Geography of WRAs

We found 20 unique national-level WRAs used by a total of 81 countries (Table 2). Twelve WRAs were used by a single country; eight were used by multiple countries. We were unable to find WRAs for 159 countries, only 4 of which (Bahrain, Cabo Verde, Colombia, Nigeria) confirmed via email that they do not have a WRA. It is concerning that we were only able to locate WRAs for 81 of 240 countries (34%). Most invasive plants are introduced intentionally as ornamental plants (Beaury et al. Reference Beaury, Patrick and Bradley2021b; Reichard and White Reference Reichard and White2001), making WRAs and subsequent regulations highly effective for proactively preventing invasions when they are used to consistently screen new species before import. WRAs and border control policies are cost-effective (Keller et al. Reference Keller, Lodge and Finnoff2007), particularly when considering the high economic costs of invasive plants (Diagne et al. Reference Diagne, Leroy, Vaissière, Gozlan, Roiz, Jarić, Salles, Bradshaw and Courchamp2021). Moreover, proactive exclusion of invasive plants prevents substantial negative impacts on native species and ecosystems (Roy et al. Reference Roy, Pauchard, Stoett and Renard Truong2024). While we only searched for WRAs in English, the spatial pattern of border control measures (or lack thereof) is consistent with a previous analysis of other proactive invasive species policies based on U.N. Convention on Biological Diversity implementation documents (Early et al. Reference Early, Bradley, Dukes, Lawler, Olden, Blumenthal, Gonzalez, Grosholz, Ibañez, Miller, Sorte and Tatem2016). Nevertheless, we acknowledge that it is possible that our methods excluded some WRAs written in languages other than English.

Table 2. National-scale weed risk assessments (WRAs) and their corresponding scores (number of “minimum standards” met out of 24; see Table 1) ^a . State-scale scores can be found in Appendix S1.

^a More details about each WRA and its score can be found in Supplementary Appendix S2. PDFs of each WRA can be found on GitHub at the following link: https://github.com/wpfadenhauer/Global-WRAs/tree/main/WRA_Documents.

Encouragingly, our analysis identifies opportunities for sharing WRA policies across neighboring countries—at least one country on every continent except Antarctica uses a WRA (Figure 1). Seven WRAs applied to countries in Europe; three WRAs applied to countries in Africa; three WRAs applied to countries in North America (including the Caribbean); two WRAs applied to countries in each of Asia, South America, and Oceania; and one WRA applied to Australia. The most-represented continents were South America (9/14 countries used a WRA; 64%), Europe (28/51; 55%), Oceania (14/26; 54%), and North America/Caribbean (20/41; 49%). In contrast, much smaller percentages of the countries in Asia (6/53; 11%) and Africa (4/60; 7%) used WRAs. The lack of WRAs in Asia and Africa is consistent with existing biases in invasive plant ecology, which illustrate that most studies focus on a subset of species that are harmful in wealthier nations (Laginhas et al. Reference Laginhas, Fertakos and Bradley2022; Pyšek et al. Reference Pyšek, Richardson, Pergl, Jarošík, Sixtová and Weber2008). Pyšek et al. (Reference Pyšek, Richardson, Pergl, Jarošík, Sixtová and Weber2008) suggested that the lack of data on invasive species in many locations is a result of an overall lack of funding invested in research by these countries. Lack of WRAs may also be related to political will and maturity of an import assessment process. For example, invasion scientists in both India (Sreekanth et al. Reference Sreekanth, Pawar, Chethan, Singh, Sondhia, Chander and Singh2022) and China (Wang et al. Reference Wang, Huang, Zhang, Li, He, Wen, Yin and Liang2024) have developed WRAs but they have not yet been implemented as regulations.

Figure 1. Weed risk assessment (WRA) standards met by nations and states. The global map (top panel) displays the number of minimum standards (out of 24) met by 20 unique WRAs used by 81 countries. The global map uses a Robinson projection and is at a scale of 1:200,000,000. The continental maps (bottom panels) display WRAs used by 38 states or provinces (37 total WRAs). The continental maps use Albers equal area conic projections and are at a scale of 1:150,000,000. We were unable to find WRAs for countries and states/provinces shown in white.

Similar patterns of lack of scientific capacity and/or lack of political will may also exist within countries. In the United States, state invasive plant councils often lack the necessary funding to conduct risk assessments for all potentially harmful species. Thus, increasing the consistency of existing WRAs across national and state/provincial borders could enable sharing of completed risk assessments, which would stretch limited resources further by reducing duplicate effort while also potentially increasing consistency in regulated species lists (Bradley et al. Reference Bradley, Beaury, Fusco, Munro, Brown-Lima, Coville, Kesler, Olmstead and Parker2022). Similarly, several states that have yet to implement a WRA or regulate invasive plants could adopt protocols from neighbors.

Countries located in tropical regions were particularly unlikely to have national WRAs. Notably, WRAs were lacking across tropical regions of Central America, Central Africa, and Southeast Asia. A recent analysis of plant invasion risk suggested that the tropics are particularly susceptible to invasions and that the low number of previously reported invasions in the tropics was more likely due to fewer introductions rather than lack of susceptibility (Pfadenhauer and Bradley Reference Pfadenhauer and Bradley2024). This analysis also noted that WRAs are likely to be most useful in areas such as these, with high ratios of invasive to established plant species. The lack of WRA protocols coupled with elevated risk to tropical regions highlights a major gap in proactive invasive plant policy and management at a global scale. However, a past analysis of the Australian WRA applied to multiple regions suggests that the same criteria effectively identify invasive plants in both temperate and tropical regions (Gordon et al. Reference Gordon, Onderdonk, Fox and Stocker2008). This finding suggests that using one of the existing national WRAs would lead to effective invasive plant identification for any country.

We found state/provincial-level WRAs for Australia, Canada, the United States (previously collected by Kesler [Reference Kesler2021]), and Ecuador. For Belgium, Brazil, Germany, Ireland, New Zealand, Norway, South Africa, South Korea, United Kingdom, and Zambia, we were unable to find WRAs for their most populous state/province, and therefore did not search for the remaining states and provinces. We found unique WRAs for 7 of the 9 states/territories in Australia, 2 of the 13 provinces/territories in Canada, and 1 of the 24 provinces in Ecuador. For comparison, Kesler (Reference Kesler2021) found unique WRAs for 31 states within the United States, with 2 additional states using a WRA that was created by a different state (we were unable to locate the WRAs for some of these states; our results include WRAs for 28 U.S. states). Although it was common for countries to use the same WRAs across borders, it was less common for states/provinces to use the same WRAs. For states/provinces that do not yet have WRAs, using an existing protocol from a neighboring state rather than creating a new one could be more efficient and enable more direct sharing of completed risk assessments. Even though state WRAs are often unique, sharing of completed risk assessments would still benefit others because risk assessment criteria often overlap (Bradley et al. Reference Bradley, Beaury, Fusco, Munro, Brown-Lima, Coville, Kesler, Olmstead and Parker2022).

Meeting Minimum Standards

The average number of standards met by each national-level WRA was 17.6 ± 3.2 (SD) out of 24 total standards. For the state/provincial-level WRAs in Canada, Australia, the United States, and Ecuador, the average was 14.4 ± 3.7 (SD). There were several standards that were commonly met by both national-level and state/provincial-level WRAs that could be shared across jurisdictions to reduce replication of effort (Figure 2). Almost all assessments included criteria related to the likelihood that the species will establish, spread, and have negative impacts within the region (Standards 2B–2D; Figure 2). Broadly, this information aligns with the general guidelines for WRAs recommended by the IPPC (FAO 2019b). This type of information could readily inform risk assessment for neighboring regions and beyond, given the many similarities between climate and ecoregions globally. Similarly, most WRAs included information about negative environmental (Standard 5) and socioeconomic impacts (Standard 7; Figure 2). Roy et al. (Reference Roy, Rabitsch, Scalera, Stewart, Gallardo, Genovesi, Essl, Adriaens, Bacher, Booy, Branquart, Brunel, Copp, Dean and D’hondt2018) noted that many of the risk assessments they evaluated did not include socioeconomic impacts, while our analysis found that most did. One possible explanation for this pattern may be that regulatory WRAs (i.e., the ones we sampled) are more likely to include a wider range of possible impacts that reflect a country’s widely varied interests. Moreover, the IPPC (FAO 2019b), which provides general guidelines for national WRAs, emphasizes socioeconomic impacts but not ecological impacts. In contrast, risk assessments developed primarily for academic purposes (which were included in the Roy et al. [Reference Roy, Rabitsch, Scalera, Stewart, Gallardo, Genovesi, Essl, Adriaens, Bacher, Booy, Branquart, Brunel, Copp, Dean and D’hondt2018] analysis) may primarily reflect a researcher’s area of expertise (e.g., ecology), without evaluating possible consequences in other disciplines. Nevertheless, given that our analysis focused only on regulatory WRAs, it is encouraging that both ecological and socioeconomic impacts are usually included. Finally, most WRAs included a list of data sources (Standard 11). This suggests that a considerable amount of similar information and number of associated sources is being compiled about invasive plant distributions and impacts across jurisdictions. We reiterate calls for global, open-source repositories of invasive species information (Barney et al. Reference Barney, Tekiela, Barrios-Garcia, Dimarco, Hufbauer, Leipzig-Scott, Nuñez, Pauchard, Pyšek, Vítková and Maxwell2015; Culina et al. Reference Culina, Baglioni, Crowther, Visser, Woutersen-Windhouwer and Manghi2018; Fusco et al. Reference Fusco, Beaury, Bradley, Cox, Jarnevich, Mahood, Nagy, Nietupski and Halofsky2023) to reduce the likelihood of redundant efforts to compile that same information (e.g., documentation of ecological impacts, introduction pathways, or potential habitat).

Figure 2. Percentage weed risk assessments (WRAs) that fulfilled each minimum standard. (A) Fulfillment of minimum standards by national WRAs (out of 20 total) and (B) fulfillment of minimum standards by state WRAs (out of 37 total; CAN = 2, AUS = 7, ECU = 1, US = 27). See Table 2 for descriptions of each minimum standard.

Conversely, there were several minimum standards that were frequently missing from WRAs. Fewer than 50% of both national and state/provincial-level WRAs included information about the native range (Standard 1C) or socioeconomic benefits (Standard 1F)—both of these standards are part of the “basic species description.” Native range is not relevant as a risk factor by itself, but instead can inform assessments of likelihood of introduction and establishment (Standards 2A and 2B). Socioeconomic benefits are also not a risk factor (Carneiro et al. Reference Carneiro, Hulme, Cuthbert, Kourantidou, Bang, Haubrock, Bradshaw, Balzani, Bacher, Latombe, Bodey, Probert, Quilodrán and Courchamp2024), but could be useful for contextualizing the motivation for introducing the species. Additionally, fewer than 50% of WRAs included information about potential impacts on ecosystem services (Standard 6) or the effects of climate change on invasion risk (Standard 9). These latter two were also sparse in the Roy et al. (Reference Roy, Rabitsch, Scalera, Stewart, Gallardo, Genovesi, Essl, Adriaens, Bacher, Booy, Branquart, Brunel, Copp, Dean and D’hondt2018) analysis, and were surprisingly absent in some of the most recently adopted regulatory WRAs in our analysis (COSAVE, Guide for Developing Pest Risk Analysis (BRA), A Framework to Support Alien Species Regulation: The Risk Analysis for Alien Taxa (RAAT), and Centre for Agriculture and Biosciences International Pest Risk Analysis Tool (CABI)). One plausible explanation for these omissions is that both impacts on ecosystem services and interactions with climate change are challenging to estimate without existing scientific data and are therefore unlikely to be observed based on expert knowledge alone.

The persistent lack of explicit consideration of climate change (Standard 9; Figure 3) is particularly concerning, because climate change is likely to change the sets of species that pose high risk within countries or states (Bradley et al. Reference Bradley, Beaury, Fusco and Lopez2023; Colberg et al. Reference Colberg, Bradley, Morelli and Brown-Lima2024) and is a top concern for invasive species management (Beaury et al. Reference Beaury, Fusco, Jackson, Laginhas, Morelli, Allen, Pasquarella and Bradley2020). The Global South in particular appears ill-prepared to assess interactions between invasive plants and climate change. Models of climate change impacts on invasive species range shifts are becoming increasingly common at both global and regional scales, such that assessors in the future could more easily find information about potential for establishment (e.g., Allen and Bradley Reference Allen and Bradley2016; Bezeng et al. Reference Bezeng, Morales-Castilla, van der Bank, Yessoufou, Daru and Davies2017; Gallagher et al. Reference Gallagher, Hughes and Leishman2013) and abundance/impact (e.g., Evans et al. Reference Evans, Jarnevich, Beaury, Engelstad, Teich, LaRoe and Bradley2024; O’Neill et al. Reference O’Neill, Bradley and Allen2021). In the absence of future distribution or abundance projections, assessors could consider whether the species currently poses a risk to ecosystems warmer than their target region and extrapolate risk. Explicitly including climate change in WRAs, even if data are currently lacking, ensures that assessors are thinking about the changing likelihood of establishment and impact due to warming. Failing to include climate change misses an opportunity to proactively prevent future invasions (Bradley et al. Reference Bradley, Beaury, Fusco and Lopez2023).

Figure 3. Climate change consideration (Standard 9) remains rare in national-level and state/provincial-level weed risk assessments (WRAs). The global map (top panel) uses a Robinson projection and is at a scale of 1:200,000,000. The continental maps (bottom panels) display WRAs used by 38 states or provinces (37 total WRAs). The continental maps use Albers equal area conic projections and are at a scale of 1:150,000,000. We were unable to find WRAs for countries and states/provinces shown in white.

Limitations

Although Roy et al.’s (Reference Roy, Rabitsch, Scalera, Stewart, Gallardo, Genovesi, Essl, Adriaens, Bacher, Booy, Branquart, Brunel, Copp, Dean and D’hondt2018) minimum standards include a comprehensive set of criteria for WRA, all of these criteria are not necessarily strong predictors of invasion risk. Past studies of risk assessment efficacy have found high levels of accuracy with fewer criteria or questions (Caley and Kuhnert Reference Caley and Kuhnert2006; Conser et al. Reference Conser, Seebacher, Fujino, Reichard and DiTomaso2015; Gordon et al. Reference Gordon, Onderdonk, Fox and Stocker2008; Koop et al. Reference Koop, Fowler, Newton and Caton2012). Criteria with high predictive power include whether a species has established and/or become invasive elsewhere (Caley and Kuhnert Reference Caley and Kuhnert2006; Conser et al. Reference Conser, Seebacher, Fujino, Reichard and DiTomaso2015; Koop et al. Reference Koop, Fowler, Newton and Caton2012), whether the species has negative ecological impacts (Conser et al. Reference Conser, Seebacher, Fujino, Reichard and DiTomaso2015; Koop et al. Reference Koop, Fowler, Newton and Caton2012), whether the species has an invasive congener (Buonaiuto et al. Reference Buonaiuto, Evans, Fertakos, Pfadenhauer, Salva and Bradley2023; Conser et al. Reference Conser, Seebacher, Fujino, Reichard and DiTomaso2015), and whether the species has broad climatic tolerance (Higgins and Richardson Reference Higgins and Richardson2014; Pfadenhauer et al. Reference Pfadenhauer, Nelson, Laginhas and Bradley2023). Including additional criteria (as is typical in the WRAs we assessed) makes it likely that more information will be available to potentially inform assessments in other jurisdictions. However, additional criteria may also make the WRA process time-prohibitive for some countries or states. For example, Verbrugge et al. (Reference Verbrugge, Leuven and van der Velde2010) estimated that some comprehensive risk assessments could take up to a week of time to complete for a single species. Given that an estimated 14,000 plants have been introduced and established outside their native ranges (van Kleunen et al. Reference Kleunen, Dawson, Essl, Pergl, Winter, Weber, Kreft, Weigelt, Kartesz, Nishino, Antonova, Barcelona, Cabezas, Cárdenas and Cárdenas-Toro2015), it might be a more realistic approach to start with a shorter risk assessment process such as horizon scanning (Kendig et al. Reference Kendig, Canavan, Anderson, Flory, Gettys, Gordon, Iannone, Kunzer, Petri, Pfingsten and Lieurance2022) and then only complete full WRAs for higher priority species that result from the initial screening process.

As implemented here, the minimum standards are also biased toward environmental impacts. Three separate standards focus on various components of environmental effects (Standard 5: biodiversity; Standard 6: ecosystem services; and Standard 8: threatened species and habitats), while only one standard focuses on socioeconomic impacts (Standard 7). To reduce bias toward environmental impacts, future reviews of WRAs may opt to split Standard 7 into three separate standards, with one for social, economic, and cultural impacts, respectively. Similarly, future assessors may choose to include separate questions about these three categories of impact in their WRAs.

Conclusions

Our analysis reveals a high degree of consistency across both national and state/provincial-level WRA criteria. Broad adoption of the same protocols (e.g., EPPO, IPPC, Hawai’i Pacific Islands Weed Risk Assessment (PAC IS), Guidelines for Pest Risk Analysis of Imported Plants and Plant Products (CARIB), and COSAVE) across national borders supports the sharing of completed risk assessments, reducing duplication of effort in a field with limited resources (Beaury et al. Reference Beaury, Fusco, Jackson, Laginhas, Morelli, Allen, Pasquarella and Bradley2020). Using the same or similar protocols and sharing results would be an important advance toward unified defenses against potential invasive plants. An approach of sharing completed assessments is also critical at state/provincial levels, where regulated species are highly inconsistent across borders (Beaury et al. Reference Beaury, Fusco, Allen and Bradley2021a; Lakoba et al. Reference Lakoba, Brooks, Haak and Barney2020), leading to patchy defenses against potential invasive plants.

Unfortunately, most nations and states do not appear to have WRAs or a regulatory process for preventing the introduction of invasive plants. While it is probable that some WRAs in non-English languages were not captured, our study, combined with previous analyses of proactive invasive species policies (Early et al. Reference Early, Bradley, Dukes, Lawler, Olden, Blumenthal, Gonzalez, Grosholz, Ibañez, Miller, Sorte and Tatem2016), nevertheless suggests that many gaps remain in phytosanitary practices globally. Countries without WRAs would benefit from adopting one of the established protocols outlined here, optimally from a neighbor, to increase regional consistency in approach. It is well known that preventing introductions is the most environmentally effective and cost-efficient approach to managing invasive species, underscoring the need for broader implementation of these critical tools. Policymakers should prioritize adopting comprehensive WRAs that account for climate change, uncertainty, and ecosystem service impacts to enhance regulatory consistency and effectiveness.