ecology and Evolution, 5, 1198-1205. 577 Naimi, B., Hamm, N.A.S., Groen, T.A., Skidmore, A.K. & Toxopeus, A.G. (2014) Where is positional 578 uncertainty a problem for species distribution modelling? Ecography, 37, 191-203. 579 Nathan, R. (2006) Long-distance dispersal of plants. Science, 313, 786-788. 580 Pearson, R.G., Raxworthy, C.J., Nakamura, M. & Townsend Peterson, A. (2007) Predicting species 581 distributions from small numbers of occurrence records: a test case using cryptic geckos in 582 Madagascar. Journal of Biogeography, 34, 102-117. 583 Peterson, T.A. & Vieglais, D.A. (2001) Predicting Species Invasions Using Ecological Niche Modeling: New 584 Approaches from Bioinformatics Attack. BioScience, 51, 363-371. 585 18 Phillips, S.J., Anderson, R.P. & Schapire, R.E. (2006) Maximum entropy modeling of species geographic 586 distributions. Ecological Modelling, 190, 231-259. 587 Phillips, S.J., Anderson, R.P., Dudík, M., Schapire, R.E. & Blair, M.E. (2017) Opening the black box: an 588 open-source release of Maxent. Ecography, 40, 887-893. 589 Phillips, S.J., Dudík, M., Elith, J., Graham, C.H., Lehmann, A., Leathwick, J., Ferrier, S. (2009) Sample 590 selection bias and presence-only distribution models: implications for background and pseudo-591 absence data. Ecological Applications, 19, 181-197. 592 Polovina, J.J., Howell, E.A. & Abecassis, M. (2008) Ocean's least productive waters are expanding. 593 Geophysical Research Letters, 35, 1-5. 594 R Core Team (2017) R: A language and environment for statistical computing. R Foundation for 595 Statistical Computing. Available at: https://www.R-project.org/ (accessed 2018). 596 Rech, S., Borrell, Y. & Garcia-Vazquez, E. (2016) Marine litter as a vector for non-native species: What we 597 need to know. Marine Pollution Bulletin, 113, 40-43. 598 Roman, J. & Darling, J.A. (2007) Paradox lost: genetic diversity and the success of aquatic invasions. 599 Trends in Ecology & Evolution, 22, 455-464. 600 Ruiz, G.M., Freestone, A.L., Fofonoff, P.W. & Simkanin, C. (2009) Habitat distribution and heterogeneity 601 in marine Invasion dynamics: the importance of hard substrate and artificial structure. Marine 602 Hard Bottom Communities, (ed. by M. Wahl), pp. 321-332. Springer-Verlag, Berlin Heidelberg. 603 Ruiz, G.M., Fofonoff, P.W., Carlton, J.T., Wonham, M.J. & Hines, A.H. (2000) Invasion of coastal marine 604 communities in North America: Apparent patterns, processes, and biases. Annual Review of 605 Ecology and Systematics, 31, 481-531. 606 Ryan, P.G. (2015) A Brief History of Marine Litter Research. Springer, Cham. 607 Sbrooco, E.J. & Barber, P.H. (2013) MARSPEC: ocean climate layers for marine spatial ecology. Ecology, 608 94, 979. 609 Seebens, H., Gastner, M.T. & Blasius, B. (2013) The risk of marine bioinvasion caused by global shipping. 610 Ecology Letters, 16, 782-790. 611 Seebens, H., Blackburn, T.M., Dyer, E.E., Genovesi, P., Hulme, P.E., Jeschke, J.M., Pagad, S., Pysek, P., van 612 Kleunen, M., Winter, M., Ansong, M., Arianoutsou, M., Bacher, S., Blasius, B., Brockerhoff, E.G., 613 Brundu, G., Capinha, C., Causton, C.E., Celesti-Grapow, L., Dawson, W., Dullinger, S., Economo, 614 E.P., Fuentes, N., Guenard, B., Jager, H., Kartesz, J., Kenis, M., Kuhn, I., Lenzner, B., Liebhold, 615 A.M., Mosena, A., Moser, D., Nentwig, W., Nishino, M., Pearman, D., Pergl, J., Rabitsch, W., 616 Rojas-Sandoval, J., Roques, A., Rorke, S., Rossinelli, S., Roy, H.E., Scalera, R., Schindler, S., 617 Stajerova, K., Tokarska-Guzik, B., Walker, K., Ward, D.F., Yamanaka, T. & Essl, F. (2018) Global 618 rise in emerging alien species results from increased accessibility of new source pools. 619 Proceedings of the National Academy of Sciences of the U.S.A., 115, E2264-E2273. 620 Smith, R.C. & Baker, K.S. (1979) Penetration of UV-B and biologically effective does-rates in natural 621 waters. Photochemistry and Photobiology, 29, 311-322. 622 Thiel, M. & Gutow, L. (2005a) The ecology of rafting in the marine environment. I. The floating substrata. 623 Oceanography and Marine Biology: An Annual Review, 42, 181-263. 624 Thiel, M. & Gutow, L. (2005b) The ecology of rafting in the marine environment. II. The rafting organisms 625 and community. Oceanography and Marine Biology: An Annual Review, 43, 279-418. 626 Thompson, R.C., Olsen, Y.S., Mitchell, R.P., Davis, A., Rowland, S.J., John, A.W.G., McGonigle, D. & Russel, 627 A.E. (2004) Lost at sea: where is all the plastic? Science, 304, 838. 628 Trakhtenbrot, A., Nathan, R., Perry, G. & Richardson, D.M. (2005) The importance of long-distance 629 dispersal in biodiversity conservation. Diversity and Distributions, 11, 173-181. 630 Tyberghein, L., Verbruggen, H., Pauly, K., Troupin, C., Mineur, F. & De Clerck, O. (2012) Bio-ORACLE: a 631 global environmental dataset for marine species distribution modelling. Global Ecology and 632 Biogeography, 21, 272–281. 633 19 Verling, E., Ruiz, G.M., Smith, D., Galil, B., Miller, A.W. & Murphy, K.R. (2005) Supply-side invasion 634 ecology: characterizing propagule pressure in coastal ecosystems. Proceedings of the Royal 635 Society of London Series B, 272, 1249-1257. 636 Wenger, S.J. & Olden, J.D. (2012) Assessing transferability of ecological models: an underappreciated 637 aspect of statistical validation. Methods in Ecology and Evolution, 3, 260-267. 638 Winston, J.E., Gregory, M.R. & Stevens, L.M. (1997) Encrusters, epibionts, and other biota associated 639 with pelagic plastics: a review of biogeographical, environmental, and conservation issues. 640 Springer, New York, NY. 641 Yackulic, C.B., Chandler, R., Zipkin, E.F., Royle, J.A., Nichols, J.D., Campbell Grant, E.H., Veran, S. (2013) 642 Presence-only modelling using MAXENT: when can we trust the inferences? Methods in Ecology 643 and Evolution, 4, 236-243. 644 Zabin, C.J., Marraffini, M., Lonhart, S.I., McCann, L., Ceballos, L., King, C., Watanabe, J., Pearse, J.S. & 645 Ruiz, G.M. (2018) Non-native species colonization of highly diverse, wave swept outer coast 646 habitats in Central California. Marine Biology, 165, 1-18. 647 648 649 650 20 Data Accessibility Statement 651 The species distribution data used in this study are available through the Dryad Digital Repository: 652 http://doi.org/10.5061/dryad.b6np614. These occurrence data were collected from GBIF.org (2016), 653 Fofonoff et al. (2016), and published articles or reports garnered from the scientific literature, and were 654 quality assessed and cleaned for accuracy prior to use. 655 656 657 658 21 Table 1: List of the 48 Japanese tsunami debris species used in species distribution models. Table 659 includes information on the species population status in Japan (i.e. native, introduced or cryptogenic – 660 meaning origin unknown), whether the species is currently known to be established on the Pacific coast 661 of North America, the total number of global occurrence records used in modelling, the total number of 662 debris items the species was record on, and the predominate planktonic phase of the species. *Species 663 in bold have a known invasion history somewhere in the world. 664 Taxa Species Status in Japan Currently established US/Canada West Coast? Global occurrence records Number of tsunami debris items Planktonic phase Amphipod Ampithoe valida Cryptogenic Yes 82 2 direct developer Sea Star Aphelasterias japonica Native No 20 1 planktotrophic Bivalve Arca boucardi Native No 29 6 planktotrophic Sea Star Asterias amurensis Native No 191 1 planktotrophic Bivalve Bankia carinata Native No 38 33 planktotrophic Bivalve Barbatia virescens Native No 40 2 planktotrophic Amphipod Caprella mutica Native Yes 144 7 direct developer Bivalve Chama dunkeri Native No 22 1 planktotrophic Bivalve Chama japonica Native No 20 2 planktotrophic Barnacle Chthamalus challengeri Native No 26 7 planktotrophic Bivalve Crassostrea gigas Native Yes 211 54 planktotrophic Gastropod Crepidula onyx Introduced Yes 90 1 planktotrophic Bryozoan Cryptosula pallasiana Cryptogenic Yes 117 10 lecithotrophic Bivalve Dendostrea folium Native No 91 2 planktotrophic Anemone Diadumene lineata Native Yes 102 3 lecithotrophic Ascidian Didemnum vexillum Native Yes 88 3 lecithotrophic Hydrozoan Eutima japonica Native No 20 6 lecithotrophic Bryozoan Exochella tricuspis Native No 21 5 lecithotrophic Bivalve Glorichlamys quadrilirata Native No 41 1 planktotrophic Algae Grateloupia turuturu Native Yes 75 3 non-flagellated spores Crab Hemigrapsus sanguineus Native No 139 1 planktotrophic Bivalve Hiatella orientalis Native No 25 38 planktotrophic Serpulid Hydroides ezoensis Native No 38 48 planktotrophic Isopod Ianiropsis serricaudis Native Yes 24 36 direct developer Bivalve Isognomon legumen Native No 44 3 planktotrophic Bivalve Laevichlamys cuneata Native No 129 3 planktotrophic Bivalve Laevichlamys squamosa Native No 121 2 planktotrophic Gastropod Lottia dorsuosa Native No 20 5 lecithotrophic Gastropod Lottia kogamogai Native No 34 2 lecithotrophic Barnacle Megabalanus rosa Native No 44 96 planktotrophic Bivalve Mizuhopecten yessoensis Native No 34 1 planktotrophic Bivalve Modiolus nipponicus Native No 24 7 planktotrophic Bivalve Mytilisepta virgata Native No 26 6 planktotrophic 22 Bivalve Mytilus coruscus Native No 17 2 planktotrophic Bivalve Mytilus galloprovincialis Introduced Yes 199 224 planktotrophic Bivalve Pascahinnites coruscans Native No 77 1 planktotrophic Sea Star Patiria pectinifera Native No 39 2 planktotrophic Sipunculid Phascolosoma scolops Native No 44 1 planktotrophic Coral Pocillopora damicornis Native No 150 18 lecithotrophic Gastropod Reishia bronni Native No 34 1 planktotrophic Bryozoan Schizoporella japonica Native Yes 44 2 lecithotrophic Bivalve Septifer bilocularis Native No 179 2 planktotrophic Gastropod Siphonaria japonica Native No 23 1 planktotrophic Gastropod Siphonaria sirius Native No 20 2 planktotrophic Crab Sphaerozius nitidus Native No 20 1 planktotrophic Bivalve Spondylus squamosus Native No 83 1 planktotrophic Barnacle Tetraclita japonica Native No 20 1 planktotrophic Algae Undaria pinnatifida Native Yes 96 1 flagellated spores 665 23 Figure Legends 666 Figure 1: Heat map showing the cumulative predicted presence of the 48 species recorded on marine 667 debris associated with the 2011 Japanese tsunami. Separate maps for each species can be found in the 668 Supporting Information Figures S1-S48. Inset shows the cumulative number of Japanese tsunami debris 669 items found and sampled along the US and Canadian Pacific coast. 670 Figure 2: The MaxEnt complementary log-log probability scores garnered from each 5-arcmin pixel along 671 the Pacific coastline of North America (represented as grey dots) for each of the 48 tsunami marine 672 debris species. Landing locations are shown as black dots. Larger versions of these graphs can be found 673 in the Supporting Information Figures S1-S48. NB: Images are meant to represent broad species groups 674 and are not diagnostic of the species themselves. 675 Figure 3: Heat-map showing the cumulative distributions of the 99 marine invertebrate and algal species 676 which are native to Japan and introduced and established along the Pacific Coast of North America. 677 678 24 Figure 1: Heat map showing the cumulative predicted presence of the 48 species recorded on marine 679 debris associated with the 2011 Japanese tsunami. Separate maps for each species can be found in the 680 Supporting Information Figures S1-S48. Inset shows the cumulative number of Japanese tsunami debris 681 items found and sampled along the US and Canadian Pacific coast. 682 683 25 Figure 2: The MaxEnt complementary log-log probability scores garnered from each 5-arcmin pixel along 684 the Pacific coastline of North America (represented as grey dots) for each of the 48 tsunami marine 685 debris species. Landing locations are shown as black dots. Larger versions of these graphs can be found 686 in the Supporting Information Figures S1-S48. NB: Images are meant to represent broad species groups 687 and are not diagnostic of the species themselves.688 689 26 Figure 2 cont. 690 691 692 693 27 Figure 3: Heat map showing the cumulative distributions of the 99 marine invertebrate and algal species 694 which are native to Japan and introduced and established along the Pacific Coast of North America. 695 696 697 698 28 699 The supporting files include: 700 Table S1: Correlation matrix of candidate Bio-ORACLE marine environmental layers for modelling marine 701 invertebrate species. 702 Table S2: Correlation matrix of candidate Bio-ORACLE marine environmental layers for modelling algal 703 species. 704 Table S3: The feature classes and regularization multiplier parameters selected to represent the 705 simplest MaxEnt model for each species. 706 Table S4: Statistics from MaxEnt species distribution models. 707 Table S5: Marine invertebrates and algae of Japanese origin that are currently introduced and 708 established on the Pacific coast of the United States and Canada. 709 Figures S1 to S48: Individual species maps showing occurrence probability along the Pacific coast of 710 North America. 711 712 713