A peer-reviewed open-access journal NeoBiota 87: 45—72 (2023) doi: 10.3897/neobiota.87.104472 RESEARCH ARTICLE ) NeoBuiota https:/ / neobi ota. pen soft. net Advancing research on alien species and biological invasions Weed wide web: characterising illegal online trade of invasive plants in Australia Jacob Maher', Oliver C. Stringham!??, Stephanie Moncayo', Lisa Wood', Charlotte R. Lassaline', John Virtue'*, Phillip Cassey' | Invasion Science & Wildlife Ecology Lab, University of Adelaide, Adelaide, SA, Australia 2 School of Mathematical Sciences, University of Adelaide, Adelaide, SA, Australia 3 Institute of Earth, Ocean, and Atmospheric Sciences, Rutgers, The State University of New Jersey, New Brunswick, NJ, USA 4 JG Virtue Biosecurity Services, Rockleigh, SA, Australia Corresponding author: Jacob Maher (jacob.maher@adelaide.edu.au) Academic editor: Angela Brandt | Received 4 April 2023 | Accepted 20 July 2023 | Published 11 August 2023 Citation: Maher J, Stringham OC, Moncayo S, Wood L, Lassaline CR, Virtue J, Cassey P (2023) Weed wide web: characterising illegal online trade of invasive plants in Australia. NeoBiota 87: 45-72. https://doi.org/10.3897/ neobiota.87.104472 Abstract Invasive plants seriously impact our environmental, agricultural and forestry assets, and the ornamental plant trade is a major introduction pathway. The variety and extent of the ornamental plant trade is grow- ing in reach and is increasingly facilitated by the internet (i-e., through e-commerce). A lack of surveillance and regulation of e-commerce has resulted in invasive species being widely traded on these platforms. Here, we investigated the extent of illegal trade in invasive plant species in Australia by collecting adver- tisements found on a popular public e-commerce website. Across a 12-month period we collected a total of 235,162 plant advertisements. From 10,000 of these advertisements (4.25% of total advertisements) we found 155 plant taxa advertised online that were prohibited to trade in at least one Australian State or Territory (12.5% of Australia’s total prohibited plant taxa). We detected 1,415 instances of invasive plants advertised, of which 411 breached local jurisdictional (i.e., State or Territory) laws. Opuntia cacti and in- vasive aquatic plants were traded in the greatest quantities. A variety of uses for plants prohibited to trade were reported by the sellers, with aquatic uses being the most popular (i.e., water filtering and habitat for aquatic animals). We used generalised linear mixed-effects models to test the effect of prohibiting the sale of invasive plants on the quantity and price of online advertisements. Despite Australia’s strict internal bi- osecurity regulations, we found that trade prohibitions had no influence on the quantity and price of trade in illegal invasive plants. Given this, and the extent of illegal invasive plants traded, we believe increased monitoring and regulation of online plant trade is warranted. We demonstrate that targeted searches using string matching is an effective tool for detecting e-commerce trade of invasive species. However, to obtain the most optimal outcomes, regulations should be coupled with increased cooperation from e-commerce Copyright Jacob Maher et al. This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. 46 Jacob Maher et al. / NeoBiota 87: 45-72 (2023) platforms and public awareness campaigns. Future weed risk assessments should consider online trade as a key factor in the long-distance dispersal and propagule pressure of a plant. Jurisdictions would also benefit from greater alignment on plant trade prohibitions and revision of associated compliance policies. Keywords Aquatic weeds, biosecurity, e-commerce, Opuntia, ornamental plants, prevention, surveillance, web scraping Introduction Invasive plants can cause serious negative impacts to biodiversity, human health, and primary resource industries (Pysek et al. 2020; Ward et al. 2021). The largest vector of new plant introductions and invasions is the global trade of ornamental plants, which is continually growing in both reach and quantity (Weber et al. 2008; Dodd et al. 2015; Faulkner et al. 2016; van Kleunen et al. 2018; Arianoutsou et al. 2021; Beaury et al. 2021; Rojas-Sandoval et al. 2022). Within this global trade, a pathway of serious concern is trade facilitated by the internet, hereafter termed e-commerce (Derraik and Phillips 2010; Lenda et al. 2014; Humair et al. 2015). E-commerce platforms facilitate long distance dispersal of invasive species and can often circumvent regulations (Giltrap et al. 2009; Derraik and Phillips 2010; Magalhaes and Avelar 2012; Lenda et al. 2014; Humair et al. 2015; Beaury et al. 2021). As a result, e-commerce has proven challeng- ing to monitor and enforce for biosecurity agencies (Derraik and Phillips 2010; Lavor- ena and Sajeva 2021). Many invasive plant species are being traded online despite leg- islative regulations (Humair et al. 2015; Munakamwe and Constantine 2017; Beaury et al. 2021). Without intervention, it is predicted that online trade will lead to further invasive plant incursions (Humair et al. 2015; Peres et al. 2018; Beaury et al. 2021). Australia has a highly endemic floral community that has been severely impacted by plant invasions (Broadhurst and Coates 2017; Bradshaw et al. 2021). Strict importa- tion measures and risk assessment processes have been implemented by the Australian government to prevent the arrival of new alien-invasive plants (Pheloung et al. 1999; Walton 2001; Keller et al. 2007; Simberloff et al. 2013). Even so, Australia already has more than 29,000 introduced alien-plant species (Gallagher and Leishman 2014). There are also native Australian plants which have become invasive outside their indig- enous range (Rose and Fairweather 1997; Morgan et al. 2002; O’Loughlin et al. 2015). Where plant species become invasive, or there is potential to be invasive, state and terri- tory governments (‘jurisdictions hereafter) have the main responsibility for their man- agement and control. A common control measure used by jurisdictions is to ‘declare’ invasive plant taxa in legislation as prohibited to trade within jurisdictional borders (simply ‘declared plant’ hereafter); with 1,236 taxa declared in one or more jurisdictions across Australia. These taxa are declared because they pose significant risks of environ- mental, economic and/or social impacts to natural ecosystems, agricultural and forestry production, and human communities. While legislation differs slightly between ju- risdictions, generally it is prohibited to supply, sell, or transport declared plants, with fines issued for offences. However, e-commerce websites could circumvent traditional Online trade of invasive plants in Australia 47 enforcement measures by trading without physical stores, sending plants by mail, or having buyers collect plants from private residences, resulting in a poorly regulated sector of the horticultural market (Munakamwe and Constantine 2017). Screening for invasive plants entering the country is also challenging due to the high volume of incoming international mail (Australian National Audit Office 2014). Therefore, sur- veillance of e-commerce is an essential tool for detecting and preventing plant invasions (Humair et al. 2015; Lavorgna et al. 2020; Duncan 2021; Stoett and Omrow 2021; Whitehead et al. 2021). E-commerce websites where members of the public post plant advertisements are particularly difficult to monitor. Some efforts have been made to monitor this trade within Australia, however the focus has been limited by time and resources to a handful of problematic species (Munakamwe and Constantine 2017). To investigate the current invasion risk of e-commerce plant trade within Australia (i.e., internal trade, not international shipments into Australia), we applied web-scraping technology to monitor and record plant trade advertisements on a popular Australian e- commerce website over the course of one year. We investigated five research aims: (i) de- termine what proportion of plants advertised are prohibited to trade; (ii) determine the quantity and taxonomic composition of declared plants traded; (iii) determine whether current regulations reduce trade quantity or influence the price of declared plants in ju- risdictions which prohibit trade versus those that permit trade; (iv) characterise the most frequently traded declared plants; and (v) document advertised plant uses to inform our understanding of the desire for declared plants. We seek to provide a clearer picture of the present risk of e-commerce trade and whether prescriptive laws reduce invasive plant trade. ‘These results will help inform future policy decisions regarding the monitoring and prevention of invasive species occurring in the Australian plant trade. Methods Compiling Australia’s declared plants In order to investigate the trade of invasive plants online, we compiled a list of declared plants in Australia. These declared plants are prohibited from trade under jurisdictional biosecurity legislation because of their current or potential impact as invasive species (Parsons and Cuthbertson 2001). Declaration is usually based on an analysis of weed risk using various post-border weed risk management systems (Virtue et al. 2006). Ju- risdictional declarations can include Australian native plant species that have invaded beyond their indigenous range, for example a Western Australian species that is invasive in eastern Australia. Hence declared native plant species are included in this study. To assemble a comprehensive list of declared plants, we used sources relevant to Australia’s eight main jurisdictions (i.e., six states plus Northern Territory and Australian Capital Territory), including government websites, online databases, legislative acts, and ga- zettes (see Suppl. material 1 for complete list of sources). Our compiled list of declared plants and relevant legislation was reviewed and endorsed by appropriate jurisdiction- al officials through the Weeds Working Group of the Australian intergovernmental 48 Jacob Maher et al. / NeoBiota 87: 45—72 (2023) Environment and Invasives Committee. We standardised the taxonomy of the declared plants using the Global Biodiversity Information Facility taxonomic database (GBIF 2021). Our finalised list of declared plant taxa contained 1,236 defined taxa compris- ing 1,178 species, 6 subspecies, and 5 varieties, as well as 47 declared genera. Twenty- two of the declared plant species are recognised as native by the Australian Plant Census and 2 species have uncertain native status (Australian National Herbarium 2023). E-commerce platform selection and building web scrapers We followed established protocols to select e-commerce websites to monitor for sales of plants (Stringham et al. 2020). Specifically, we conducted a systematic web search of invasive plant species names (common and scientific) with an appropriate phrase e.g., “Vinca major for sale Australia” and “Periwinkle for sale Australia’. To optimise the search effort in selecting e-commerce websites for further investigation, we created a short-list of declared species known to be popular in horticulture (Suppl. material 2) (Nursery & Garden Industry Australia 2009). A total of 38 nursery websites and 4 public e-commerce websites were reviewed. We defined nursery websites as private on- line businesses. Public e-commerce websites host online classifieds where members of the public can post personal advertisements. We found plants considered to be invasive on nursery websites, but we did not find any that were declared in the jurisdiction the nursery was located in (i.e., no prohibited advertisements). In contrast, our initial in- vestigations of public e-commerce yielded many prohibited advertisements for declared plants. Alongside our internet search, we consulted with biosecurity officers from each jurisdiction who had experience monitoring the online plant trade. They identified public e-commerce websites over private nursery websites as their primary concern, cit- ing regular detections of declared plants on the former in their own investigations. The risk of public e-commerce is an under-assessed aspect of the ornamental plant trade as it is dificult to monitor and regulate. Based on this recommendation and the findings of our web search, we concentrated our study on one highly popular public e-commerce website. This allowed us to construct a reliable and consistent web scraper for a popular e-commerce website that included seller location data and which frequently traded de- clared plants, based on expert opinion and our preliminary search. This website hosts trade within Australia and is not specific to ornamental plants. However, the website has a ‘plant’ category from which we collected advertisements. Sellers advertise plants, and sales are conducted through private exchanges between traders either online, over the phone, or in person. Therefore, it is important to note that we could not determine how many plants were actually sold from the data we collected. Similarly, we could not determine how many advertisements were relisted plants that had previously failed to sell. We have kept the identity of this website anonymous in accordance with our eth- ics approval (Ethics approval H-2020-184). Personal and identifiable information of traders is available on this website and while publicly available our ethics involve taking a cautious approach to avoid revealing behaviour which may have legal ramifications. Additionally, identifying the website could alter the behaviour of traders which would reduce the value of ongoing surveillance research (Stringham et al. 2020). Online trade of invasive plants in Australia 49 To collect online advertisement data, we constructed a custom web scraper in Python Programming Language (version 3.8.1; Python Software Foundation 2020) using the libraries bs4 (Richardson 2020), requests (Reitz 2020), and selenium (Selenium Main Repository 2020). The web scraper ran daily and collected all advertisements from the designated plant category of the website. Plant advertisement data was stored on a local SQL database. For this study, we explored 12 months of plant advertisements between 1 February 2020 and 31 January 2021. Duplicate collections of advertisements were common because the web scraper ran on a daily basis. We removed these duplicate advertisements based on a unique listing identifier generated by the website. ‘This re- sulted in 235,162 unique advertisements collected over the 12-month period. For our analysis we removed any advertisements that did not provide a seller location, leaving us a dataset of 233,694 advertisements. Sampling and detecting declared plant trade The data we collected were not immediately ready for analysis because the advertise- ments from the website were composed of free-form text boxes completed by the users, and thus the taxonomic names could not be automatically retrieved (i.e., no stand- ardization in names). Identification of plants was conducted manually using text and pictures, provided by the seller, which was a time-consuming process. Subsequently, we explored a subset of the advertisements. For our study, we extracted two samples of 5,000 advertisements each. The first sample was a random sample of all plants traded stratified by jurisdiction. For the second sample we utilised natural language processing to focus specifically on detecting declared plants. The first sample was untargeted; it sampled from all the advertisements we col- lected and did not intentionally target declared plants. This sample was stratified by jurisdiction with 625 unique advertisements randomly sampled from each jurisdic- tion, providing 5,000 advertisements in total. We used this dataset to estimate the underlying proportion of declared plant trade in each jurisdiction and to compare the effectiveness of our targeted sampling method. For the second sample we targeted declared plant advertisements. Our objec- tive was to identify frequently traded declared plants, and capture the composition of declared plants traded. We aimed to capture declared plants traded anywhere in Australia regardless of whether they were advertised in a prohibited jurisdiction. ‘This was to capture the full extent of declared plant trade in Australia. To do this we used string matching to generate a targeted sample aimed at detecting declared plant ad- vertisements (Stringham et al. 2021). String matching is a natural language processing method of finding a sequence of characters, called a string, that match a given charac- ter pattern. In our case the character patterns were the scientific and common names of declared plants. In total, we used 10,573 names to search for the 1,236 declared taxa within the text of collected advertisements. We initially sourced common names from jurisdiction legislation, followed by broader internet searches if necessary (Suppl. material 1) (Shepherd et al. 2001). We cleaned names by removing parentheses and punctuation, converted to lower case, and also pluralised and singularised the names. 50 Jacob Maher et al. / NeoBiota 87: 45—72 (2023) Based on findings by Munakamwe and Constantine (2017), we included common terms for some aquatic species. String matching helped reduce the number of adver- tisements down to a more manageable data set with a higher probability of detecting declared plants. However, common and generic plant names are non-specific and can be shared by many species. This resulted in false positives in the targeted sample. Our pilot investigation revealed some frequent false positives due to the inclusion of certain broad search terms (e.g., ‘lily’ returned many non-target species). We created a list of match exceptions to remove the bulk of the false positives (Suppl. material 3). There- fore, if an advertisement contained the word ‘lily and contained a match exception such as ‘peace lily’ (a non-target species) it would be removed, but an advertisement for ‘arum lily’ would remain. This approach helped us to reduce the number of false positives while retaining the use of certain generic search terms. Out of 233,694 total advertisements, text in the title or description matched to 12,751 advertisements for declared plants. From this, we took a sample of 5,000 unique advertisements. Given our interest in characterising the legality of online trade across Australian jurisdictions, we stratified the sample by jurisdiction. Three jurisdictions had substantially fewer ad- vertisements: Australian Capital Territory, Northern Territory, and Tasmania (Table 1). To help capture trade from these three smaller jurisdictions all advertisements that matched declared plant search terms were analysed. The remaining jurisdictions were randomly sampled until 5,000 unique advertisements was reached (Table 1). We cleaned the sampled datasets by identifying the plants in each advertisement using photos and text provided by the seller. Advertisements would often contain multiple species for sale so we recorded each plant species (or lowest taxonomic rank possible) as a separate identification within an advertisement. We recorded the price and quantity for each plant identified, and the location of the advertisement. It is important to note that recorded locations were seller locations and not where a plant may have been transported to after it had been purchased. Predominately, advertise- ments were for live plants, however we also captured trade of seeds and other prop- agules. We documented and categorised advertisements that stated uses for plants when specified by sellers (i.e., used for purposes other than as a live ornamental plant, including propagules). Once we identified the plant taxa in the advertisements, we cross referenced them with our dataset of 1,236 declared plants. We recorded the number of plant taxa iden- tified and how many were declared plants. We used species accumulation curves to assess how well our samples captured the diversity of plant taxa and declared plant taxa traded online. We measured the number of advertisements containing declared plants and identified advertisements that were prohibited (i.e., the advertisement contained a plant that was declared in the jurisdiction where it was advertised). However, multiple declared plant taxa could appear in a single advertisement. To account for this, we also recorded each detection of a declared plant taxon in any single advertisement. To help explain these different types of trade observations an example with term definitions is provided in Fig. 1. By using these observation metrics, we were able to capture pro- hibited trade of a declared plant and the broader extent of its trade within Australia. Online trade of invasive plants in Australia 51 Table |. The number of advertisements collected and sampled from an e-commerce website stratified by jurisdiction. The table provides the number of advertisements from: (i) 12 months of web scraping (Total dataset); (ii) the untargeted sample (Untargeted); (iii) the string-matching for declared plant taxa (Matched); and (iv) the targeted sample (Targeted). The targeted sample was weighted to better capture trade in three jurisdictions with comparatively lower quantities of matched advertisements: Australian Capital Territory, Northern Territory, and Tasmania (* indicates weighted samples). All advertisements that matched search terms for declared plants in these jurisdictions were cleaned. The remaining advertise- ments were sampled randomly across the remaining jurisdictions to total 5,000 advertisements. Jurisdiction Total dataset | Untargeted Matched Targeted Australian Capital Territory (ACT) 7,362 625 String matching using 420 *420 New South Wales (NSW) 64,641 625 declared plant search 3,351 1,031 Northern Territory (NT) 859 625 nS at 66 *66 Queensland (Qld) 48,909 625 2,893 948 South Australia (SA) 21,121 625 1,073 539 Tasmania (Tas.) 5,991 625 308 *308 Victoria (Vic.) 41,186 625 2,567 921 Western Australia (WA) 43,625 625 2,073 767 Total 233,694 5,000 12,751 5,000 Prohibited Prohibited advertisments (n= 1) Advertisements containing declared plants prohibited to trade in the jurisdiction where the advertisement is located. There can be multiple declared plant taxa in a single advertisement. jurisdiction Individual instances of declared plant taxa identified within advertisements located in jurisdictions which prohibits its trade. Permitted Total declared advertisments (n = 2) jurisdiction i Referring to all advertisements for declared plants in jurisdictions that prohibit and permit trade. :--Total declared detections (n = 3) ? Referring to all detections of declared plants identified in jurisdictions that prohibit and permit trade. Figure |. A diagram explaining the terms we used to define the different types of plant trade observa- tions. This example shows two advertisements and two species of declared plant (plants prohibited to trade in a given jurisdiction). The number of observations for each term in this scenario are provided in parentheses. In the ‘prohibited jurisdiction’ there is one advertisement with two plant species, both species are prohibited to trade in this jurisdiction. One of these plant species is sold by itself in the ‘permitted jurisdiction’. In this case we refer to it as a declared plant, but it is permitted to trade in that jurisdiction. Analysis of prohibited trade on quantity and price We used generalised linear mixed-effects models to test whether prohibited trade had an effect on the trade quantity and price of declared plants. These models considered declared status as the binary explanatory variable and taxa identity as a random effect 52 Jacob Maher et al. / NeoBiota 87: 45—72 (2023) (i.e., random intercept). For quantity, we hypothesised fewer declared plants are ad- vertised in jurisdictions that prohibit their trade compared to jurisdictions that permit their trade. We based our rationale on the notion that laws prohibiting trade would re- duce the number of advertisements online. For price, we hypothesised that in jurisdic- tions that prohibit trade, prices for declared plants would be higher compared to juris- dictions that permit trade. Our rationale was that laws prohibiting trade would result in an increased price to offset their risk; i.e., buyers paying a premium for prohibited plants. We measured the performance of the models using Nakagawa and Schielzeth’s conditional R-squared (Rc2) (Nakagawa and Schielzeth 2013). For all models, we used the targeted dataset, which had the greater number of total declared advertisements compared to the untargeted. For these models, we removed nationally declared taxa, ie., taxa declared in all jurisdictions (n = 130 taxa remaining for quantity compari- son). In the quantity models, we defined quantity as the proportion of advertisements within each jurisdiction’s sample. This approach was to account for differing sample sizes among jurisdictions in the target dataset (see Table 1 for sample sizes). For analys- ing price differences, we used unit prices (price per plant) gathered from the targeted and untargeted datasets. Further, for these price models, we excluded taxa with fewer than two advertisements in each legality category (i.e., prohibited or permitted); this limited the model to 20 taxa. There were two factors that contributed to this reduction. Firstly, a price per plant could not be determined for many advertisements. Either no clear price was provided or plants (particularly aquatic species) were priced by incon- sistent container volumes (i.e., $5 for a full take-away container). Secondly, for some plants price data was absent from a legal category (i.e., no prices recorded in either a prohibited or permitted jurisdiction). We took an additional approach to assess and visualise the difference in quantity and price by exploring the distribution of differences in quantity and price. We calcu- lated the difference of mean quantity and price of each declared plant taxon traded in prohibited jurisdictions compared to permitted jurisdictions (i.e., the mean quantity of taxon A pooled across all prohibited jurisdictions minus the mean quantity of taxon A pooled across all permitted jurisdictions). We used this distribution to determine the degree that prohibited trade affected trade quantity and price, where a distribution centred around zero with low variation suggests little to no influence. Data and software resources We conducted data analysis and visualisation using the R software environment for statistical and graphical computing (version 4.1.1; R Core Team 2022) and used the following packages for our analyses. We verified taxonomy by using the ‘taxize’ package (Scott Chamberlain 2013) and to acquire information from the Global Biodiversity Information Facility taxonomic database. Plant search terms were pluralised using the ‘pluralize’ package (Rudis and Embrey 2020) and string matching was performed us- ing the ‘stringr’ package (Wickham 2019). Collected data was accessed from MySQL database using the ‘DBI’ package (Wickham and Miiller 2022). Regression model Online trade of invasive plants in Australia 53 coeflicients were summarised and extracted using the ‘broom’ package (Robinson et al. 2021). Shapefiles were obtained from the Australian Bureau of Statistics (2021) and visualised using the ‘sf’ package (Pebesma 2018). Species accumulation curves were cal- culated using the ‘vegan’ package (Oksanen et al. 2020). The following packages were used for handling and manipulating data: ‘tidyverse’ (Wickham et al. 2019), “dbplyr (Wickham et al. 2021), ‘lubridate’ (Grolemund and Wickham 2011), and ‘sampler’ (Baldassaro 2019). To create and assess models we used: ‘Ime4’ (Barton 2020), ‘ImerT- est’ (Kuznetsova et al. 2017), and ‘MuMIr’ (Bates et al. 2015) packages. The following packages were used for data visualisation: ‘tidyverse’ (Wickham et al. 2019), ‘cowplot’ (Wilke 2020), ‘ggalluvial’ (Brunson and Read 2020), ‘ggrepel’ (Slowikowski 2021), ‘gepubr’ (Kassambara 2020), and ‘scales’ (Wickham and Seidel 2022). The data un- derpinning the methods and analysis of this study have been deposited on the Figshare Repository at https://doi.org/10.6084/m9.figshare.22493944 (Maher et al. 2023). Results Overall richness, trade proportion, and detection rate From the 10,000 advertisements we examined (i.e., 5,000 each for the untargeted and targeted samples), we made 13,619 plant identifications (average c. 1.4 identifications per advertisement). We identified 1,777 unique plant taxa (Fig. 2a) of which 78 were declared plants prohibited to trade in the jurisdictions where they were advertised (c. 6% of declared plants). A further 77 declared plants were advertised legally in juris- A B 150 1800+ Se 1600 4 8 © 1400-7 o ~< o al © 12007 aoe o © 40007 © o _o} 2 800 4 ‘5 s Hl ye 50:5 = 600 g 400+ = = 2005 2 0 ) 0 2,000 4,000 6,000 8,000 10,000 0 1,000 2,000 3,000 4,000 5,000 Number of advertisments Number of advertisements Figure 2. Accumulation curves of plant taxa identified from sampling 10,000 online advertisements A accumulation curve of all plant taxa identified. There were 1,777 taxa observed from 10,000 advertise- ments B accumulation curves of declared plant taxa identified. The red line represents a targeted sample that utilised search terms to locate declared plant advertisements and the blue line represents an untarget- ed sample that did not use search terms (i.e., random sampling). There were 155 declared taxa identified in 1,415 detections of declared plants. 54 Jacob Maher et al. / NeoBiota 87: 45-72 (2023) dictions that do not prohibit their trade. This brought the overall number of declared plants traded to 155 taxa (c. 12.5% of all declared plants in Australia) (Fig. 2b). We did not observe any of the species accumulation curves approaching a clear limit (Fig. 2). From the 10,000 advertisements examined, we made 411 prohibited detections (from 374 advertisements) within 1,415 total declared detections (from 1,296 adver- tisements). From our untargeted sample, we found 59 prohibited advertisements (c. 1%) and 150 total declared advertisements (detection rate of 3%). In comparison, our targeted sample contained 328 prohibited advertisements (c. 7%) and 1,183 total de- clared advertisements (detection rate of c. 24%) (Fig. 3). New South Wales (NSW) and Victoria (Vic.) are the most populous jurisdictions in Australia (Australian Bureau of Statistics 2020) and had the greatest number of total declared advertisements. Western Australia (WA) declares the greatest number of plant taxa of any Australia jurisdiction (877 plant taxa) and had the greatest number of prohibited advertisements (Fig. 3). Prohibited | Tm Total declared advertisments (%) 3 6 9 12 15 advertisments (%) 18 20 22 24 26 28 Cc Jurisdiction Population Plants declared Australian Capital Territory (ACT) 431,484 288 New South Wales (NSW) 8,172,505 402 Northern Territory (NT) 246,561 372 Queensland (QlId) 5,194,879 354 South Australia (SA) 1,770,790 391 Tasmania (Tas.) 541,506 434 Victoria (Vic.) 6,661,736 361 Western Australia (WA) 2,670,241 877 Figure 3. The number of advertisements for declared plants detected on an e-commerce platform over a 12-month period. These detections were made from a sample of 5,000 advertisements that had been matched to search terms for declared plants (i.e., targeted sample) A the number of prohibited declared plant advertise- ments detected within the jurisdiction (i-e., prohibited in that jurisdiction, refer Fig. 2). The colour refers to the percentage of advertisements that were prohibited B the total number of declared plant advertisements detected in that jurisdiction that are declared anywhere in Australia. The colour refers to the percentage of advertisements that contained declared plants C the 2020 resident population (Population) and number of plant taxa declared in each jurisdiction (Plants declared). Population data was sourced from Australian Bureau of Statistics (2020). Online trade of invasive plants in Australia 55 Influence of trade prohibition on quantity and price The generalised linear mixed-effects models revealed no statistically significant effect on the quantity and price of declared plants between jurisdictions that prohibited trade and those that did not. The model for quantity had a p-value of 0.58 for the quantity coefficient, with a sample size of 1040, which covered 130 declared taxa (quantity coef- ficient estimate = -0.000266 + SE = 0.000479; t = -0.56; Rc2 = 0.32). The model for price had a p-value of 0.13 for the price coefficient, with a sample size of 652, covering 20 declared taxa (price coefhicient estimate = -6.25 + SE = 4.11; t= -1.52; Re2 = 0.24). For over 80% (104/130 taxa) of declared taxa analysed, the mean difference in the number of advertisements between prohibited and permitted jurisdictions was less than one advertisement (Fig. 4). The declared plants with the greatest mean differences were Drimia maitima (mean absolute difference c. 5 plants) which had higher quantities in 1.00 0.75 0.50 0.25 0.00 10 -8 6 4 =D 0 2 4 6 Mean difference in quantity (n advertisements) Proportion of adverti Figure 4. Distribution of the mean difference in the number of advertisements for declared plant taxa between prohibited and permitted jurisdictions. The black curve overlaying the histogram represents the cumulative distribution of mean differences in advertisement quantities. A positive mean difference trans- lates to comparatively more advertisements in prohibited jurisdictions and fewer in permitted jurisdictions. A negative mean difference translates to comparatively more advertisements in permitted jurisdictions and fewer in prohibited jurisdictions. The distribution represents 130 plant taxa and each bar represents one advertisement. We removed taxa that are declared in all jurisdictions and those with fewer than two adver- tisements in each legality category (i-e., prohibited or permitted) as there was nothing to compare against. 56 Jacob Maher et al. / NeoBiota 87: 45—72 (2023) prohibited jurisdictions, and Opuntia ficus-indica (mean absolute difference c. 11 plants) with higher quantities in permitted jurisdictions. We found far fewer advertisements for declared plants in the untargeted sample compared to the targeted sample (Table 2). Across jurisdictions the proportion of prohibited advertisements was c. 0.2—2% and total de- clared advertisements was c. 1-5% in the untargeted sample (Table 2). The highest pro- portion of prohibited advertisements was observed in South Australia (SA) and NSW for the untargeted sample. In comparison, the detection rate in the targeted sample rose to c. 3-15% for prohibited and c. 16-28% for total declared advertisements across jurisdictions (Table 2 and Fig. 3). The highest proportion of prohibited advertisements was observed in Northern Territory (NT), SA, and WA for the targeted sample (Table 2 and Fig. 3). The distribution of plant prices was similar across jurisdictions, typically ranging from $5 to $40 for a potted plant (Australian dollars; AUD) (Suppl. material 4). On average, prices were only $1.25 more in prohibited jurisdictions with 60% (12/20 taxa) of observed taxa having a mean price difference within $5 (Suppl. material 5). However, the sample size for the price model was greatly reduced compared to the quantity model, with only 20 declared plant taxa included. Most frequently traded declared plants and advertised uses The most frequently advertised declared plants were Opuntia cacti and aquatic weeds (Fig. 5). The declared plant with the greatest number of prohibited advertisements was Opuntia microdasys (bunny ears cactus) (Fig. 5b). Other Opuntia species were fre- quently traded, including Opuntia monacantha (drooping prickly pear) and Opuntia Table 2. Summary of advertisements for declared plants in Australia’s eight jurisdictions. Results are pre- sented from two samples collected across 12 months of e-commerce activity. The untargeted sample repre- sents a consistent number of plant advertisements sampled for each jurisdiction, based on the location of the seller. The targeted sample is a focused search for advertisements matching declared plant search terms, resulting in a variable number of advertisements sampled for each jurisdiction. The ‘Prohibited’ column indicates the count of advertisements (Ads) containing plants declared within the respective jurisdiction where the advertisement is located. The “Total Declared’ presents the number of advertisements (Ads) containing plants declared anywhere in Australia. The percentages (%) are calculated based on these ob- servations and the respective sample sizes, with darker colours for higher relative percentages. The sample sizes represent the total number of advertisements considered in each jurisdiction. Jurisdiction Untargeted Sample Targeted Sample Prohibited Total declared Sample Prohibited Total declared Sample Ads % Ads % size Ads % Ads % size Australian Capital Territory (ACT) New South Wales (NSW) 11 1031 Northern Territory (NT) 9 66 Queensland (Qld) 948 South Australia (SA) 13 539 ‘Tasmania (Tas) 308 Victoria (Vic) 921 Western Australia (WA) 2. 767 Total Online trade of invasive plants in Australia 57 ae | a Opuntia microdasys'- Zantedeschia aethiopica’- ian) Zantedeschia aethiopica?- Opuntia ficus-indica®- oo eee | Eichhornia crassipes*- Gazania spp..- 9 Opuntia monacantha’- Limnobium laevigatum®- ree | Opuntia spp. - Hedera helix’: Opuntia ficus-indica*- Opuntia microdasys'- | Limnobium laevigatum®- Orbea variegata®- _ Gazania spp.°- Lavandula stoechas '°- a Orbea variegata®- Eichhornia crassipes*- a Rubus fruticosus agg. '?- Azadirachta indica‘*- om 0 10 20 30 40 0 50 100 150 200 Number of Detections Number of Detections Figure 5. Invasive plants most frequently advertised on an e-commerce platform during a 12-month pe- riod. These plants are prohibited to trade in one or more Australian jurisdictions (i-e., declared plants) A the size of the declared plant photos is approximately scaled by their relative frequency in trade B lists the 10 de- clared plants that were most frequently advertised in jurisdictions where they are prohibited to trade (i-e., ad- vertised illegally) C lists the 10 most frequently advertised plants declared in any jurisdiction. The superscript numbers next to species names correspond to the plant photos. Photos are sourced from Getty Images and are credited to: (1) Boonsom, (2) TopPhotolmages, (3) Wjarek, (4) Igaguri_1, (5) Reginaldo Bergamo, (6) Jonnyjto, (7) ePhotocorp, (8) Radka Danailova, (9) Belizar73, (10) Membio, (11) Bdspnimage, (12) Paulfjs. 58 Jacob Maher et al. / NeoBiota 87: 45—72 (2023) ficus-indica (Indian fig). Aquatic weed species were particularly common, including Eichhornia crassipes (water hyacinth) and Limnobium laevigatum (Amazon frogbit). Zantedeschia aethiopica (arum lily), an invasive geophyte, had the highest total number of advertisements for a declared plant, and the second highest number of prohibited advertisements (Fig. 5). Other frequently detected invasive plants were Gazania spp. (gazanias), Hedera helix (English ivy), Lavandula stoechas (topped lavender), Rubus fru- ticosus (blackberry), Orbea variegata (carrion flower), and Azadirachta indica (neem) (Fig. 5). Limnobium laevigatum was an example of a highly traded declared species with a far greater number of detections in jurisdictions that did not declare it. We made 19 detections for L. laevigatum in three prohibited jurisdictions and 69 in five permitted jurisdictions. A complete list of all declared species found and the number of prohibited and total declared detections are provided in Suppl. material 6. We recorded the following eleven suggested uses for declared plants (Fig. 6): 1. Aquatic — filters and conditions water and provides habitat for aquatic animals (n= 72) 2. Decorative — floral arrangements, bonsai, and materials for craft projects (n = 32). 3. Groundcover — grows and covers ground well, may inhibit other plant growth or prevent erosion (n = 22). 4, Food-edible fruits, vegetables, herbs, spices, or advertised as a superfood (n = 17). Medicinal — provides medicinal benefit (n = 11). Screening — privacy screening, hedging, or a wind break (n = 10). Cosmetic — used for cosmetic purposes such as skin care (n = 4). Insectary — attracts pollinating insects (n = 4). Pon EGS 9. Insecticide — kills or repels insects (n = 3). 10. Air — provides oxygen and purifies air (n = 2). 11. Spiritual — incorporated into spiritual beliefs and practices (n = 1). Sellers explicitly mentioned uses for plants in only 148 of the 1,296 advertisements of declared plants (c. 11%; 50 taxa). The most advertised use was for aquatic purposes, which encompassed actions such as improving or maintaining water quality and provid- ing habitat for aquatic animals (n = 72). L. laevigatum was the declared plant most often advertised with a use, all of which were for aquatic purposes (Fig. 6). The invasive at- tributes of some plants interplayed with their proposed uses. For example, gazanias were advertised as groundcovers as they spread easily and form dense mats, and Ligustrum vulgare (privet), known for its dense vegetation, was promoted as a screening plant. A complete list of all declared species advertised with uses is provided in Suppl. material 7. Discussion Ornamental plant trade is the world’s leading pathway for invasive plant introductions and is greatly facilitated by internet e-commerce (Humair et al. 2015; Munakamwe and Constantine 2017; Peres et al. 2018; van Kleunen et al. 2018; Beaury et al. 2021). Online trade of invasive plants in Australia 59 Use Taxa 125 Azadirachta indica 100 Aquatic Ligustrum vulgare 2] — © 75 = £ Limnobium laevigatum 0) > - S (ra co) fom 2 Oenanthe javanica E 50 nee | _Oenanthe javanica Pistia stratiotes : Pyracantha coccinea Salvinia minima 25 Thalia geniculata Medicinal Zantedeschia aethiopica 0 Figure 6. Thirteen invasive plant taxa prohibited to trade (termed declared plants) that were most fre- quently advertised with a use. In total, 50 declared plant taxa had uses reported in advertisements. The number of advertisements is stratified by the promoted use for the plant. These uses were reported by traders and were not verified in this study. Our study represents the first investigation into the presence of the complete set of Australia’s declared invasive plants on e-commerce. On a single popular e-commerce website, we found hundreds of opportunities to purchase a wide variety of declared plants over the course of one year. This is despite the country’s strict biosecurity policies and a weed risk assessment that has been adopted by other countries (Gordon et al. 2008). Trade of invasive plants through e-commerce has been documented in other re- gions such as New Zealand (Derraik and Phillips 2010), the United States (US) (Maki and Galatowitsch 2004; Beaury et al. 2021) and European Union (EU) (Lenda et al. 2014; Humair et al. 2015). Australia shares similarities with the US and EU, having 60 Jacob Maher et al. / NeoBiota 87: 45—72 (2023) accessible e-commerce platforms and easily facilitated trade across jurisdictions with differing biosecurity regulations. Our findings contribute to this growing body of evi- dence calling attention to e-commerce as an invasion risk pathway that is establishing globally. In particular, we have quantified the risk of illegal online plant trade conduct- ed by individuals rather than commercial nurseries, which is a challenging aspect of e-commerce to monitor and regulate. We highlight the need to review our approaches to managing invasive species in the face of an increasingly interconnected world. The pace of the ornamental plant trade in Australia is increasing, where 2020 saw a record high number of plant sales in the nursery industry (Horticulture Innovation Australia 2021). Given this growth and the availability of invasive plants, online trade poses a serious invasion threat and demands greater scrutiny. Since declared plant taxa have already been determined as serious biosecurity concerns (i.e., declared in State/ Territory laws), we argue that monitoring and interception of this trade is certainly warranted and should continue (Munakamwe and Constantine 2017). Low detection rates emphasise the challenge of capturing and regulating this trade. Given that our species accumulation did not approach a limit, it is likely that we have not captured the full diversity of declared plants traded online. It should also be noted that our study focussed on a narrow group of invasive plants (i.e., those that are currently declared as illegal to trade). Beyond the declared plants there are likely many other non-regulated, invasive plant species being traded on these e-commerce platforms that may still cause environmental harm (Beaury et al. 2021). Additionally, we only studied one e-com- merce platform. A broader analysis of additional e-commerce platforms may reveal more declared invasive plant species that are available to the public. In addition to the prohibited trade, declared plants were widely advertised in ju- risdictions where they are currently permitted to trade. Just under half of the declared taxa and more than double the number of detections we found were located in the ju- risdictions that did not prohibit sale. Some of the most frequently traded declared spe- cies are only prohibited to trade in one or two jurisdictions, despite many being known to be invasive in permitted jurisdictions. Some examples of invasive populations in permitted jurisdictions include: Lavandula stoechas in SA (Nicholson 2006), Orbea variegata in NSW (Hamilton et al. 2013), and Limnobium laevigatum in Queensland (Bickel et al. 2022). L. laevigatum was a particularly concerning example traded to a much larger extent in jurisdictions that did not prohibit its trade. We made 19 detec- tions for L. daevigatum in prohibited jurisdictions and 69 in permitted jurisdictions. By using online trade data, we argue that jurisdictions should reconsider the risk of invasive species like L. Jaevigatum to determine if prohibition is warranted. A similar situation has been observed in the US (Beaury et al. 2021), another geographically large country with multiple states with their own governing legislations. Like in the US, we argue this type of trade can compromise the biosecurity of neighbouring ju- risdictions (Beaury et al. 2021). For example, we found NSW and Vic. traded large quantities of species declared in neighbouring jurisdictions. ‘This is especially concern- ing because the plant trade facilitates long-distance dispersal from plants mailed over long distances (Maki and Galatowitsch 2004). Despite the limitations of online trade Online trade of invasive plants in Australia 61 data, it is still a valuable resource to help identify species or areas of concern (Kikillus et al. 2012). Thus, we suggest future weed risk assessments utilise data collected from monitoring e-commerce to factor in trade of invasive plants as a risk factor, even if oc- curring in other jurisdictions. Incorporating this may lead jurisdictions to consider a nationally consistent approach to plant declarations, similar to other control programs which have benefited from cross-border coordination (Pluess et al. 2012). As long as the trade of invasive plants persists somewhere with a country, the risk of natural or human mediated dispersal into vulnerable landscapes will remain. While more consistent regulations among jurisdictions would provide the legal framework to address invasive plant trade, our results may suggest this is not a cure- all. We found that across declared plant taxa, there was no difference in the quantities of advertisements observed in prohibited and permitted jurisdictions. We also saw no significant effect on price, however our sample size was reduced to 20 declared taxa, making it difficult to draw a meaningful conclusion across all declared taxa traded. It is likely that jurisdictional regulations are reducing the total abundance of declared taxa in Australian plant trade, through compliance from traditional “brick-and-mortar” nurseries. It is important to note that the lack of effect on quantity we saw could be due to the limited size of our sample. Investigations across larger datasets, and across more e-commerce platforms, may reveal different results. However, if trade prohibition is not having an effect on the quantity of online trade, explanations from other plant trade studies may provide an answer. For one, sellers may perceive online trading of declared plants as low risk. This perception may be in part due to limited enforcement of e-commerce due to surveillance and legal challenges (Lavorgna and Sajeva 2021; Whitehead et al. 2021). Another reason may be a lack of awareness that these plants are invasive and that their trade is prohibited. Public awareness has been suggested by oth- er studies into invasive plant trade, reporting that people are often unaware, lack the ability to correctly identify plants, or are misinformed about relevant legislation rather than knowingly breaking the law (Derraik and Phillips 2010; Martin and Coetzee 2011; Munakamwe and Constantine 2017). We suggest implementing web scraping surveillance tools to improve enforcement and to enhance public knowledge through awareness campaigns which improve invasive species management (Novoa et al. 2017; Cordeiro et al. 2020; Li et al. 2021). Further, e-commerce platforms can also play a role in prevention and should be engaged as a biosecurity stakeholder. Specifically, in agreement with other studies of the illegal plant trade, we recommend that relevant governments coordinate with e-commerce websites to prevent illegal trade (Derraik and Phillips 2010; Munakamwe and Constantine 2017). For example, e-commerce websites could provide information to people creating plant advertisements, warning them of plants that cannot be sold and to help identify those plants. Given that plant trade is fundamentally human driven, we expected to observe a higher number of advertisements matching search terms and corresponding to de- clared plants in jurisdictions with larger populations. Consequently, in the targeted sample, we observed this trend with NSW and Vic. having the greatest number of total declared advertisements. Interestingly, NSW and Vic. also had the greatest propor- 62 Jacob Maher et al. / NeoBiota 87: 45—72 (2023) tion of total declared advertisements. However, in terms of prohibited advertisements, WA, SA, and NT had the highest proportions in the targeted sample. To explain this, we should consider the plants that jurisdictions have chosen to declare. Regulations are jurisdiction based, therefore differences in declarations arise between jurisdictions. WA declares the greatest number of plant taxa of any Australia jurisdiction (877 plant taxa), more than double that of the next highest jurisdiction. As a result, WA prohibits a larger proportion of Australia’s assemblage of declared plants. Complementary to this is that NT, SA, and WA declare highly traded declared species that other jurisdictions do not. Zantedeschia aethiopica is only declared in SA and WA, Opuntia ficus-indica is only declared in NT and WA, and Gazania spp. are only declared in SA. These species were frequently traded in SA and WA, thus the higher proportions are indicative of the regulations of these jurisdictions. However, NT prohibited advertisements were predominately for aquatic declared plants that are not exclusively declared in the juris- diction. Evidently this is a popular group of plants traded in the jurisdiction, one that may benefit from targeted management campaigns. We found that Opuntia cacti and aquatic invasive plants were among the most fre- quently traded declared plants. This is concerning given the historical extent of Opun- tia impact on the Australian environment (Freeman 1992), and the invasiveness of the traded aquatic weeds Eichhornia crassipes and L. laevigatum (Riches 2001; Tidwell and O’Donnell 2010; Villamagna and Murphy 2010). It is possible that some traits that aid their invasion success could also lend to their popularity in trade. Opuntia cacti are easily propagated from cuttings and will do so readily when discarded from gardens (Smith 2006; Smith et al. 2011). &. crassipes and L. laevigatum can also repro- duce vegetatively and in good conditions growers will quickly have an overabundance (Madsen and Morgan 2021; Prasetyo et al. 2021). This ease of excess could present sale as an attractive option to get rid of surplus plants, thus facilitating invasions. However, without further investigation into seller behaviour we cannot say how common this is. Similarly, it has been suggested that some Opuntia protective traits (e.g., spines and glochids) eventually lead owners to dispose of them. Smith et al. (2011) suggested that the irritating hairs (glochids) of Opuntia microdasys drive owners to dispose of the plants through dumping. We spoke with a compliance officer investigating Opuntia sales, who reported that sellers mention a desire to sell the plants in order to be rid of them (D. Swan 2021, pers. comm., 3 November). The high number of advertisements we observed of these taxa may indicate selling plants is an attractive alternative to dis- posal but this would require further investigation. We demonstrated that targeted searches using string matching was a more effec- tive means of detection than random sampling. We took a conservative approach by including common and generic names (e.g., pond plant) alongside scientific names in our effort to detect declared plants. Common and generic names are non-specific and can be shared by many plant species, contributing to a higher rate of false posi- tives. However, we believe this approach is necessary to reduce the chance of missing advertisements for invasive species. Image recognition technology could be employed to further increase detection rate (Di Minin et al. 2019). However, the accuracy of Online trade of invasive plants in Australia 63 image recognition is dependent on large, pre-identified image datasets and the quality of images provided (Xiong et al. 2021). The quality of images that we observed in ad- vertisements varied greatly in resolution and often had complex backgrounds, a feature known to hinder the accuracy of image recognition (Xiong et al. 2021). We propose that string matching and other natural language processing methods are a cost-effective means for the semi-automated detection of invasive plants on e-commerce platforms. The advertised uses for declared plants revealed some reasons why people de- sire them, which may complicate their management. We discovered a variety of uses advertised for declared plants, including food, medicine, cosmetics, and decoration (e.g., floral arrangements). However, the most commonly advertised uses fell into the ‘aquatic’ category, uses such as water-conditioning and providing habitat for aquatic pets. Perceived water-conditioning abilities could encourage people to introduce the plant into waterbodies (e.g., ponds and dams), risking dispersal into the surrounding environment. For example, we found E. crassipes traded which has been known to be intentionally introduced into waterbodies to help prevent algal blooms (Villamagna and Murphy 2010). It is important to consider people’s intended use of an invasive plant because prevention is often more of a cultural challenge than biological (Pfeiffer and Voeks 2008). Understanding the public’s desire for an invasive plant could help to tailor education campaigns or promote non-invasive alternatives. It is also important that public attitudes are understood to establish collaborative efforts between invested communities and policymakers, which will lead to optimal social and biosecurity out- comes (Virtue et al. 2004; Head 2017). Conclusion We observed the prohibited advertisement of invasive plants online in all Australian ju- risdictions. This online trade creates many opportunities for the public to purchase and spread declared invasive plants around the country. As it stands, laws prohibiting the trade of declared plants have not halted prohibited advertisements of declared plants on public e-commerce. We suggest enhancing detection methods of illegal trade using web scraping techniques to improve enforcement. Jurisdictions should also focus on educating the public that certain plants are prohibited to trade while considering the desire that people have for these plants to help promote safe alternatives. Cooperation should be sought from e-commerce websites to prevent instances of illegal trade being facilitated on their platforms. For now, monitoring e-commerce is still needed and we have demonstrated that web-scraping is an effective tool. Data collected from moni- toring e-commerce could also be utilised in future weed risk assessments with online availability incorporated as a risk factor. Beyond surveillance, jurisdictions should seek to better align the taxa they choose to regulate as the existing legal disparities could contribute to the persistence of invasive species being distributed within a country. Australia’s biosecurity, and that of other countries and regions, would benefit from more coordinated approaches to controlling the online trade of invasive species. 64 Jacob Maher et al. / NeoBiota 87: 45-72 (2023) Acknowledgements We, the authors, acknowledge we are all living and working on colonised land. Jacob Maher, Lisa Wood, Charlotte R. Lassaline, and Phillip Cassey live and work on Kaurna land. Oliver C. Stringham and Stephanie Moncayo live and work on Lenape land. John Virtue lives and works on Peramangk land. In all instances, we acknowledge and recognize the longstanding significance of these lands for these nations. The Kaurna, Lenape, and Peramangk people were violently displaced as a result of European settler colonialism yet remain closely connected with these lands and are their rightful stew- ards. We respect their custodianship of the land, value their past, present, and ongoing connection to the land and their cultural beliefs. This work was supported by funding from the Australian Research Training Pro- gram and the Centre for Invasive Species Solutions (P01-W-003: Biosecurity surveil- lance of e-commerce and other online platforms for illegal trade in declared plants). References Arianoutsou M, Bazos I, Christopoulou A, Kokkoris Y, Zikos A, Zervou S, Delipetrou P, Car- doso AC, Deriu I, Gervasini E, Tsiamis K (2021) Alien plants of Europe: Introduction pathways, gateways and time trends. Peer] 9: e11270. https://doi.org/10.7717/peerj.11270 Australian Bureau of Statistics (2020) National, state and territory population, December 2020. https://population.gov.au/data-and-forecasts/key-data-releases/national-state-and- territory-population-december-2020 [Accessed 12 September 2022] Australian Bureau of Statistics (2021) Digital boundary files Australian Statistical Geography Standard. https://www.abs.gov.au/statistics/standards/australian-statistical-geography- standard-asgs-edition-3/jul202 1-jun2026/access-and-downloads/digital-boundary-files [Accessed 31 August.2020] Australian National Audit Office (2014) Audit Report No.42 2013-14 Screening of Interna- tional Mail. Australian Government, Canberra, ACT, 17 pp. Australian National Herbarium (2023) Australian Plant Census. https://biodiversity.org.au/ nsl/services/search/taxonomy [Accessed 23 Feb 2023] Baldassaro M (2019) sampler: Sample Design, Drawing & Data Analysis Using Data Frames. R package version 0.2.4. https://CRAN.R-project.org/package=sampler Barton K (2020) MuMIn: Multi-Model Inference. R package version 1.43.17. https:// CRAN.R-project.org/packages=MuMIn Bates D, Machler M, Bolker B, Walker S (2015) Fitting Linear Mixed-Effects Models Using Ime4. Journal of Statistical Software 67(1): 1-48. https://doi.org/10.18637/jss.v067.i01 Beaury EM, Patrick M, Bradley BA (2021) Invaders for sale: The ongoing spread of invasive species by the plant trade industry. Frontiers in Ecology and the Environment 19(10): 550-556. https://doi.org/10.1002/fee.2392 Bickel TO, Farahani B, Perrett C, Xu J, Vitelli JS (2022) Control of the emerging aquatic weed Amazon frogbit with flumioxazin. In: Melland R, Brodie C, Emms J, Feuerherdt L, Ivory Online trade of invasive plants in Australia 65 S, Potter S (Eds) 22 Australasian Weeds Conference. Weed Management Society of South Australia Inc., Adelaide, 110 pp. Bradshaw CJA, Hoskins AJ, Haubrock PJ, Cuthbert RN, Diagne C, Leroy B, Andrews L, Page B, Cassey P, Sheppard AW, Courchamp F (2021) Detailed assessment of the report- ed economic costs of invasive species in Australia. NeoBiota 67: 511-550. https://doi. org/10.3897/neobiota.67.58834 Broadhurst L, Coates D (2017) Plant conservation in Australia: Current directions and future challenges. Plant Diversity 39(6): 348-356. https://doi.org/10.1016/j.pld.2017.09.005 Brunson JC, Read QD (2020) ggalluvial: Alluvial Plots in ‘ggplot2’. R package version 0.12.3. http://corybrunson.github.io/ggalluvial/ Cordeiro B, Marchante H, Castro P, Marchante E (2020) Does public awareness about invasive plants pays off? An analysis of knowledge and perceptions of environmentally aware citi- zens in Portugal. Biological Invasions 22(7): 2267-2281. https://doi.org/10.1007/s10530- 020-02247-z Derraik JGB, Phillips S (2010) Online trade poses a threat to biosecurity in New Zealand. Bio- logical Invasions 12(6): 1477-1480. https://doi.org/10.1007/s10530-009-9595-0 Di Minin E, Fink C, Hiippala T, Tenkanen H (2019) A framework for investigating illegal wildlife trade on social media with machine learning. Conservation Biology 33(1): 210- 213. https://doi.org/10.1111/cobi.13104 Dodd AJ, Burgman MA, McCarthy MA, Ainsworth N (2015) The changing patterns of plant naturalization in Australia. Diversity & Distributions 21(9): 1038-1050. https://doi. org/10.1111/ddi.12351 Duncan RP (2021) Time lags and the invasion debt in plant naturalisations. Ecology Letters 24(7): 1363-1374. https://doi.org/10.1111/ele.13751 Faulkner KT, Robertson MP, Rouget M, Wilson JRU (2016) Understanding and managing the introduction pathways of alien taxa: South Africa as a case study. Biological Invasions 18(1): 73-87. https://doi.org/10.1007/s10530-015-0990-4 Freeman DB (1992) Prickly pear menace in eastern Australia 1880-1940. Geographical Re- view 82(4): 413-429. https://doi.org/10.2307/215199 Gallagher RV, Leishman MR (2014) Invasive plants and invaded ecosystems in Australia: im- plications for biodiversity. In: Stow A, Maclean N, Holwell G (Eds) Austral Ark: The State of Wildlife in Australia and New Zealand. Cambridge University Press, 105-133. https:// doi.org/10.1017/CBO9781139519960.008 GBIF (2021) The Global Biodiversity Information Facility (2021) What is GBIF? https://www. ebif.org/what-is-gbif [Accessed 24-08-2021] Giltrap N, Eyre D, Reed P (2009) Internet sales of plants for planting — an increasing trend and threat?1. Bulletin OEPP. EPPO Bulletin. European and Mediterranean Plant Protection Organisation 39(2): 168-170. https://doi.org/10.1111/j.1365-2338.2009.02283.x Gordon DR, Onderdonk DA, Fox AM, Stocker RK (2008) Consistent accuracy of the Aus- tralian weed risk assessment system across varied geographies. Diversity & Distributions 14(2): 234-242. https://doi.org/10.1111/j.1472-4642.2007.00460.x Grolemund G, Wickham H (2011) Dates and Times Made Easy with {lubridate}. Journal of Statistical Software 40(3): 1-25. https://doi.org/10.18637/jss.v040.i03 66 Jacob Maher et al. / NeoBiota 87: 45—72 (2023) Hamilton MA, Turner PJ, Wurst D (2013) Carrion flower, a novel invasive species in NSW. In: Wu H (Ed.) Proceedings of the 17° Biennial NSW Weeds Conference. 133-136. Head L (2017) The social dimensions of invasive plants. Nature Plants 3(6): e17075. https:// doi.org/10.1038/nplants.2017.75 Horticulture Innovation Australia (2021) Supporting the horticulture sector now and into the future, Annual Report 2020/21. https://www.horticulture.com.au/globalassets/hort-inno- vation/corporate-documents/hort-innovation-company-annual-report-2020-21.pdf Humair F, Humair L, Kuhn E Kueffer C (2015) E-commerce trade in invasive plants. Conser- vation Biology 29(6): 1658-1665. https://doi.org/10.1111/cobi.12579 Kassambara A (2020) ggpubr: ‘ggplot2’ Based Publication Ready Plots. R package version 0.4.0. https://CRAN.R-project.org/package=ggpubr Keller RP, Lodge DM, Finnoff DC (2007) Risk assessment for invasive species produces net bioeconomic benefits. Proceedings of the National Academy of Sciences of the United States of America 104(1): 203-207. https://doi.org/10.1073/pnas.0605787 104 Kikillus KH, Hare KM, Hartley S (2012) Online trading tools as a method of estimating propagule pressure via the pet-release pathway. Biological Invasions 14(12): 2657-2664. https://doi.org/10.1007/s10530-012-0262-5 Kuznetsova A, Brockhoff PB, Christensen RHB (2017) ImerTest Package: Tests in Linear Mixed Effects Models. Journal of Statistical Software 82(13): 1-26. https://doi.org/10.18637/jss. v082.i113 Lavorgna A, Sajeva M (2021) Studying Illegal Online Trades in Plants: Market Characteristics, Or- ganisational and Behavioural Aspects, and Policing Challenges. European Journal on Crimi- nal Policy and Research 27(4): 451-470. https://doi.org/10.1007/s10610-020-09447-2 Lavorgna A, Middleton SE, Pickering B, Neumann G (2020) FloraGuard: Tackling the Online Illegal Trade in Endangered Plants Through a Cross-Disciplinary ICT-Enabled Methodology. Journal of Contemporary Criminal Justice 36(3): 428-450. https://doi. org/10.1177/1043986220910297 Lenda M, Skérka P, Knops JMH, Moron D, Sutherland WJ, Kuszewska K, Woyciechowski M (2014) Effect of the internet commerce on dispersal modes of invasive alien species. PLoS ONE 9(6): e99786. https://doi.org/10.137 1/journal.pone.0099786 Li Y, Liu X, Zeng H, Zhang J, Zhang L (2021) Public education improves farmers knowledge and management of invasive alien species. Biological Invasions 23(6): 2003-2017. https:// doi.org/10.1007/s10530-021-02486-8 Madsen JD, Morgan CM (2021) Water temperature controls the growth of waterhyacinth and South American sponge plant. Journal of Aquatic Plant Management 59s: 28-32. Magalhaes AL, Avelar V (2012) Illegal trade on non-native amphibians and reptiles in southeast Brazil: The status of e-commerce. Phyllomedusa 11(2): 155-160. https://doi. org/10.11606/issn.23 16-9079.v11i2p155-160 Maher J, Stringham O, Moncayo S, Wood L, Lassaline C, Virtue J, Cassey P (2023) Illegal online trade of invasive plants in Australia. figshare. https://doi.org/10.6084/m9.figshare.22493944 Maki K, Galatowitsch S (2004) Movement of invasive aquatic plants into Minnesota (USA) through horticultural trade. Biological Conservation 118(3): 389-396. https://doi. org/10.1016/j.biocon.2003.09.015 Online trade of invasive plants in Australia 67 Martin GD, Coetzee JA (2011) Pet stores, aquarists and the internet trade as modes of intro- duction and spread of invasive macrophytes in South Africa. Water S.A. 37(3): 371-380. https://doi.org/10.4314/wsa.v37i3.68488 Morgan A, Carthew SM, Sedgley M (2002) Breeding system, reproductive efficiency and weed potential of A. baileyana. Australian Journal of Botany 50(3): 357-364. https://doi. org/10.1071/BT01088 Munakamwe Z, Constantine A (2017) Illegal Online Trade of Noxious Weeds in Australia Monitoring and regulation E-Commerce. 19 NSW Biennial Weeds Conference Papers. The Weed Society of New South Wales Inc., Armidale NSW, 58-62. Nakagawa S, Schielzeth H (2013) A general and simple method for obtaining R2 from gen- eralized linear mixed-effects models. Methods in Ecology and Evolution 4(2): 133-142. https://doi.org/10.1111/j.2041-210x.2012.00261.x Nicholson H (2006) Conflicting values of topped lavender Lavandula stoechas L.: the essential oil on a complex issue. In: Preston C, Watts J, Crossman N (Eds) 15% Australian Weeds Conference: Papers and Proceedings. Adelaide, South Australia, 191-194. Novoa A, Dehnen-Schmutz K, Fried J, Vimercati G (2017) Does public awareness increase sup- port for invasive species management? Promising evidence across taxa and landscape types. Biological Invasions 19(12): 3691-3705. https://doi.org/10.1007/s10530-017-1592-0 Nursery & Garden Industry Australia (2009) Grow Me Instead. https://www.growmeinstead. com.au/ [Accessed 23"! of September 2020] O’Loughlin LS, Green PT, Morgan JW (2015) The rise and fall of Leptospermum laevigatum: Plant community change associated with the invasion and senescence of a range-expand- ing native species. Applied Vegetation Science 18(2): 323-331. https://doi.org/10.1111/ avse. L213] Oksanen J, Blanchet FG, Friendly M, Kindt R, Legendre PR. McGlinn D, Minchin PR, O’Hara RB, Simpson GL, Solymos P, Stevens MHH, Szoecs E, Wagner H (2020) vegan: Communi- ty Ecology Package. R package version 2.5-7. https://CRAN.R-project.org/package=vegan Parsons WT, Cuthbertson EG (2001) Noxious Weeds of Australia. CSIRO Publishing, Colling- wood, 698 pp. Pebesma E (2018) Simple features for R: Standardized support for spatial vector data. The R Journal 10(1): 439-446. https://doi.org/10.32614/RJ-2018-009 Peres CK, Lambrecht RW, Tavares DA, Chiba de Castro WA (2018) Alien express: The threat of aquarium e-commerce introducing invasive aquatic plants in Brazil. Perspectives in Ecol- ogy and Conservation 16(4): 221-227. https://doi.org/10.1016/j.pecon.2018.10.001 Pfeiffer JM, Voeks RA (2008) Biological invasions and biocultural diversity: Linking eco- logical and cultural systems. Environmental Conservation 35(04): 281-293. https://doi. org/10.1017/S0376892908005 146 Pheloung PC, Williams PA, Halloy SR (1999) A weed risk assessment model for use as a bios- ecurity tool evaluating plant introductions. Journal of Environmental Management 57(4): 239-251. https://doi.org/10.1006/jema.1999.0297 Pluess T, Jarosik V, PySek P, Cannon R, Pergl J, Breukers A, Bacher S (2012) Which Factors Affect the Success or Failure of Eradication Campaigns against Alien Species? PLoS ONE 7(10): e48157. https://doi.org/10.1371/journal.pone.0048157 68 Jacob Maher et al. / NeoBiota 87: 45—72 (2023) Prasetyo S, Anggoro S, Soeprobowati TR (2021) The growth rate of Water Hyacinth (Eichhornia crassipes (Mart.) Solms) in Rawapening Lake, Central Java. Journal of Ecological Engineer- ing 22(6): 222-231. https://doi.org/10.12911/22998993/137678 PySek P, Hulme PE, Simberloff D, Bacher S, Blackburn TM, Carlton JT, Dawson W, Essl FE Foxcroft LC, Genovesi P, Jeschke JM, Kithn I, Liebhold AM, Mandrak NE, Meyerson LA, Pauchard A, Pergl J, Roy HE, Seebens H, van Kleunen M, Vila M, Wingfield MJ, Richardson DM (2020) Scientists’ warning on invasive alien species. Biological Reviews of the Cambridge Philosophical Society 95(6): 1511-1534. https://doi.org/10.1111/ brv. 12627 Python Software Foundation (2020) Python Programming Language. Version 3.8.1. https:// www.python.org/ R Core Team (2022) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna. https://www.R-project.org/ Reitz K (2020) requests. https://pypi.org/project/requests/ [Accessed 1 February 2020] Richardson L (2020) beautifulsoup4. https://pypi.org/project/beautifulsoup4/ [Accessed 1 February 2020] Riches CR (2001) The world’s worst weeds. Proceedings of an International Symposium. Brit- ish Crop Protection, Hilton Brighton Metropole Hotel, UK, 118 pp. Robinson D, Hayes A, Couch S (2021) broom: Convert Statistical Objects into Tidy Tibbles. R package version 0.7.9. https://CRAN.R-project.org/package=broom Rojas-Sandoval J, Ferrufino-Acosta L, Flores R, Galan P, Lopez O, MacVean A, Rodriguez Delcid D, Ruiz Y, Chacén-Madrigal E (2022) Flora introduced and naturalized in Central America. Alien Floras and Faunas 18. Biological Invasions 25: 1007-1021. https://doi. org/10.1007/s10530-022-02968-3 Rose S, Fairweather PG (1997) Changes in floristic composition of urban bushland invaded by Pittosporum undulatum in Northern Sydney, Australia. Australian Journal of Botany 45(1): 123-149. https://doi.org/10.1071/BT95058 Rudis B, Embrey B (2020) pluralize: Pluralize and “Singularize’ Any (English) Word. R package version 0.2.0. https://CRAN.R-project.org/package=pluralize Chamberlain ES, Foster Z, Arendsee Z, Boettiger C, Ram K, et al. (2013) taxize — taxonomic search and retrieval in R. R package version 0.9.98. https://github.com/ropensci/taxize Selenium Main Repository (2020) selenium. https://www.selenium.dev/ [Accessed 1 February 2020} Shepherd RCH, Richardson RG, Richardson FJ (2001) Plants of importance to Australia: a checklist. R.G. & FJ. Richardson, Meredith, 358 pp. Simberloff D, Martin J-L, Genovesi P, Maris V, Wardle DA, Aronson J, Courchamp EF, Galil B, Garcia-Berthou E, Pascal M, PySek P, Sousa R, Tabacchi E, Vila M (2013) Impacts of biological invasions: What's what and the way forward. Trends in Ecology & Evolution 28(1): 58-66. https://doi.org/10.1016/j.tree.2012.07.013 Slowikowski K (2021) ggrepel: Automatically Position Non-Overlapping Text Labels with ‘ge- plot2’. R package version 0.9.1, https://CRAN.R-project.org/package=ggrepel Smith GF (2006) Cacti and Succulents: A Complete Guide to Species, Cultivation and Care. Ball Publishing, Batavia, 160 pp. Online trade of invasive plants in Australia 69 Smith GF, Figueiredo E, Boatwright JS, Crouch NR (2011) South Africa's ongoing Opun- tia Mill. (Cactaceae) problem: The case of Opuntia microdasys (Lehm.) Pfeiff. Bradleya 2011(29): 73-78. https://doi.org/10.25223/brad.n29.2011.a9 Stoett P, Omrow DA (2021) Floral Transnational Ecoviolence. Spheres of Transnational Ecovi- olence: Environmental Crime, Human Security, and Justice. Springer International Pub- lishing, Cham, 127-154. https://doi.org/10.1007/978-3-030-58561-7_5 Stringham O, Toomes A, Kanishka A, Mitchell L, Heinrich S, Ross J, Cassey P (2020) A guide to using the Internet to monitor and quantify the wildlife trade. Conservation Biology 35: 113-1139. https://doi.org/10.1111/cobi.13675 Stringham OC, Moncayo S, Hill KGW, Toomes A, Mitchell L, Ross JV, Cassey P (2021) Text classification to streamline online wildlife trade analyses. PLoS ONE 16(7): e0254007. https://doi.org/10.1371/journal.pone.0254007 Tidwell T, O’Donnell M (2010) Plant pest diagnostic center annual report. California Depart- ment of Food and Agriculture. California, 14—23. van Kleunen M, Essl F, Pergl J, Brundu G, Carboni M, Dullinger S, Early R, Gonzalez-Moreno P, Groom QJ, Hulme PE, Kueffer C, Ktihn I, Maguas C, Maurel N, Novoa A, Parepa M, Pysek PB, Seebens H, Tanner R, Touza J, Verbrugge L, Weber E, Dawson W, Kreft H, Weigelt P, Winter M, Klonner G, Talluto MV, Dehnen-Schmutz K (2018) The changing role of ornamental horticulture in alien plant invasions. Biological Reviews of the Cam- bridge Philosophical Society 93(3): 1421-1437. https://doi.org/10.1111/brv.12402 Villamagna A, Murphy B (2010) Ecological and socio-economic impacts of invasive water hyacinth (Eichhornia crassipes): A review. Freshwater Biology 55(2): 282-298. https://doi. org/10.1111/j.1365-2427.2009.02294.x Virtue JG, Bennett SJ, Randall RP (2004) Plant introductions in Australia: how can we resolve ‘weedy’conflicts of interest? In: Sindel BM, Johnson SB (Eds) Proceedings of the 14° Aus- tralian Weeds Conference. 42—48. Virtue J, Cunningham D, Hanson C, Hosking J, Miller I, Panetta F, Phleoung P, Randall R, Timmins S, Walton C, Weiss J, Williams P (2006) HB 294-2006 National Post-Border, Weed Risk Management Protocol. Standards Australia. International Ltd., Sydney, Standards New Zealand, Auckland and CRC Australian Weed Management, Adelaide, 76 pp. Walton CS (2001) Implementation of a permitted list approach to plant introductions. Weed Risk Assessment. CSIRO Publishing, Collingwood, 93-100. Ward M, Carwardine J, Yong CJ, Watson JEM, Silcock J, Taylor GS, Lintermans M, Gillespie GR, Garnett ST, Woinarski J, Tingley R, Fensham RJ, Hoskin CJ, Hines HB, Roberts JD, Kennard MJ, Harvey MS, Chapple DG, Reside AE (2021) A national-scale dataset for threats impacting Australia’s imperiled flora and fauna. Ecology and Evolution 11(17): 11749-11761. https://doi.org/10.1002/ece3.7920 Weber E, Sun S-G, Li B (2008) Invasive alien plants in China: Diversity and ecological insights. Biological Invasions 10(8): 1411-1429. https://doi.org/10.1007/s10530-008-9216-3 Whitehead D, Cowell CR, Lavorgna A, Middleton SE (2021) Countering plant crime online: Cross-disciplinary collaboration in the FloraGuard study. Forensic Science International. Animals and Environments 1: e100007. https://doi.org/10.1016/j.fsiae.202 1.100007 70 Jacob Maher et al. / NeoBiota 87: 45—72 (2023) Wickham H (2019) stringr: Simple, Consistent Wrappers for Common String Operations. R package version 1.4.0. https://CRAN.R-project.org/package=stringr Wickham H, Miiller KR, Special Interest Group on Databases [R-SIG-DB] (2022) DBI: R Database Interface. R package version 1.1.3, https://CRAN.R-project.org/package=DBI Wickham H, Seidel D (2022) scales: Scale Functions for Visualization. R package version 1.2.1. https://CRAN.R-project.org/package=scales Wickham H, Averick M, Bryan J, Chang W, McGowan LDA, Francois R, Grolemund G, Hayes A, Henry L, Hester J, Kuhn M, Pedersen TL, Miller E, Bache SM, Miiller K, Ooms J, Robinson D, Seidel DP, Spinu V, Takahashi K, Vaughan D, Wilke C, Woo K, Yutani H (2019) Welcome to the {tidyverse}. Journal of Open Source Software 4(43): €1686. https:// doi.org/10.21105/joss.01686 Wickham H, Girlich M, Ruiz E (2021) dbplyr: A ‘dplyr’ Back End for Databases. R package version 2.1.1. https://CRAN.R-project.org/package=dbplyr Wilke CO (2020) cowplot: Streamlined Plot Theme and Plot Annotations for ‘ggplot2’. R package version 1.1.1. https://CRAN.R-project.org/package=cowplot Xiong J, Yu D, Liu S, Shu L, Wang X, Liu Z (2021) A review of plant phenotypic image recog- nition technology based on deep learning. Electronics 10(1): e81. https://doi.org/10.3390/ electronics10010081 Supplementary material | This table details the relevant legislation identifying declared plants in each jurisdiction Author: Jacob Maher Data type: Legislation references Copyright notice: This dataset is made available under the Open Database License (http://opendatacommons.org/licenses/odbl/1.0/). The Open Database License (ODDbL) is a license agreement intended to allow users to freely share, modify, and use this Dataset while maintaining this same freedom for others, provided that the original source and author(s) are credited. Link: https://doi.org/10.3897/neobiota.87.104472.suppl1 Supplementary material 2 Short-list of invasive plants used for surveying candidate Australian websites Author: Jacob Maher Data type: species list (PDF file) Copyright notice: This dataset is made available under the Open Database License (http://opendatacommons.org/licenses/odbl/1.0/). The Open Database License (ODDbL) is a license agreement intended to allow users to freely share, modify, and use this Dataset while maintaining this same freedom for others, provided that the original source and author(s) are credited. Link: https://doi.org/10.3897/neobiota.87.104472.suppl2 Online trade of invasive plants in Australia 71 Supplementary material 3 List of search term exceptions used to remove the majority of false positives in target sample dataset Author: Jacob Maher Data type: Search term exceptions (PDF file) Copyright notice: This dataset is made available under the Open Database License (http://opendatacommons.org/licenses/odbl/1.0/). The Open Database License (ODDbL) is a license agreement intended to allow users to freely share, modify, and use this Dataset while maintaining this same freedom for others, provided that the original source and author(s) are credited. Link: https://doi.org/10.3897/neobiota.87.104472.suppl3 Supplementary material 4 The price of plants advertised online in Australia from a random sample of 625 advertisements from each jurisdiction Author: Jacob Maher Data type: Boxplot (PDF file) Copyright notice: This dataset is made available under the Open Database License (http://opendatacommons.org/licenses/odbl/1.0/). The Open Database License (ODDbL) is a license agreement intended to allow users to freely share, modify, and use this Dataset while maintaining this same freedom for others, provided that the original source and author(s) are credited. Link: https://doi.org/10.3897/neobiota.87.104472.suppl4 Supplementary material 5 Distribution of the mean difference in price for declared plant taxa between pro- hibited and permitted jurisdictions Author: Jacob Maher Data type: image (PDF file) Copyright notice: This dataset is made available under the Open Database License (http://opendatacommons.org/licenses/odbl/1.0/). The Open Database License (ODDbL) is a license agreement intended to allow users to freely share, modify, and use this Dataset while maintaining this same freedom for others, provided that the original source and author(s) are credited. Link: https://doi.org/10.3897/neobiota.87.104472.suppl5 ip Jacob Maher et al. / NeoBiota 87: 45—72 (2023) Supplementary material 6 Total number of detections for invasive plants which are prohibited to trade in at least one Australian jurisdictions Author: Jacob Maher Data type: Table: Species detections (PDF file) Copyright notice: This dataset is made available under the Open Database License (http://opendatacommons.org/licenses/odbl/1.0/). The Open Database License (ODDbL) is a license agreement intended to allow users to freely share, modify, and use this Dataset while maintaining this same freedom for others, provided that the original source and author(s) are credited. Link: https://doi.org/10.3897/neobiota.87.104472.suppl6 Supplementary material 7 The number of observations for plant taxa prohibited to trade (i.e., declared plants) that were advertised with uses by traders Author: Jacob Maher Data type: table: Species detections with uses (PDF file) Copyright notice: This dataset is made available under the Open Database License (http://opendatacommons.org/licenses/odbl/1.0/). The Open Database License (ODDbL) is a license agreement intended to allow users to freely share, modify, and use this Dataset while maintaining this same freedom for others, provided that the original source and author(s) are credited. Link: https://doi.org/10.3897/neobiota.87.104472.suppl7